1 systemd System and Service Manager
3 CHANGES WITH 232 in spe
5 * The new RemoveIPC= option can be used to remove IPC objects owned by
6 the user or group of a service when that service exits.
8 * The new ProtectKernelModules= option can be used to disable explicit
9 load and unload operations of kernel modules by a service. In
10 addition access to /usr/lib/modules is removed if this option is set.
12 * ProtectSystem= option gained a new value "strict", which causes the
13 whole file system tree with the exception of /dev, /proc, and /sys,
14 to be remounted read-only for a service.
16 * The new ProtectKernelTunables= option can be used to disable
17 modification of configuration files in /sys and /proc by a service.
18 Various directories and files are remounted read-only, so access is
19 restricted even if the file permissions would allow it.
21 * The new ProtectControlGroups= option can be used to disable write
22 access by a service to /sys/fs/cgroup.
24 * Various systemd services have been hardened with
25 ProtectKernelTunables=yes, ProtectControlGroups=yes,
26 RestrictAddressFamilies=.
28 * Support for dynamically creating users for the lifetime of a service
29 has been added. If DynamicUser=yes is specified, user and group IDs
30 will be allocated from the range 61184..65519 for the lifetime of the
31 service. They can be resolved using the new nss-systemd.so NSS
32 module. The module must be enabled in /etc/nsswitch.conf. Services
33 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
34 any resources allocated by the service will be cleaned up when the
35 service exits. They also have ProtectHome=read-only and
36 ProtectSystem=strict enabled, so they are not able to make any
37 permanent modifications to the system.
39 * The nss-systemd module also always resolves root and nobody, making
40 it possible to have no /etc/passwd or /etc/group files in minimal
41 container or chroot environments.
43 * Services may be started with their own user namespace using the new
44 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
45 under which the service is running are mapped. All other users are
48 * Support for the cgroup namespace has been added to systemd-nspawn. If
49 supported by kernel, the container system started by systemd-nspawn
50 will have its own view of the cgroup hierarchy. This new behaviour
51 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
53 * The new MemorySwapMax= option can be used to limit the maximum swap
54 usage under the unified cgroup hierarchy.
56 * Support for the CPU controller in the unified cgroup hierarchy has
57 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
58 options. This controller requires out-of-tree patches for the kernel
59 and the support is provisional.
61 * Mount and automount units may now be created transiently
62 (i.e. dynamically at runtime via the bus API, instead of requiring
63 unit files in the file system).
65 * systemd-mount is a new tool which may mount file systems – much like
66 mount(8), optionally pulling in additional dependencies through
67 transient .mount and .automount units. For example, this tool
68 automatically runs fsck on a backing block device before mounting,
69 and allows the automount logic to be used dynamically from the
70 command line for establishing mount points. This tool is particularly
71 useful when dealing with removable media, as it will ensure fsck is
72 run – if necessary – before the first access and that the file system
73 is quickly unmounted after each access by utilizing the automount
74 logic. This maximizes the chance that the file system on the
75 removable media stays in a clean state, and if it isn't in a clean
76 state is fixed automatically.
78 * LazyUnmount=yes option for mount units has been added to expose the
79 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
82 * /efi will be used as the mount point of the EFI boot partition, if
83 the directory is present, and the mount point was not configured
84 through other means (e.g. fstab). If /efi directory does not exist,
85 /boot will be used as before. This makes it easier to automatically
86 mount the EFI partition on systems where /boot is used for something
89 * When operating on GPT disk images for containers, systemd-nspawn will
90 now mount the ESP to /boot or /efi according to the same rules as PID
91 1 running on a host. This allows tools like "bootctl" to operate
92 correctly within such containers, in order to make container images
93 bootable on physical systems.
95 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
97 * Two new user session targets have been added to support running
98 graphical sessions under the systemd --user instance:
99 graphical-session.target and graphical-session-pre.target. See
100 systemd.special(7) for a description of how those targets should be
103 * The vconsole initialization code has been significantly reworked to
104 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
105 support unicode keymaps. Font and keymap configuration will now be
106 copied to all allocated virtual consoles.
108 * FreeBSD's bhyve virtualization is now detected.
110 * Information recorded in the journal for core dumps now includes the
111 contents of /proc/mountinfo and the command line of the process at
112 the top of the process hierarchy (which is usually the init process
115 * systemd-journal-gatewayd learned the --directory= option to serve
116 files from the specified location.
118 * journalctl --root=… can be used to peruse the journal in the
119 /var/log/ directories inside of a container tree. This is similar to
120 the existing --machine= option, but does not require the container to
123 * The hardware database has been extended to support
124 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
127 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
128 specify the click rate for mice which include a horizontal wheel with
129 a click rate that is different than the one for the vertical wheel.
131 * systemd-run gained a new --wait option that makes service execution
132 synchronous. (Specifically, the command will not return until the
133 specified service binary exited.)
135 * systemctl gained a new --wait option that causes the start command to
136 wait until the units being started have terminated again.
138 * A new journal output mode "short-full" has been added which displays
139 timestamps with abbreviated English day names and adds a timezone
140 suffix. Those timestamps include more information than the default
141 "short" output mode, and can be passed directly to journalctl's
142 --since= and --until= options.
144 * /etc/resolv.conf will be bind-mounted into containers started by
145 systemd-nspawn, if possible, so any changes to resolv.conf contents
146 are automatically propagated to the container.
148 * The number of instances for socket-activated services originating
149 from a single IP address can be limited with
150 MaxConnectionsPerSource=, extending the existing setting of
153 * systemd-networkd gained support for vcan ("Virtual CAN") interface
156 * .netdev and .network configuration can now be extended through
159 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
160 Segmentation Offload, Generic Receive Offload, Large Receive Offload
161 can be enabled and disabled using the new UDPSegmentationOffload=,
162 TCPSegmentationOffload=, GenericSegmentationOffload=,
163 GenericReceiveOffload=, LargeReceiveOffload= options in the
164 [Link] section of .link files.
166 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
167 Port VLAN ID can be configured for bridge devices using the new STP=,
168 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
169 section of .netdev files.
171 * The route table to which routes received over DHCP or RA should be
172 added can be configured with the new RouteTable= option in the [DHCP]
173 and [IPv6AcceptRA] sections of .network files.
175 * The Address Resolution Protocol can be disabled on links managed by
176 systemd-networkd using the ARP=no setting in the [Link] section of
179 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
180 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
181 encode information about the result and exit codes of the current
182 service runtime cycle.
184 * systemd-sysctl will now configure kernel parameters in the order
185 they occur in the configuration files. This matches what sysctl
186 has been traditionally doing.
188 * kernel-install "plugins" that are executed to perform various
189 tasks after a new kernel is added and before an old one is removed
190 can now return a special value to terminate the procedure and
191 prevent any later plugins from running.
193 * Journald's SplitMode=login setting has been deprecated. It has been
194 removed from documentation, and its use is discouraged. In a future
195 release it will be completely removed, and made equivalent to current
196 default of SplitMode=uid.
198 * Storage=both option setting in /etc/systemd/coredump.conf has been
199 removed. With fast LZ4 compression storing the core dump twice is not
202 * The --share-system systemd-nspawn option has been replaced with an
203 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
204 this functionality is discouraged. In addition the variables
205 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
206 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
207 individual namespaces.
209 * "machinectl list" now shows the IP address of running containers in
210 the output, as well as OS release information.
212 * "loginctl list" now shows the TTY of each session in the output.
214 * sd-bus gained new API calls sd_bus_track_set_recursive(),
215 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
216 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
217 tracking objects in a "recursive" mode, where a single client can be
218 counted multiple times, if it takes multiple references.
220 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
221 sd_bus_get_exit_on_disconnect(). They may be used to to make a
222 process using sd-bus automatically exit if the bus connection is
225 * Bus clients of the service manager may now "pin" loaded units into
226 memory, by taking an explicit reference on them. This is useful to
227 ensure the client can retrieve runtime data about the service even
228 after the service completed execution. Taking such a reference is
229 available only for privileged clients and should be helpful to watch
230 running services in a race-free manner, and in particular collect
231 information about exit statuses and results.
233 * The nss-resolve module has been changed to strictly return UNAVAIL
234 when communication via D-Bus with resolved failed, and NOTFOUND when
235 a lookup completed but was negative. This means it is now possible to
236 neatly configure fallbacks using nsswitch.conf result checking
237 expressions. Taking benefit of this, the new recommended
238 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
240 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
242 * A new setting CtrlAltDelBurstAction= has been added to
243 /etc/systemd/system.conf which may be used to configure the precise
244 behaviour if the user on the console presses Ctrl-Alt-Del more often
245 than 7 times in 2s. Previously this would unconditionally result in
246 an expedited, immediate reboot. With this new setting the precise
247 operation may be configured in more detail, and also turned off
250 * In .netdev files two new settings RemoteChecksumTx= and
251 RemoteChecksumRx= are now understood that permit configuring the
252 remote checksumming logic for VXLAN networks.
254 * The service manager learnt a new "invocation ID" concept for invoked
255 services. Each runtime cycle of a service will get a new invocation
256 ID (a 128bit random UUID) assigned that identifies the current
257 run of the service uniquely and globally. A new invocation ID
258 is generated each time a service starts up. The journal will store
259 the invocation ID of a service along with any logged messages, thus
260 making the invocation ID useful for matching the online runtime of a
261 service with the offline log data it generated in a safe way without
262 relying on synchronized timestamps. In many ways this new service
263 invocation ID concept is similar to the kernel's boot ID concept that
264 uniquely and globally identifies the runtime of each boot. The
265 invocation ID of a service is passed to the service itself via an
266 environment variable ($INVOCATION_ID). A new bus call
267 GetUnitByInvocationID() has been added that is similar to GetUnit()
268 but instead of retrieving the bus path for a unit by its name
269 retrieves it by its invocation ID. The returned path is valid only as
270 long as the passed invocation ID is current.
272 * systemd-resolved gained a new "DNSStubListener" setting in
273 resolved.conf. It either takes a boolean value or the special values
274 "udp" and "tcp", and configures whether to enable the stub DNS
275 listener on 127.0.0.53:53.
277 * IP addresses configured via networkd may now carry additional
278 configuration settings supported by the kernel. New options include:
279 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
280 PrefixRoute=, AutoJoin=.
282 * The PAM configuration fragment file for "user@.service" shipped with
283 systemd (i.e. the --user instance of systemd) has been stripped to
284 the minimum necessary to make the system boot. Previously, it
285 contained Fedora-specific stanzas that did not apply to other
286 distributions. It is expected that downstream distributions add
287 additional configuration lines, matching their needs to this file,
288 using it only as rough template of what systemd itself needs. Note
289 that this reduced fragment does not even include an invocation of
290 pam_limits which most distributions probably want to add, even though
291 systemd itself does not need it. (There's also the new build time
292 option --with-pamconfdir=no to disable installation of the PAM
295 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
296 capability is now also dropped from its set (in addition to
297 CAP_SYS_MKNOD as before).
299 * In service unit files it is now possible to connect a specific named
300 file descriptor with stdin/stdout/stdout of an executed service. The
301 name may be specified in matching .socket units using the
302 FileDescriptorName= setting.
304 * A number of journal settings may now be configured on the kernel
305 command line. Specifically, the following options are now understood:
306 systemd.journald.max_level_console=,
307 systemd.journald.max_level_store=,
308 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
309 systemd.journald.max_level_wall=.
311 * "systemctl is-enabled --full" will now show by which symlinks a unit
312 file is enabled in the unit dependency tree.
316 * In service units the various ExecXYZ= settings have been extended
317 with an additional special character as first argument of the
318 assigned value: if the character '+' is used the specified command
319 line it will be run with full privileges, regardless of User=,
320 Group=, CapabilityBoundingSet= and similar options. The effect is
321 similar to the existing PermissionsStartOnly= option, but allows
322 configuration of this concept for each executed command line
325 * Services may now alter the service watchdog timeout at runtime by
326 sending a WATCHDOG_USEC= message via sd_notify().
328 * MemoryLimit= and related unit settings now optionally take percentage
329 specifications. The percentage is taken relative to the amount of
330 physical memory in the system (or in case of containers, the assigned
331 amount of memory). This allows scaling service resources neatly with
332 the amount of RAM available on the system. Similarly, systemd-logind's
333 RuntimeDirectorySize= option now also optionally takes percentage
336 * In similar fashion TasksMax= takes percentage values now, too. The
337 value is taken relative to the configured maximum number of processes
338 on the system. The per-service task maximum has been changed to 15%
339 using this functionality. (Effectively this is an increase of 512 →
340 4915 for service units, given the kernel's default pid_max setting.)
342 * Calendar time specifications in .timer units now understand a ".."
343 syntax for time ranges. Example: "4..7:10" may now be used for
344 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
347 * The InaccessableDirectories=, ReadOnlyDirectories= and
348 ReadWriteDirectories= unit file settings have been renamed to
349 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
350 applied to all kinds of file nodes, and not just directories, with
351 the exception of symlinks. Specifically these settings may now be
352 used on block and character device nodes, UNIX sockets and FIFOS as
353 well as regular files. The old names of these settings remain
354 available for compatibility.
356 * systemd will now log about all service processes it kills forcibly
357 (using SIGKILL) because they remained after the clean shutdown phase
358 of the service completed. This should help identifying services that
359 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
360 systemd-logind's configuration a similar log message is generated for
361 processes killed at the end of each session due to this setting.
363 * systemd will now set the $JOURNAL_STREAM environment variable for all
364 services whose stdout/stderr are connected to the Journal (which
365 effectively means by default: all services). The variable contains
366 the device and inode number of the file descriptor used for
367 stdout/stderr. This may be used by invoked programs to detect whether
368 their stdout/stderr is connected to the Journal, in which case they
369 can switch over to direct Journal communication, thus being able to
370 pass extended, structured metadata along with their log messages. As
371 one example, this is now used by glib's logging primitives.
373 * When using systemd's default tmp.mount unit for /tmp, the mount point
374 will now be established with the "nosuid" and "nodev" options. This
375 avoids privilege escalation attacks that put traps and exploits into
376 /tmp. However, this might cause problems if you e. g. put container
377 images or overlays into /tmp; if you need this, override tmp.mount's
378 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
381 * systemd now supports the "memory" cgroup controller also on
384 * The systemd-cgtop tool now optionally takes a control group path as
385 command line argument. If specified, the control group list shown is
386 limited to subgroups of that group.
388 * The SystemCallFilter= unit file setting gained support for
389 pre-defined, named system call filter sets. For example
390 SystemCallFilter=@clock is now an effective way to make all clock
391 changing-related system calls unavailable to a service. A number of
392 similar pre-defined groups are defined. Writing system call filters
393 for system services is simplified substantially with this new
394 concept. Accordingly, all of systemd's own, long-running services now
395 enable system call filtering based on this, by default.
397 * A new service setting MemoryDenyWriteExecute= has been added, taking
398 a boolean value. If turned on, a service may no longer create memory
399 mappings that are writable and executable at the same time. This
400 enhances security for services where this is enabled as it becomes
401 harder to dynamically write and then execute memory in exploited
402 service processes. This option has been enabled for all of systemd's
403 own long-running services.
405 * A new RestrictRealtime= service setting has been added, taking a
406 boolean argument. If set the service's processes may no longer
407 acquire realtime scheduling. This improves security as realtime
408 scheduling may otherwise be used to easily freeze the system.
410 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
411 value. This may be used for requesting that the system manager inside
412 of the container reports start-up completion to nspawn which then
413 propagates this notification further to the service manager
414 supervising nspawn itself. A related option NotifyReady= in .nspawn
415 files has been added too. This functionality allows ordering of the
416 start-up of multiple containers using the usual systemd ordering
419 * machinectl gained a new command "stop" that is an alias for
422 * systemd-resolved gained support for contacting DNS servers on
423 link-local IPv6 addresses.
425 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
426 its caches. A method call for requesting the same operation has been
427 added to the bus API too, and is made available via "systemd-resolve
430 * systemd-resolve gained a new --status switch. If passed a brief
431 summary of the used DNS configuration with per-interface information
434 * resolved.conf gained a new Cache= boolean option, defaulting to
435 on. If turned off local DNS caching is disabled. This comes with a
436 performance penalty in particular when DNSSEC is enabled. Note that
437 resolved disables its internal caching implicitly anyway, when the
438 configured DNS server is on a host-local IP address such as ::1 or
439 127.0.0.1, thus automatically avoiding double local caching.
441 * systemd-resolved now listens on the local IP address 127.0.0.53:53
442 for DNS requests. This improves compatibility with local programs
443 that do not use the libc NSS or systemd-resolved's bus APIs for name
444 resolution. This minimal DNS service is only available to local
445 programs and does not implement the full DNS protocol, but enough to
446 cover local DNS clients. A new, static resolv.conf file, listing just
447 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
448 now recommended to make /etc/resolv.conf a symlink to this file in
449 order to route all DNS lookups to systemd-resolved, regardless if
450 done via NSS, the bus API or raw DNS packets. Note that this local
451 DNS service is not as fully featured as the libc NSS or
452 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
453 used to deliver link-local address information (as this implies
454 sending a local interface index along), LLMNR/mDNS support via this
455 interface is severely restricted. It is thus strongly recommended for
456 all applications to use the libc NSS API or native systemd-resolved
459 * systemd-networkd's bridge support learned a new setting
460 VLANFiltering= for controlling VLAN filtering. Moreover a new section
461 in .network files has been added for configuring VLAN bridging in
462 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
464 * systemd-networkd's IPv6 Router Advertisement code now makes use of
465 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
466 now be acquired without relying on DHCPv6. Two new options
467 UseDomains= and UseDNS= have been added to configure this behaviour.
469 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
470 renamed IPv6AcceptRA=, without altering its behaviour. The old
471 setting name remains available for compatibility reasons.
473 * The systemd-networkd VTI/VTI6 tunneling support gained new options
474 Key=, InputKey= and OutputKey=.
476 * systemd-networkd gained support for VRF ("Virtual Routing Function")
477 interface configuration.
479 * "systemctl edit" may now be used to create new unit files by
480 specifying the --force switch.
482 * sd-event gained a new function sd_event_get_iteration() for
483 requesting the current iteration counter of the event loop. It starts
484 at zero and is increased by one with each event loop iteration.
486 * A new rpm macro %systemd_ordering is provided by the macros.systemd
487 file. It can be used in lieu of %systemd_requires in packages which
488 don't use any systemd functionality and are intended to be installed
489 in minimal containers without systemd present. This macro provides
490 ordering dependencies to ensure that if the package is installed in
491 the same rpm transaction as systemd, systemd will be installed before
492 the scriptlets for the package are executed, allowing unit presets
495 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
496 been added to simplify packaging of generators.
498 * The os-release file gained VERSION_CODENAME field for the
499 distribution nickname (e.g. VERSION_CODENAME=woody).
501 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
502 can be set to disable parsing of metadata and the creation
503 of persistent symlinks for that device.
505 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
506 to make them available to logged-in users has been reverted.
508 * Much of the common code of the various systemd components is now
509 built into an internal shared library libsystemd-shared-231.so
510 (incorporating the systemd version number in the name, to be updated
511 with future releases) that the components link to. This should
512 decrease systemd footprint both in memory during runtime and on
513 disk. Note that the shared library is not for public use, and is
514 neither API not ABI stable, but is likely to change with every new
515 released update. Packagers need to make sure that binaries
516 linking to libsystemd-shared.so are updated in step with the
519 * Configuration for "mkosi" is now part of the systemd
520 repository. mkosi is a tool to easily build legacy-free OS images,
521 and is available on github: https://github.com/systemd/mkosi. If
522 "mkosi" is invoked in the build tree a new raw OS image is generated
523 incorporating the systemd sources currently being worked on and a
524 clean, fresh distribution installation. The generated OS image may be
525 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
526 UEFI PC. This functionality is particularly useful to easily test
527 local changes made to systemd in a pristine, defined environment. See
530 * configure learned the --with-support-url= option to specify the
531 distribution's bugtracker.
533 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
534 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
535 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
536 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
537 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
538 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
539 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
540 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
541 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
542 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
543 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
544 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
545 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
546 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
547 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
548 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
549 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
550 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
551 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
557 * DNSSEC is now turned on by default in systemd-resolved (in
558 "allow-downgrade" mode), but may be turned off during compile time by
559 passing "--with-default-dnssec=no" to "configure" (and of course,
560 during runtime with DNSSEC= in resolved.conf). We recommend
561 downstreams to leave this on at least during development cycles and
562 report any issues with the DNSSEC logic upstream. We are very
563 interested in collecting feedback about the DNSSEC validator and its
564 limitations in the wild. Note however, that DNSSEC support is
565 probably nothing downstreams should turn on in stable distros just
566 yet, as it might create incompatibilities with a few DNS servers and
567 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
568 automatically whenever we detect such incompatible setups, but there
569 might be systems we do not cover yet. Hence: please help us testing
570 the DNSSEC code, leave this on where you can, report back, but then
571 again don't consider turning this on in your stable, LTS or
572 production release just yet. (Note that you have to enable
573 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
574 and its DNSSEC mode for host name resolution from local
577 * systemd-resolve conveniently resolves DANE records with the --tlsa
578 option and OPENPGPKEY records with the --openpgp option. It also
579 supports dumping raw DNS record data via the new --raw= switch.
581 * systemd-logind will now by default terminate user processes that are
582 part of the user session scope unit (session-XX.scope) when the user
583 logs out. This behavior is controlled by the KillUserProcesses=
584 setting in logind.conf, and the previous default of "no" is now
585 changed to "yes". This means that user sessions will be properly
586 cleaned up after, but additional steps are necessary to allow
587 intentionally long-running processes to survive logout.
589 While the user is logged in at least once, user@.service is running,
590 and any service that should survive the end of any individual login
591 session can be started at a user service or scope using systemd-run.
592 systemd-run(1) man page has been extended with an example which shows
593 how to run screen in a scope unit underneath user@.service. The same
594 command works for tmux.
596 After the user logs out of all sessions, user@.service will be
597 terminated too, by default, unless the user has "lingering" enabled.
598 To effectively allow users to run long-term tasks even if they are
599 logged out, lingering must be enabled for them. See loginctl(1) for
600 details. The default polkit policy was modified to allow users to
601 set lingering for themselves without authentication.
603 Previous defaults can be restored at compile time by the
604 --without-kill-user-processes option to "configure".
606 * systemd-logind gained new configuration settings SessionsMax= and
607 InhibitorsMax=, both with a default of 8192. It will not register new
608 user sessions or inhibitors above this limit.
610 * systemd-logind will now reload configuration on SIGHUP.
612 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
613 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
614 enable. Also, support for the "io" cgroup controller in the unified
615 hierarchy has been added, so that the "memory", "pids" and "io" are
616 now the controllers that are supported on the unified hierarchy.
618 WARNING: it is not possible to use previous systemd versions with
619 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
620 is necessary to also update systemd in the initramfs if using the
621 unified hierarchy. An updated SELinux policy is also required.
623 * LLDP support has been extended, and both passive (receive-only) and
624 active (sender) modes are supported. Passive mode ("routers-only") is
625 enabled by default in systemd-networkd. Active LLDP mode is enabled
626 by default for containers on the internal network. The "networkctl
627 lldp" command may be used to list information gathered. "networkctl
628 status" will also show basic LLDP information on connected peers now.
630 * The IAID and DUID unique identifier sent in DHCP requests may now be
631 configured for the system and each .network file managed by
632 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
634 * systemd-networkd gained support for configuring proxy ARP support for
635 each interface, via the ProxyArp= setting in .network files. It also
636 gained support for configuring the multicast querier feature of
637 bridge devices, via the new MulticastQuerier= setting in .netdev
638 files. Similarly, snooping on the IGMP traffic can be controlled
639 via the new setting MulticastSnooping=.
641 A new setting PreferredLifetime= has been added for addresses
642 configured in .network file to configure the lifetime intended for an
645 The systemd-networkd DHCP server gained the option EmitRouter=, which
646 defaults to yes, to configure whether the DHCP Option 3 (Router)
649 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
650 systemd-socket-activate and installed into /usr/bin. It is now fully
653 * systemd-journald now uses separate threads to flush changes to disk
654 when closing journal files, thus reducing impact of slow disk I/O on
657 * The sd-journal API gained two new calls
658 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
659 can be used to open journal files using file descriptors instead of
660 file or directory paths. sd_journal_open_container() has been
661 deprecated, sd_journal_open_directory_fd() should be used instead
662 with the flag SD_JOURNAL_OS_ROOT.
664 * journalctl learned a new output mode "-o short-unix" that outputs log
665 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
666 UTC). It also gained support for a new --no-hostname setting to
667 suppress the hostname column in the family of "short" output modes.
669 * systemd-ask-password now optionally skips printing of the password to
670 stdout with --no-output which can be useful in scripts.
672 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
673 (devices tagged with ID_MAKER_TOOL) are now tagged with
674 "uaccess" and are available to logged in users.
676 * The DeviceAllow= unit setting now supports specifiers (with "%").
678 * "systemctl show" gained a new --value switch, which allows print a
679 only the contents of a specific unit property, without also printing
680 the property's name. Similar support was added to "show*" verbs
681 of loginctl and machinectl that output "key=value" lists.
683 * A new unit type "generated" was added for files dynamically generated
684 by generator tools. Similarly, a new unit type "transient" is used
685 for unit files created using the runtime API. "systemctl enable" will
686 refuse to operate on such files.
688 * A new command "systemctl revert" has been added that may be used to
689 revert to the vendor version of a unit file, in case local changes
690 have been made by adding drop-ins or overriding the unit file.
692 * "machinectl clean" gained a new verb to automatically remove all or
693 just hidden container images.
695 * systemd-tmpfiles gained support for a new line type "e" for emptying
696 directories, if they exist, without creating them if they don't.
698 * systemd-nspawn gained support for automatically patching the UID/GIDs
699 of the owners and the ACLs of all files and directories in a
700 container tree to match the UID/GID user namespacing range selected
701 for the container invocation. This mode is enabled via the new
702 --private-users-chown switch. It also gained support for
703 automatically choosing a free, previously unused UID/GID range when
704 starting a container, via the new --private-users=pick setting (which
705 implies --private-users-chown). Together, these options for the first
706 time make user namespacing for nspawn containers fully automatic and
707 thus deployable. The systemd-nspawn@.service template unit file has
708 been changed to use this functionality by default.
710 * systemd-nspawn gained a new --network-zone= switch, that allows
711 creating ad-hoc virtual Ethernet links between multiple containers,
712 that only exist as long as at least one container referencing them is
713 running. This allows easy connecting of multiple containers with a
714 common link that implements an Ethernet broadcast domain. Each of
715 these network "zones" may be named relatively freely by the user, and
716 may be referenced by any number of containers, but each container may
717 only reference one of these "zones". On the lower level, this is
718 implemented by an automatically managed bridge network interface for
719 each zone, that is created when the first container referencing its
720 zone is created and removed when the last one referencing its zone
723 * The default start timeout may now be configured on the kernel command
724 line via systemd.default_timeout_start_sec=. It was already
725 configurable via the DefaultTimeoutStartSec= option in
726 /etc/systemd/system.conf.
728 * Socket units gained a new TriggerLimitIntervalSec= and
729 TriggerLimitBurst= setting to configure a limit on the activation
730 rate of the socket unit.
732 * The LimitNICE= setting now optionally takes normal UNIX nice values
733 in addition to the raw integer limit value. If the specified
734 parameter is prefixed with "+" or "-" and is in the range -20..19 the
735 value is understood as UNIX nice value. If not prefixed like this it
736 is understood as raw RLIMIT_NICE limit.
738 * Note that the effect of the PrivateDevices= unit file setting changed
739 slightly with this release: the per-device /dev file system will be
740 mounted read-only from this version on, and will have "noexec"
741 set. This (minor) change of behavior might cause some (exceptional)
742 legacy software to break, when PrivateDevices=yes is set for its
743 service. Please leave PrivateDevices= off if you run into problems
746 * systemd-bootchart has been split out to a separate repository:
747 https://github.com/systemd/systemd-bootchart
749 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
750 merged into the kernel in its current form.
752 * The compatibility libraries libsystemd-daemon.so,
753 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
754 which have been deprecated since systemd-209 have been removed along
755 with the corresponding pkg-config files. All symbols provided by
756 those libraries are provided by libsystemd.so.
758 * The Capabilities= unit file setting has been removed (it is ignored
759 for backwards compatibility). AmbientCapabilities= and
760 CapabilityBoundingSet= should be used instead.
762 * A new special target has been added, initrd-root-device.target,
763 which creates a synchronization point for dependencies of the root
764 device in early userspace. Initramfs builders must ensure that this
765 target is now included in early userspace.
767 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
768 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
769 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
770 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
771 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
772 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
773 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
774 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
775 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
776 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
777 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
778 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
779 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
780 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
781 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
782 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
783 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
784 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
785 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
786 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
787 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
788 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
789 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
790 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
791 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
794 — Fairfax, 2016-05-21
798 * The systemd-resolved DNS resolver service has gained a substantial
799 set of new features, most prominently it may now act as a DNSSEC
800 validating stub resolver. DNSSEC mode is currently turned off by
801 default, but is expected to be turned on by default in one of the
802 next releases. For now, we invite everybody to test the DNSSEC logic
803 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
804 service also gained a full set of D-Bus interfaces, including calls
805 to configure DNS and DNSSEC settings per link (for use by external
806 network management software). systemd-resolved and systemd-networkd
807 now distinguish between "search" and "routing" domains. The former
808 are used to qualify single-label names, the latter are used purely
809 for routing lookups within certain domains to specific links.
810 resolved now also synthesizes RRs for all entries from /etc/hosts.
812 * The systemd-resolve tool (which is a client utility for
813 systemd-resolved) has been improved considerably and is now fully
814 supported and documented. Hence it has moved from /usr/lib/systemd to
817 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
820 * The coredump collection logic has been reworked: when a coredump is
821 collected it is now written to disk, compressed and processed
822 (including stacktrace extraction) from a new instantiated service
823 systemd-coredump@.service, instead of directly from the
824 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
825 processing large coredumps can take up a substantial amount of
826 resources and time, and this previously happened entirely outside of
827 systemd's service supervision. With the new logic the core_pattern
828 hook only does minimal metadata collection before passing off control
829 to the new instantiated service, which is configured with a time
830 limit, a nice level and other settings to minimize negative impact on
831 the rest of the system. Also note that the new logic will honour the
832 RLIMIT_CORE setting of the crashed process, which now allows users
833 and processes to turn off coredumping for their processes by setting
836 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
837 and all forked processes by default. Previously, PID 1 would leave
838 the setting at "0" for all processes, as set by the kernel. Note that
839 the resource limit traditionally has no effect on the generated
840 coredumps on the system if the /proc/sys/kernel/core_pattern hook
841 logic is used. Since the limit is now honoured (see above) its
842 default has been changed so that the coredumping logic is enabled by
843 default for all processes, while allowing specific opt-out.
845 * When the stacktrace is extracted from processes of system users, this
846 is now done as "systemd-coredump" user, in order to sandbox this
847 potentially security sensitive parsing operation. (Note that when
848 processing coredumps of normal users this is done under the user ID
849 of process that crashed, as before.) Packagers should take notice
850 that it is now necessary to create the "systemd-coredump" system user
851 and group at package installation time.
853 * The systemd-activate socket activation testing tool gained support
854 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
855 and --seqpacket switches. It also has been extended to support both
856 new-style and inetd-style file descriptor passing. Use the new
857 --inetd switch to request inetd-style file descriptor passing.
859 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
860 variable, which takes a boolean value. If set to false, ANSI color
861 output is disabled in the tools even when run on a terminal that
864 * The VXLAN support in networkd now supports two new settings
865 DestinationPort= and PortRange=.
867 * A new systemd.machine_id= kernel command line switch has been added,
868 that may be used to set the machine ID in /etc/machine-id if it is
869 not initialized yet. This command line option has no effect if the
870 file is already initialized.
872 * systemd-nspawn gained a new --as-pid2 switch that invokes any
873 specified command line as PID 2 rather than PID 1 in the
874 container. In this mode PID 1 is a minimal stub init process that
875 implements the special POSIX and Linux semantics of PID 1 regarding
876 signal and child process management. Note that this stub init process
877 is implemented in nspawn itself and requires no support from the
878 container image. This new logic is useful to support running
879 arbitrary commands in the container, as normal processes are
880 generally not prepared to run as PID 1.
882 * systemd-nspawn gained a new --chdir= switch for setting the current
883 working directory for the process started in the container.
885 * "journalctl /dev/sda" will now output all kernel log messages for
886 specified device from the current boot, in addition to all devices
887 that are parents of it. This should make log output about devices
888 pretty useful, as long as kernel drivers attach enough metadata to
889 the log messages. (The usual SATA drivers do.)
891 * The sd-journal API gained two new calls
892 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
893 that report whether log data from /run or /var has been found.
895 * journalctl gained a new switch "--fields" that prints all journal
896 record field names currently in use in the journal. This is backed
897 by two new sd-journal API calls sd_journal_enumerate_fields() and
898 sd_journal_restart_fields().
900 * Most configurable timeouts in systemd now expect an argument of
901 "infinity" to turn them off, instead of "0" as before. The semantics
902 from now on is that a timeout of "0" means "now", and "infinity"
903 means "never". To maintain backwards compatibility, "0" continues to
904 turn off previously existing timeout settings.
906 * "systemctl reload-or-try-restart" has been renamed to "systemctl
907 try-reload-or-restart" to clarify what it actually does: the "try"
908 logic applies to both reloading and restarting, not just restarting.
909 The old name continues to be accepted for compatibility.
911 * On boot-up, when PID 1 detects that the system clock is behind the
912 release date of the systemd version in use, the clock is now set
913 to the latter. Previously, this was already done in timesyncd, in order
914 to avoid running with clocks set to the various clock epochs such as
915 1902, 1938 or 1970. With this change the logic is now done in PID 1
916 in addition to timesyncd during early boot-up, so that it is enforced
917 before the first process is spawned by systemd. Note that the logic
918 in timesyncd remains, as it is more comprehensive and ensures
919 clock monotonicity by maintaining a persistent timestamp file in
920 /var. Since /var is generally not available in earliest boot or the
921 initrd, this part of the logic remains in timesyncd, and is not done
924 * Support for tweaking details in net_cls.class_id through the
925 NetClass= configuration directive has been removed, as the kernel
926 people have decided to deprecate that controller in cgroup v2.
927 Userspace tools such as nftables are moving over to setting rules
928 that are specific to the full cgroup path of a task, which obsoletes
929 these controllers anyway. The NetClass= directive is kept around for
930 legacy compatibility reasons. For a more in-depth description of the
931 kernel change, please refer to the respective upstream commit:
933 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
935 * A new service setting RuntimeMaxSec= has been added that may be used
936 to specify a maximum runtime for a service. If the timeout is hit, the
937 service is terminated and put into a failure state.
939 * A new service setting AmbientCapabilities= has been added. It allows
940 configuration of additional Linux process capabilities that are
941 passed to the activated processes. This is only available on very
944 * The process resource limit settings in service units may now be used
945 to configure hard and soft limits individually.
947 * The various libsystemd APIs such as sd-bus or sd-event now publicly
948 expose support for gcc's __attribute__((cleanup())) C extension.
949 Specifically, for many object destructor functions alternative
950 versions have been added that have names suffixed with "p" and take a
951 pointer to a pointer to the object to destroy, instead of just a
952 pointer to the object itself. This is useful because these destructor
953 functions may be used directly as parameters to the cleanup
954 construct. Internally, systemd has been a heavy user of this GCC
955 extension for a long time, and with this change similar support is
956 now available to consumers of the library outside of systemd. Note
957 that by using this extension in your sources compatibility with old
958 and strictly ANSI compatible C compilers is lost. However, all gcc or
959 LLVM versions of recent years support this extension.
961 * Timer units gained support for a new setting RandomizedDelaySec= that
962 allows configuring some additional randomized delay to the configured
963 time. This is useful to spread out timer events to avoid load peaks in
964 clusters or larger setups.
966 * Calendar time specifications now support sub-second accuracy.
968 * Socket units now support listening on SCTP and UDP-lite protocol
971 * The sd-event API now comes with a full set of man pages.
973 * Older versions of systemd contained experimental support for
974 compressing journal files and coredumps with the LZ4 compressor that
975 was not compatible with the lz4 binary (due to API limitations of the
976 lz4 library). This support has been removed; only support for files
977 compatible with the lz4 binary remains. This LZ4 logic is now
978 officially supported and no longer considered experimental.
980 * The dkr image import logic has been removed again from importd. dkr's
981 micro-services focus doesn't fit into the machine image focus of
982 importd, and quickly got out of date with the upstream dkr API.
984 * Creation of the /run/lock/lockdev/ directory was dropped from
985 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
986 been available for many years. If you still need this, you need to
987 create your own tmpfiles.d config file with:
989 d /run/lock/lockdev 0775 root lock -
991 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
992 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
993 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
994 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
995 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
996 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
997 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
998 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
999 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
1000 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
1001 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
1002 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
1003 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
1004 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
1005 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
1006 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
1007 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
1008 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
1009 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
1011 — Berlin, 2016-02-11
1015 * A number of properties previously only settable in unit
1016 files are now also available as properties to set when
1017 creating transient units programmatically via the bus, as it
1018 is exposed with systemd-run's --property=
1019 setting. Specifically, these are: SyslogIdentifier=,
1020 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
1021 EnvironmentFile=, ReadWriteDirectories=,
1022 ReadOnlyDirectories=, InaccessibleDirectories=,
1023 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
1025 * When creating transient services via the bus API it is now
1026 possible to pass in a set of file descriptors to use as
1027 STDIN/STDOUT/STDERR for the invoked process.
1029 * Slice units may now be created transiently via the bus APIs,
1030 similar to the way service and scope units may already be
1031 created transiently.
1033 * Wherever systemd expects a calendar timestamp specification
1034 (like in journalctl's --since= and --until= switches) UTC
1035 timestamps are now supported. Timestamps suffixed with "UTC"
1036 are now considered to be in Universal Time Coordinated
1037 instead of the local timezone. Also, timestamps may now
1038 optionally be specified with sub-second accuracy. Both of
1039 these additions also apply to recurring calendar event
1040 specification, such as OnCalendar= in timer units.
1042 * journalctl gained a new "--sync" switch that asks the
1043 journal daemon to write all so far unwritten log messages to
1044 disk and sync the files, before returning.
1046 * systemd-tmpfiles learned two new line types "q" and "Q" that
1047 operate like "v", but also set up a basic btrfs quota
1048 hierarchy when used on a btrfs file system with quota
1051 * tmpfiles' "v", "q" and "Q" will now create a plain directory
1052 instead of a subvolume (even on a btrfs file system) if the
1053 root directory is a plain directory, and not a
1054 subvolume. This should simplify things with certain chroot()
1055 environments which are not aware of the concept of btrfs
1058 * systemd-detect-virt gained a new --chroot switch to detect
1059 whether execution takes place in a chroot() environment.
1061 * CPUAffinity= now takes CPU index ranges in addition to
1064 * The various memory-related resource limit settings (such as
1065 LimitAS=) now understand the usual K, M, G, ... suffixes to
1066 the base of 1024 (IEC). Similar, the time-related resource
1067 limit settings understand the usual min, h, day, ...
1070 * There's a new system.conf setting DefaultTasksMax= to
1071 control the default TasksMax= setting for services and
1072 scopes running on the system. (TasksMax= is the primary
1073 setting that exposes the "pids" cgroup controller on systemd
1074 and was introduced in the previous systemd release.) The
1075 setting now defaults to 512, which means services that are
1076 not explicitly configured otherwise will only be able to
1077 create 512 processes or threads at maximum, from this
1078 version on. Note that this means that thread- or
1079 process-heavy services might need to be reconfigured to set
1080 TasksMax= to a higher value. It is sufficient to set
1081 TasksMax= in these specific unit files to a higher value, or
1082 even "infinity". Similar, there's now a logind.conf setting
1083 UserTasksMax= that defaults to 4096 and limits the total
1084 number of processes or tasks each user may own
1085 concurrently. nspawn containers also have the TasksMax=
1086 value set by default now, to 8192. Note that all of this
1087 only has an effect if the "pids" cgroup controller is
1088 enabled in the kernel. The general benefit of these changes
1089 should be a more robust and safer system, that provides a
1090 certain amount of per-service fork() bomb protection.
1092 * systemd-nspawn gained the new --network-veth-extra= switch
1093 to define additional and arbitrarily-named virtual Ethernet
1094 links between the host and the container.
1096 * A new service execution setting PassEnvironment= has been
1097 added that allows importing select environment variables
1098 from PID1's environment block into the environment block of
1101 * Timer units gained support for a new RemainAfterElapse=
1102 setting which takes a boolean argument. It defaults to on,
1103 exposing behaviour unchanged to previous releases. If set to
1104 off, timer units are unloaded after they elapsed if they
1105 cannot elapse again. This is particularly useful for
1106 transient timer units, which shall not stay around longer
1107 than until they first elapse.
1109 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
1110 default now (the kernel default is 16). This is beneficial
1111 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
1112 allows substantially larger numbers of queued
1113 datagrams. This should increase the capability of systemd to
1114 parallelize boot-up, as logging and sd_notify() are unlikely
1115 to stall execution anymore. If you need to change the value
1116 from the new defaults, use the usual sysctl.d/ snippets.
1118 * The compression framing format used by the journal or
1119 coredump processing has changed to be in line with what the
1120 official LZ4 tools generate. LZ4 compression support in
1121 systemd was considered unsupported previously, as the format
1122 was not compatible with the normal tools. With this release
1123 this has changed now, and it is hence safe for downstream
1124 distributions to turn it on. While not compressing as well
1125 as the XZ, LZ4 is substantially faster, which makes
1126 it a good default choice for the compression logic in the
1127 journal and in coredump handling.
1129 * Any reference to /etc/mtab has been dropped from
1130 systemd. The file has been obsolete since a while, but
1131 systemd refused to work on systems where it was incorrectly
1132 set up (it should be a symlink or non-existent). Please make
1133 sure to update to util-linux 2.27.1 or newer in conjunction
1134 with this systemd release, which also drops any reference to
1135 /etc/mtab. If you maintain a distribution make sure that no
1136 software you package still references it, as this is a
1137 likely source of bugs. There's also a glibc bug pending,
1138 asking for removal of any reference to this obsolete file:
1140 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
1142 Note that only util-linux versions built with
1143 --enable-libmount-force-mountinfo are supported.
1145 * Support for the ".snapshot" unit type has been removed. This
1146 feature turned out to be little useful and little used, and
1147 has now been removed from the core and from systemctl.
1149 * The dependency types RequiresOverridable= and
1150 RequisiteOverridable= have been removed from systemd. They
1151 have been used only very sparingly to our knowledge and
1152 other options that provide a similar effect (such as
1153 systemctl --mode=ignore-dependencies) are much more useful
1154 and commonly used. Moreover, they were only half-way
1155 implemented as the option to control behaviour regarding
1156 these dependencies was never added to systemctl. By removing
1157 these dependency types the execution engine becomes a bit
1158 simpler. Unit files that use these dependencies should be
1159 changed to use the non-Overridable dependency types
1160 instead. In fact, when parsing unit files with these
1161 options, that's what systemd will automatically convert them
1162 too, but it will also warn, asking users to fix the unit
1163 files accordingly. Removal of these dependency types should
1164 only affect a negligible number of unit files in the wild.
1166 * Behaviour of networkd's IPForward= option changed
1167 (again). It will no longer maintain a per-interface setting,
1168 but propagate one way from interfaces where this is enabled
1169 to the global kernel setting. The global setting will be
1170 enabled when requested by a network that is set up, but
1171 never be disabled again. This change was made to make sure
1172 IPv4 and IPv6 behaviour regarding packet forwarding is
1173 similar (as the Linux IPv6 stack does not support
1174 per-interface control of this setting) and to minimize
1177 * In unit files the behaviour of %u, %U, %h, %s has
1178 changed. These specifiers will now unconditionally resolve
1179 to the various user database fields of the user that the
1180 systemd instance is running as, instead of the user
1181 configured in the specific unit via User=. Note that this
1182 effectively doesn't change much, as resolving of these
1183 specifiers was already turned off in the --system instance
1184 of systemd, as we cannot do NSS lookups from PID 1. In the
1185 --user instance of systemd these specifiers where correctly
1186 resolved, but hardly made any sense, since the user instance
1187 lacks privileges to do user switches anyway, and User= is
1188 hence useless. Moreover, even in the --user instance of
1189 systemd behaviour was awkward as it would only take settings
1190 from User= assignment placed before the specifier into
1191 account. In order to unify and simplify the logic around
1192 this the specifiers will now always resolve to the
1193 credentials of the user invoking the manager (which in case
1194 of PID 1 is the root user).
1196 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
1197 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
1198 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
1199 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
1200 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
1201 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
1202 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
1203 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
1204 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
1205 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
1206 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
1207 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
1208 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
1209 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
1212 — Berlin, 2015-11-18
1216 * systemd now depends on util-linux v2.27. More specifically,
1217 the newly added mount monitor feature in libmount now
1218 replaces systemd's former own implementation.
1220 * libmount mandates /etc/mtab not to be regular file, and
1221 systemd now enforces this condition at early boot.
1222 /etc/mtab has been deprecated and warned about for a very
1223 long time, so systems running systemd should already have
1224 stopped having this file around as anything else than a
1225 symlink to /proc/self/mounts.
1227 * Support for the "pids" cgroup controller has been added. It
1228 allows accounting the number of tasks in a cgroup and
1229 enforcing limits on it. This adds two new setting
1230 TasksAccounting= and TasksMax= to each unit, as well as a
1231 global option DefaultTasksAccounting=.
1233 * Support for the "net_cls" cgroup controller has been added.
1234 It allows assigning a net class ID to each task in the
1235 cgroup, which can then be used in firewall rules and traffic
1236 shaping configurations. Note that the kernel netfilter net
1237 class code does not currently work reliably for ingress
1238 packets on unestablished sockets.
1240 This adds a new config directive called NetClass= to CGroup
1241 enabled units. Allowed values are positive numbers for fixed
1242 assignments and "auto" for picking a free value
1245 * 'systemctl is-system-running' now returns 'offline' if the
1246 system is not booted with systemd. This command can now be
1247 used as a substitute for 'systemd-notify --booted'.
1249 * Watchdog timeouts have been increased to 3 minutes for all
1250 in-tree service files. Apparently, disk IO issues are more
1251 frequent than we hoped, and user reported >1 minute waiting
1254 * 'machine-id-commit' functionality has been merged into
1255 'machine-id-setup --commit'. The separate binary has been
1258 * The WorkingDirectory= directive in unit files may now be set
1259 to the special value '~'. In this case, the working
1260 directory is set to the home directory of the user
1261 configured in User=.
1263 * "machinectl shell" will now open the shell in the home
1264 directory of the selected user by default.
1266 * The CrashChVT= configuration file setting is renamed to
1267 CrashChangeVT=, following our usual logic of not
1268 abbreviating unnecessarily. The old directive is still
1269 supported for compat reasons. Also, this directive now takes
1270 an integer value between 1 and 63, or a boolean value. The
1271 formerly supported '-1' value for disabling stays around for
1274 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
1275 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
1276 RootDirectory= properties can now be set for transient
1279 * The systemd-analyze tool gained a new "set-log-target" verb
1280 to change the logging target the system manager logs to
1281 dynamically during runtime. This is similar to how
1282 "systemd-analyze set-log-level" already changes the log
1285 * In nspawn /sys is now mounted as tmpfs, with only a selected
1286 set of subdirectories mounted in from the real sysfs. This
1287 enhances security slightly, and is useful for ensuring user
1288 namespaces work correctly.
1290 * Support for USB FunctionFS activation has been added. This
1291 allows implementation of USB gadget services that are
1292 activated as soon as they are requested, so that they don't
1293 have to run continuously, similar to classic socket
1296 * The "systemctl exit" command now optionally takes an
1297 additional parameter that sets the exit code to return from
1298 the systemd manager when exiting. This is only relevant when
1299 running the systemd user instance, or when running the
1300 system instance in a container.
1302 * sd-bus gained the new API calls sd_bus_path_encode_many()
1303 and sd_bus_path_decode_many() that allow easy encoding and
1304 decoding of multiple identifier strings inside a D-Bus
1305 object path. Another new call sd_bus_default_flush_close()
1306 has been added to flush and close per-thread default
1309 * systemd-cgtop gained support for a -M/--machine= switch to
1310 show the control groups within a certain container only.
1312 * "systemctl kill" gained support for an optional --fail
1313 switch. If specified the requested operation will fail of no
1314 processes have been killed, because the unit had no
1315 processes attached, or similar.
1317 * A new systemd.crash_reboot=1 kernel command line option has
1318 been added that triggers a reboot after crashing. This can
1319 also be set through CrashReboot= in systemd.conf.
1321 * The RuntimeDirectory= setting now understands unit
1322 specifiers like %i or %f.
1324 * A new (still internal) library API sd-ipv4acd has been added,
1325 that implements address conflict detection for IPv4. It's
1326 based on code from sd-ipv4ll, and will be useful for
1327 detecting DHCP address conflicts.
1329 * File descriptors passed during socket activation may now be
1330 named. A new API sd_listen_fds_with_names() is added to
1331 access the names. The default names may be overridden,
1332 either in the .socket file using the FileDescriptorName=
1333 parameter, or by passing FDNAME= when storing the file
1334 descriptors using sd_notify().
1336 * systemd-networkd gained support for:
1338 - Setting the IPv6 Router Advertisement settings via
1339 IPv6AcceptRouterAdvertisements= in .network files.
1341 - Configuring the HelloTimeSec=, MaxAgeSec= and
1342 ForwardDelaySec= bridge parameters in .netdev files.
1344 - Configuring PreferredSource= for static routes in
1347 * The "ask-password" framework used to query for LUKS harddisk
1348 passwords or SSL passwords during boot gained support for
1349 caching passwords in the kernel keyring, if it is
1350 available. This makes sure that the user only has to type in
1351 a passphrase once if there are multiple objects to unlock
1352 with the same one. Previously, such password caching was
1353 available only when Plymouth was used; this moves the
1354 caching logic into the systemd codebase itself. The
1355 "systemd-ask-password" utility gained a new --keyname=
1356 switch to control which kernel keyring key to use for
1357 caching a password in. This functionality is also useful for
1358 enabling display managers such as gdm to automatically
1359 unlock the user's GNOME keyring if its passphrase, the
1360 user's password and the harddisk password are the same, if
1361 gdm-autologin is used.
1363 * When downloading tar or raw images using "machinectl
1364 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
1365 file is now also downloaded, if it is available and stored
1366 next to the image file.
1368 * Units of type ".socket" gained a new boolean setting
1369 Writable= which is only useful in conjunction with
1370 ListenSpecial=. If true, enables opening the specified
1371 special file in O_RDWR mode rather than O_RDONLY mode.
1373 * systemd-rfkill has been reworked to become a singleton
1374 service that is activated through /dev/rfkill on each rfkill
1375 state change and saves the settings to disk. This way,
1376 systemd-rfkill is now compatible with devices that exist
1377 only intermittendly, and even restores state if the previous
1378 system shutdown was abrupt rather than clean.
1380 * The journal daemon gained support for vacuuming old journal
1381 files controlled by the number of files that shall remain,
1382 in addition to the already existing control by size and by
1383 date. This is useful as journal interleaving performance
1384 degrades with too many separate journal files, and allows
1385 putting an effective limit on them. The new setting defaults
1386 to 100, but this may be changed by setting SystemMaxFiles=
1387 and RuntimeMaxFiles= in journald.conf. Also, the
1388 "journalctl" tool gained the new --vacuum-files= switch to
1389 manually vacuum journal files to leave only the specified
1390 number of files in place.
1392 * udev will now create /dev/disk/by-path links for ATA devices
1393 on kernels where that is supported.
1395 * Galician, Serbian, Turkish and Korean translations were added.
1397 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
1398 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
1399 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
1400 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
1401 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
1402 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
1403 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
1404 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
1405 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
1406 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
1407 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
1408 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
1409 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
1410 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
1411 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
1412 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
1413 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
1414 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
1416 — Berlin, 2015-10-07
1420 * The DHCP implementation of systemd-networkd gained a set of
1423 - The DHCP server now supports emitting DNS and NTP
1424 information. It may be enabled and configured via
1425 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
1426 and NTP information is enabled, but no servers are
1427 configured, the corresponding uplink information (if there
1428 is any) is propagated.
1430 - Server and client now support transmission and reception
1431 of timezone information. It can be configured via the
1432 newly introduced network options UseTimezone=,
1433 EmitTimezone=, and Timezone=. Transmission of timezone
1434 information is enabled between host and containers by
1435 default now: the container will change its local timezone
1436 to what the host has set.
1438 - Lease timeouts can now be configured via
1439 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
1441 - The DHCP server improved on the stability of
1442 leases. Clients are more likely to get the same lease
1443 information back, even if the server loses state.
1445 - The DHCP server supports two new configuration options to
1446 control the lease address pool metrics, PoolOffset= and
1449 * The encapsulation limit of tunnels in systemd-networkd may
1450 now be configured via 'EncapsulationLimit='. It allows
1451 modifying the maximum additional levels of encapsulation
1452 that are permitted to be prepended to a packet.
1454 * systemd now supports the concept of user buses replacing
1455 session buses, if used with dbus-1.10 (and enabled via dbus
1456 --enable-user-session). It previously only supported this on
1457 kdbus-enabled systems, and this release expands this to
1458 'dbus-daemon' systems.
1460 * systemd-networkd now supports predictable interface names
1463 * systemd now optionally supports the new Linux kernel
1464 "unified" control group hierarchy. If enabled via the kernel
1465 command-line option 'systemd.unified_cgroup_hierarchy=1',
1466 systemd will try to mount the unified cgroup hierarchy
1467 directly on /sys/fs/cgroup. If not enabled, or not
1468 available, systemd will fall back to the legacy cgroup
1469 hierarchy setup, as before. Host system and containers can
1470 mix and match legacy and unified hierarchies as they
1471 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
1472 environment variable to individually select the hierarchy to
1473 use for executed containers. By default, nspawn will use the
1474 unified hierarchy for the containers if the host uses the
1475 unified hierarchy, and the legacy hierarchy otherwise.
1476 Please note that at this point the unified hierarchy is an
1477 experimental kernel feature and is likely to change in one
1478 of the next kernel releases. Therefore, it should not be
1479 enabled by default in downstream distributions yet. The
1480 minimum required kernel version for the unified hierarchy to
1481 work is 4.2. Note that when the unified hierarchy is used
1482 for the first time delegated access to controllers is
1483 safe. Because of this systemd-nspawn containers will get
1484 access to controllers now, as will systemd user
1485 sessions. This means containers and user sessions may now
1486 manage their own resources, partitioning up what the system
1489 * A new special scope unit "init.scope" has been introduced
1490 that encapsulates PID 1 of the system. It may be used to
1491 determine resource usage and enforce resource limits on PID
1492 1 itself. PID 1 hence moved out of the root of the control
1495 * The cgtop tool gained support for filtering out kernel
1496 threads when counting tasks in a control group. Also, the
1497 count of processes is now recursively summed up by
1498 default. Two options -k and --recursive= have been added to
1499 revert to old behaviour. The tool has also been updated to
1500 work correctly in containers now.
1502 * systemd-nspawn's --bind= and --bind-ro= options have been
1503 extended to allow creation of non-recursive bind mounts.
1505 * libsystemd gained two new calls sd_pid_get_cgroup() and
1506 sd_peer_get_cgroup() which return the control group path of
1507 a process or peer of a connected AF_UNIX socket. This
1508 function call is particularly useful when implementing
1509 delegated subtrees support in the control group hierarchy.
1511 * The "sd-event" event loop API of libsystemd now supports
1512 correct dequeuing of real-time signals, without losing
1515 * When systemd requests a PolicyKit decision when managing
1516 units it will now add additional fields to the request,
1517 including unit name and desired operation. This enables more
1518 powerful PolicyKit policies, that make decisions depending
1519 on these parameters.
1521 * nspawn learnt support for .nspawn settings files, that may
1522 accompany the image files or directories of containers, and
1523 may contain additional settings for the container. This is
1524 an alternative to configuring container parameters via the
1525 nspawn command line.
1527 Contributions from: Cristian Rodríguez, Daniel Mack, David
1528 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
1529 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
1530 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
1531 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
1532 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
1533 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
1534 Andersen, Tom Gundersen, Torstein Husebø
1536 — Berlin, 2015-09-08
1540 * machinectl gained a new verb 'shell' which opens a fresh
1541 shell on the target container or the host. It is similar to
1542 the existing 'login' command of machinectl, but spawns the
1543 shell directly without prompting for username or
1544 password. The pseudo machine '.host' now refers to the local
1545 host and is used by default. Hence, 'machinectl shell' can
1546 be used as replacement for 'su -' which spawns a session as
1547 a fresh systemd unit in a way that is fully isolated from
1548 the originating session.
1550 * systemd-networkd learned to cope with private-zone DHCP
1551 options and allows other programs to query the values.
1553 * SELinux access control when enabling/disabling units is no
1554 longer enforced with this release. The previous
1555 implementation was incorrect, and a new corrected
1556 implementation is not yet available. As unit file operations
1557 are still protected via PolicyKit and D-Bus policy this is
1558 not a security problem. Yet, distributions which care about
1559 optimal SELinux support should probably not stabilize on
1562 * sd-bus gained support for matches of type "arg0has=", that
1563 test for membership of strings in string arrays sent in bus
1566 * systemd-resolved now dumps the contents of its DNS and LLMNR
1567 caches to the logs on reception of the SIGUSR1 signal. This
1568 is useful to debug DNS behaviour.
1570 * The coredumpctl tool gained a new --directory= option to
1571 operate on journal files in a specific directory.
1573 * "systemctl reboot" and related commands gained a new
1574 "--message=" option which may be used to set a free-text
1575 wall message when shutting down or rebooting the
1576 system. This message is also logged, which is useful for
1577 figuring out the reason for a reboot or shutdown a
1580 * The "systemd-resolve-host" tool's -i switch now takes
1581 network interface numbers as alternative to interface names.
1583 * A new unit file setting for services has been introduced:
1584 UtmpMode= allows configuration of how precisely systemd
1585 handles utmp and wtmp entries for the service if this is
1586 enabled. This allows writing services that appear similar to
1587 user sessions in the output of the "w", "who", "last" and
1590 * systemd-resolved will now locally synthesize DNS resource
1591 records for the "localhost" and "gateway" domains as well as
1592 the local hostname. This should ensure that clients querying
1593 RRs via resolved will get similar results as those going via
1594 NSS, if nss-myhostname is enabled.
1596 Contributions from: Alastair Hughes, Alex Crawford, Daniel
1597 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
1598 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
1599 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
1600 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
1601 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
1602 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
1603 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
1604 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
1605 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
1606 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
1607 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
1609 — Berlin, 2015-08-27
1613 * The systemd-efi-boot-generator functionality was merged into
1614 systemd-gpt-auto-generator.
1616 * systemd-networkd now supports Group Policy for vxlan
1617 devices. It can be enabled via the new boolean configuration
1618 option called 'GroupPolicyExtension='.
1620 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
1621 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
1622 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
1624 — Berlin, 2015-07-31
1628 * The python-systemd code has been removed from the systemd repository.
1629 A new repository has been created which accommodates the code from
1630 now on, and we kindly ask distributions to create a separate package
1631 for this: https://github.com/systemd/python-systemd
1633 * The systemd daemon will now reload its main configuration
1634 (/etc/systemd/system.conf) on daemon-reload.
1636 * sd-dhcp now exposes vendor specific extensions via
1637 sd_dhcp_lease_get_vendor_specific().
1639 * systemd-networkd gained a number of new configuration options.
1641 - A new boolean configuration option for TAP devices called
1642 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
1643 device, thus allowing to send and receive GSO packets.
1645 - A new tunnel configuration option called 'CopyDSCP='.
1646 If enabled, the DSCP field of ip6 tunnels is copied into the
1647 decapsulated packet.
1649 - A set of boolean bridge configuration options were added.
1650 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
1651 and 'UnicastFlood=' are now parsed by networkd and applied to the
1652 respective bridge link device via the respective IFLA_BRPORT_*
1655 - A new string configuration option to override the hostname sent
1656 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
1657 is true, networkd will use the configured hostname instead of the
1658 system hostname when sending DHCP requests.
1660 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
1661 networkd will configure the IPv6 flow-label of the tunnel device
1662 according to RFC2460.
1664 - The 'macvtap' virtual network devices are now supported, similar to
1665 the already supported 'macvlan' devices.
1667 * systemd-resolved now implements RFC5452 to improve resilience against
1668 cache poisoning. Additionally, source port randomization is enabled
1669 by default to further protect against DNS spoofing attacks.
1671 * nss-mymachines now supports translating UIDs and GIDs of running
1672 containers with user-namespaces enabled. If a container 'foo'
1673 translates a host uid 'UID' to the container uid 'TUID', then
1674 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
1675 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
1676 mapped as 'vg-foo-TGID'.
1678 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
1679 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
1680 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
1681 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
1682 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
1683 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
1684 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
1685 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
1686 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
1687 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
1689 — Berlin, 2015-07-29
1693 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
1694 There are no known issues with current sysfs, and udev does not need
1695 or should be used to work around such bugs.
1697 * udev does no longer enable USB HID power management. Several reports
1698 indicate, that some devices cannot handle that setting.
1700 * The udev accelerometer helper was removed. The functionality
1701 is now fully included in iio-sensor-proxy. But this means,
1702 older iio-sensor-proxy versions will no longer provide
1703 accelerometer/orientation data with this systemd version.
1704 Please upgrade iio-sensor-proxy to version 1.0.
1706 * networkd gained a new configuration option IPv6PrivacyExtensions=
1707 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
1708 for Stateless Address") on selected networks.
1710 * For the sake of fewer build-time dependencies and less code in the
1711 main repository, the python bindings are about to be removed in the
1712 next release. A new repository has been created which accommodates
1713 the code from now on, and we kindly ask distributions to create a
1714 separate package for this. The removal will take place in v223.
1716 https://github.com/systemd/python-systemd
1718 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
1719 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
1720 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
1721 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
1722 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
1723 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
1724 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
1725 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
1726 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
1727 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
1729 — Berlin, 2015-07-07
1733 * The sd-bus.h and sd-event.h APIs have now been declared
1734 stable and have been added to the official interface of
1735 libsystemd.so. sd-bus implements an alternative D-Bus client
1736 library, that is relatively easy to use, very efficient and
1737 supports both classic D-Bus as well as kdbus as transport
1738 backend. sd-event is a generic event loop abstraction that
1739 is built around Linux epoll, but adds features such as event
1740 prioritization or efficient timer handling. Both APIs are good
1741 choices for C programs looking for a bus and/or event loop
1742 implementation that is minimal and does not have to be
1743 portable to other kernels.
1745 * kdbus support is no longer compile-time optional. It is now
1746 always built-in. However, it can still be disabled at
1747 runtime using the kdbus=0 kernel command line setting, and
1748 that setting may be changed to default to off, by specifying
1749 --disable-kdbus at build-time. Note though that the kernel
1750 command line setting has no effect if the kdbus.ko kernel
1751 module is not installed, in which case kdbus is (obviously)
1752 also disabled. We encourage all downstream distributions to
1753 begin testing kdbus by adding it to the kernel images in the
1754 development distributions, and leaving kdbus support in
1757 * The minimal required util-linux version has been bumped to
1760 * Support for chkconfig (--enable-chkconfig) was removed in
1761 favor of calling an abstraction tool
1762 /lib/systemd/systemd-sysv-install. This needs to be
1763 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
1764 in README for details.
1766 * If there's a systemd unit and a SysV init script for the
1767 same service name, and the user executes "systemctl enable"
1768 for it (or a related call), then this will now enable both
1769 (or execute the related operation on both), not just the
1772 * The libudev API documentation has been converted from gtkdoc
1775 * gudev has been removed from the systemd tree, it is now an
1778 * The systemd-cgtop tool learnt a new --raw switch to generate
1779 "raw" (machine parsable) output.
1781 * networkd's IPForwarding= .network file setting learnt the
1782 new setting "kernel", which ensures that networkd does not
1783 change the IP forwarding sysctl from the default kernel
1786 * The systemd-logind bus API now exposes a new boolean
1787 property "Docked" that reports whether logind considers the
1788 system "docked", i.e. connected to a docking station or not.
1790 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
1791 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
1792 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
1793 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
1794 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
1795 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
1796 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
1797 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
1798 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
1799 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
1800 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
1801 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
1802 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
1803 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
1804 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
1805 Fink, Zbigniew Jędrzejewski-Szmek
1807 — Berlin, 2015-06-19
1811 * The gudev library has been extracted into a separate repository
1812 available at: https://git.gnome.org/browse/libgudev/
1813 It is now managed as part of the Gnome project. Distributions
1814 are recommended to pass --disable-gudev to systemd and use
1815 gudev from the Gnome project instead. gudev is still included
1816 in systemd, for now. It will be removed soon, though. Please
1817 also see the announcement-thread on systemd-devel:
1818 http://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
1820 * systemd now exposes a CPUUsageNSec= property for each
1821 service unit on the bus, that contains the overall consumed
1822 CPU time of a service (the sum of what each process of the
1823 service consumed). This value is only available if
1824 CPUAccounting= is turned on for a service, and is then shown
1825 in the "systemctl status" output.
1827 * Support for configuring alternative mappings of the old SysV
1828 runlevels to systemd targets has been removed. They are now
1829 hardcoded in a way that runlevels 2, 3, 4 all map to
1830 multi-user.target and 5 to graphical.target (which
1831 previously was already the default behaviour).
1833 * The auto-mounter logic gained support for mount point
1834 expiry, using a new TimeoutIdleSec= setting in .automount
1835 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
1837 * The EFI System Partition (ESP) as mounted to /boot by
1838 systemd-efi-boot-generator will now be unmounted
1839 automatically after 2 minutes of not being used. This should
1840 minimize the risk of ESP corruptions.
1842 * New /etc/fstab options x-systemd.requires= and
1843 x-systemd.requires-mounts-for= are now supported to express
1844 additional dependencies for mounts. This is useful for
1845 journalling file systems that support external journal
1846 devices or overlay file systems that require underlying file
1847 systems to be mounted.
1849 * systemd does not support direct live-upgrades (via systemctl
1850 daemon-reexec) from versions older than v44 anymore. As no
1851 distribution we are aware of shipped such old versions in a
1852 stable release this should not be problematic.
1854 * When systemd forks off a new per-connection service instance
1855 it will now set the $REMOTE_ADDR environment variable to the
1856 remote IP address, and $REMOTE_PORT environment variable to
1857 the remote IP port. This behaviour is similar to the
1858 corresponding environment variables defined by CGI.
1860 * systemd-networkd gained support for uplink failure
1861 detection. The BindCarrier= option allows binding interface
1862 configuration dynamically to the link sense of other
1863 interfaces. This is useful to achieve behaviour like in
1866 * systemd-networkd gained support for configuring the DHCP
1867 client identifier to use when requesting leases.
1869 * systemd-networkd now has a per-network UseNTP= option to
1870 configure whether NTP server information acquired via DHCP
1871 is passed on to services like systemd-timesyncd.
1873 * systemd-networkd gained support for vti6 tunnels.
1875 * Note that systemd-networkd manages the sysctl variable
1876 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
1877 it is configured for since v219. The variable controls IP
1878 forwarding, and is a per-interface alternative to the global
1879 /proc/sys/net/ipv[46]/ip_forward. This setting is
1880 configurable in the IPForward= option, which defaults to
1881 "no". This means if networkd is used for an interface it is
1882 no longer sufficient to set the global sysctl option to turn
1883 on IP forwarding! Instead, the .network file option
1884 IPForward= needs to be turned on! Note that the
1885 implementation of this behaviour was broken in v219 and has
1888 * Many bonding and vxlan options are now configurable in
1891 * systemd-nspawn gained a new --property= setting to set unit
1892 properties for the container scope. This is useful for
1893 setting resource parameters (e.g. "CPUShares=500") on
1894 containers started from the command line.
1896 * systemd-nspawn gained a new --private-users= switch to make
1897 use of user namespacing available on recent Linux kernels.
1899 * systemd-nspawn may now be called as part of a shell pipeline
1900 in which case the pipes used for stdin and stdout are passed
1901 directly to the process invoked in the container, without
1902 indirection via a pseudo tty.
1904 * systemd-nspawn gained a new switch to control the UNIX
1905 signal to use when killing the init process of the container
1908 * systemd-nspawn gained a new --overlay= switch for mounting
1909 overlay file systems into the container using the new kernel
1912 * When a container image is imported via systemd-importd and
1913 the host file system is not btrfs, a loopback block device
1914 file is created in /var/lib/machines.raw with a btrfs file
1915 system inside. It is then mounted to /var/lib/machines to
1916 enable btrfs features for container management. The loopback
1917 file and btrfs file system is grown as needed when container
1918 images are imported via systemd-importd.
1920 * systemd-machined/systemd-importd gained support for btrfs
1921 quota, to enforce container disk space limits on disk. This
1922 is exposed in "machinectl set-limit".
1924 * systemd-importd now can import containers from local .tar,
1925 .raw and .qcow2 images, and export them to .tar and .raw. It
1926 can also import dkr v2 images now from the network (on top
1929 * systemd-importd gained support for verifying downloaded
1930 images with gpg2 (previously only gpg1 was supported).
1932 * systemd-machined, systemd-logind, systemd: most bus calls
1933 are now accessible to unprivileged processes via
1934 PolicyKit. Also, systemd-logind will now allow users to kill
1935 their own sessions without further privileges or
1938 * systemd-shutdownd has been removed. This service was
1939 previously responsible for implementing scheduled shutdowns
1940 as exposed in /usr/bin/shutdown's time parameter. This
1941 functionality has now been moved into systemd-logind and is
1942 accessible via a bus interface.
1944 * "systemctl reboot" gained a new switch --firmware-setup that
1945 can be used to reboot into the EFI firmware setup, if that
1946 is available. systemd-logind now exposes an API on the bus
1947 to trigger such reboots, in case graphical desktop UIs want
1948 to cover this functionality.
1950 * "systemctl enable", "systemctl disable" and "systemctl mask"
1951 now support a new "--now" switch. If specified the units
1952 that are enabled will also be started, and the ones
1953 disabled/masked also stopped.
1955 * The Gummiboot EFI boot loader tool has been merged into
1956 systemd, and renamed to "systemd-boot". The bootctl tool has been
1957 updated to support systemd-boot.
1959 * An EFI kernel stub has been added that may be used to create
1960 kernel EFI binaries that contain not only the actual kernel,
1961 but also an initrd, boot splash, command line and OS release
1962 information. This combined binary can then be signed as a
1963 single image, so that the firmware can verify it all in one
1964 step. systemd-boot has special support for EFI binaries created
1965 like this and can extract OS release information from them
1966 and show them in the boot menu. This functionality is useful
1967 to implement cryptographically verified boot schemes.
1969 * Optional support has been added to systemd-fsck to pass
1970 fsck's progress report to an AF_UNIX socket in the file
1973 * udev will no longer create device symlinks for all block
1974 devices by default. A blacklist for excluding special block
1975 devices from this logic has been turned into a whitelist
1976 that requires picking block devices explicitly that require
1979 * A new (currently still internal) API sd-device.h has been
1980 added to libsystemd. This modernized API is supposed to
1981 replace libudev eventually. In fact, already much of libudev
1982 is now just a wrapper around sd-device.h.
1984 * A new hwdb database for storing metadata about pointing
1985 stick devices has been added.
1987 * systemd-tmpfiles gained support for setting file attributes
1988 similar to the "chattr" tool with new 'h' and 'H' lines.
1990 * systemd-journald will no longer unconditionally set the
1991 btrfs NOCOW flag on new journal files. This is instead done
1992 with tmpfiles snippet using the new 'h' line type. This
1993 allows easy disabling of this logic, by masking the
1994 journal-nocow.conf tmpfiles file.
1996 * systemd-journald will now translate audit message types to
1997 human readable identifiers when writing them to the
1998 journal. This should improve readability of audit messages.
2000 * The LUKS logic gained support for the offset= and skip=
2001 options in /etc/crypttab, as previously implemented by
2004 * /usr/lib/os-release gained a new optional field VARIANT= for
2005 distributions that support multiple variants (such as a
2006 desktop edition, a server edition, ...)
2008 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
2009 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
2010 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
2011 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
2012 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
2013 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
2014 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
2015 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
2016 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
2017 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
2018 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
2019 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
2020 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
2021 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
2022 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
2023 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
2024 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
2025 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
2026 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
2027 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
2028 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
2029 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
2030 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
2031 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
2032 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
2033 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
2034 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
2036 — Berlin, 2015-05-22
2040 * Introduce a new API "sd-hwdb.h" for querying the hardware
2041 metadata database. With this minimal interface one can query
2042 and enumerate the udev hwdb, decoupled from the old libudev
2043 library. libudev's interface for this is now only a wrapper
2044 around sd-hwdb. A new tool systemd-hwdb has been added to
2045 interface with and update the database.
2047 * When any of systemd's tools copies files (for example due to
2048 tmpfiles' C lines) a btrfs reflink will attempted first,
2049 before bytewise copying is done.
2051 * systemd-nspawn gained a new --ephemeral switch. When
2052 specified a btrfs snapshot is taken of the container's root
2053 directory, and immediately removed when the container
2054 terminates again. Thus, a container can be started whose
2055 changes never alter the container's root directory, and are
2056 lost on container termination. This switch can also be used
2057 for starting a container off the root file system of the
2058 host without affecting the host OS. This switch is only
2059 available on btrfs file systems.
2061 * systemd-nspawn gained a new --template= switch. It takes the
2062 path to a container tree to use as template for the tree
2063 specified via --directory=, should that directory be
2064 missing. This allows instantiating containers dynamically,
2065 on first run. This switch is only available on btrfs file
2068 * When a .mount unit refers to a mount point on which multiple
2069 mounts are stacked, and the .mount unit is stopped all of
2070 the stacked mount points will now be unmounted until no
2071 mount point remains.
2073 * systemd now has an explicit notion of supported and
2074 unsupported unit types. Jobs enqueued for unsupported unit
2075 types will now fail with an "unsupported" error code. More
2076 specifically .swap, .automount and .device units are not
2077 supported in containers, .busname units are not supported on
2078 non-kdbus systems. .swap and .automount are also not
2079 supported if their respective kernel compile time options
2082 * machinectl gained support for two new "copy-from" and
2083 "copy-to" commands for copying files from a running
2084 container to the host or vice versa.
2086 * machinectl gained support for a new "bind" command to bind
2087 mount host directories into local containers. This is
2088 currently only supported for nspawn containers.
2090 * networkd gained support for configuring bridge forwarding
2091 database entries (fdb) from .network files.
2093 * A new tiny daemon "systemd-importd" has been added that can
2094 download container images in tar, raw, qcow2 or dkr formats,
2095 and make them available locally in /var/lib/machines, so
2096 that they can run as nspawn containers. The daemon can GPG
2097 verify the downloads (not supported for dkr, since it has no
2098 provisions for verifying downloads). It will transparently
2099 decompress bz2, xz, gzip compressed downloads if necessary,
2100 and restore sparse files on disk. The daemon uses privilege
2101 separation to ensure the actual download logic runs with
2102 fewer privileges than the daemon itself. machinectl has
2103 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
2104 make the functionality of importd available to the
2105 user. With this in place the Fedora and Ubuntu "Cloud"
2106 images can be downloaded and booted as containers unmodified
2107 (the Fedora images lack the appropriate GPG signature files
2108 currently, so they cannot be verified, but this will change
2109 soon, hopefully). Note that downloading images is currently
2110 only fully supported on btrfs.
2112 * machinectl is now able to list container images found in
2113 /var/lib/machines, along with some metadata about sizes of
2114 disk and similar. If the directory is located on btrfs and
2115 quota is enabled, this includes quota display. A new command
2116 "image-status" has been added that shows additional
2117 information about images.
2119 * machinectl is now able to clone container images
2120 efficiently, if the underlying file system (btrfs) supports
2121 it, with the new "machinectl clone" command. It also
2122 gained commands for renaming and removing images, as well as
2123 marking them read-only or read-write (supported also on
2124 legacy file systems).
2126 * networkd gained support for collecting LLDP network
2127 announcements, from hardware that supports this. This is
2128 shown in networkctl output.
2130 * systemd-run gained support for a new -t (--pty) switch for
2131 invoking a binary on a pty whose input and output is
2132 connected to the invoking terminal. This allows executing
2133 processes as system services while interactively
2134 communicating with them via the terminal. Most interestingly
2135 this is supported across container boundaries. Invoking
2136 "systemd-run -t /bin/bash" is an alternative to running a
2137 full login session, the difference being that the former
2138 will not register a session, nor go through the PAM session
2141 * tmpfiles gained support for a new "v" line type for creating
2142 btrfs subvolumes. If the underlying file system is a legacy
2143 file system, this automatically degrades to creating a
2144 normal directory. Among others /var/lib/machines is now
2145 created like this at boot, should it be missing.
2147 * The directory /var/lib/containers/ has been deprecated and
2148 been replaced by /var/lib/machines. The term "machines" has
2149 been used in the systemd context as generic term for both
2150 VMs and containers, and hence appears more appropriate for
2151 this, as the directory can also contain raw images bootable
2154 * systemd-nspawn when invoked with -M but without --directory=
2155 or --image= is now capable of searching for the container
2156 root directory, subvolume or disk image automatically, in
2157 /var/lib/machines. systemd-nspawn@.service has been updated
2158 to make use of this, thus allowing it to be used for raw
2161 * A new machines.target unit has been introduced that is
2162 supposed to group all containers/VMs invoked as services on
2163 the system. systemd-nspawn@.service has been updated to
2164 integrate with that.
2166 * machinectl gained a new "start" command, for invoking a
2167 container as a service. "machinectl start foo" is mostly
2168 equivalent to "systemctl start systemd-nspawn@foo.service",
2169 but handles escaping in a nicer way.
2171 * systemd-nspawn will now mount most of the cgroupfs tree
2172 read-only into each container, with the exception of the
2173 container's own subtree in the name=systemd hierarchy.
2175 * journald now sets the special FS_NOCOW file flag for its
2176 journal files. This should improve performance on btrfs, by
2177 avoiding heavy fragmentation when journald's write-pattern
2178 is used on COW file systems. It degrades btrfs' data
2179 integrity guarantees for the files to the same levels as for
2180 ext3/ext4 however. This should be OK though as journald does
2181 its own data integrity checks and all its objects are
2182 checksummed on disk. Also, journald should handle btrfs disk
2183 full events a lot more gracefully now, by processing SIGBUS
2184 errors, and not relying on fallocate() anymore.
2186 * When journald detects that journal files it is writing to
2187 have been deleted it will immediately start new journal
2190 * systemd now provides a way to store file descriptors
2191 per-service in PID 1. This is useful for daemons to ensure
2192 that fds they require are not lost during a daemon
2193 restart. The fds are passed to the daemon on the next
2194 invocation in the same way socket activation fds are
2195 passed. This is now used by journald to ensure that the
2196 various sockets connected to all the system's stdout/stderr
2197 are not lost when journald is restarted. File descriptors
2198 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
2199 an extension to sd_notify(). Note that a limit is enforced
2200 on the number of fds a service can store in PID 1, and it
2201 defaults to 0, so that no fds may be stored, unless this is
2202 explicitly turned on.
2204 * The default TERM variable to use for units connected to a
2205 terminal, when no other value is explicitly is set is now
2206 vt220 rather than vt102. This should be fairly safe still,
2207 but allows PgUp/PgDn work.
2209 * The /etc/crypttab option header= as known from Debian is now
2212 * "loginctl user-status" and "loginctl session-status" will
2213 now show the last 10 lines of log messages of the
2214 user/session following the status output. Similar,
2215 "machinectl status" will show the last 10 log lines
2216 associated with a virtual machine or container
2217 service. (Note that this is usually not the log messages
2218 done in the VM/container itself, but simply what the
2219 container manager logs. For nspawn this includes all console
2222 * "loginctl session-status" without further argument will now
2223 show the status of the session of the caller. Similar,
2224 "lock-session", "unlock-session", "activate",
2225 "enable-linger", "disable-linger" may now be called without
2226 session/user parameter in which case they apply to the
2227 caller's session/user.
2229 * An X11 session scriptlet is now shipped that uploads
2230 $DISPLAY and $XAUTHORITY into the environment of the systemd
2231 --user daemon if a session begins. This should improve
2232 compatibility with X11 enabled applications run as systemd
2235 * Generators are now subject to masking via /etc and /run, the
2236 same way as unit files.
2238 * networkd .network files gained support for configuring
2239 per-link IPv4/IPv6 packet forwarding as well as IPv4
2240 masquerading. This is by default turned on for veth links to
2241 containers, as registered by systemd-nspawn. This means that
2242 nspawn containers run with --network-veth will now get
2243 automatic routed access to the host's networks without any
2244 further configuration or setup, as long as networkd runs on
2247 * systemd-nspawn gained the --port= (-p) switch to expose TCP
2248 or UDP posts of a container on the host. With this in place
2249 it is possible to run containers with private veth links
2250 (--network-veth), and have their functionality exposed on
2251 the host as if their services were running directly on the
2254 * systemd-nspawn's --network-veth switch now gained a short
2255 version "-n", since with the changes above it is now truly
2256 useful out-of-the-box. The systemd-nspawn@.service has been
2257 updated to make use of it too by default.
2259 * systemd-nspawn will now maintain a per-image R/W lock, to
2260 ensure that the same image is not started more than once
2261 writable. (It's OK to run an image multiple times
2262 simultaneously in read-only mode.)
2264 * systemd-nspawn's --image= option is now capable of
2265 dissecting and booting MBR and GPT disk images that contain
2266 only a single active Linux partition. Previously it
2267 supported only GPT disk images with proper GPT type
2268 IDs. This allows running cloud images from major
2269 distributions directly with systemd-nspawn, without
2272 * In addition to collecting mouse dpi data in the udev
2273 hardware database, there's now support for collecting angle
2274 information for mouse scroll wheels. The database is
2275 supposed to guarantee similar scrolling behavior on mice
2276 that it knows about. There's also support for collecting
2277 information about Touchpad types.
2279 * udev's input_id built-in will now also collect touch screen
2280 dimension data and attach it to probed devices.
2282 * /etc/os-release gained support for a Distribution Privacy
2285 * networkd gained support for creating "ipvlan", "gretap",
2286 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
2288 * systemd-tmpfiles gained support for "a" lines for setting
2291 * systemd-nspawn will now mount /tmp in the container to
2292 tmpfs, automatically.
2294 * systemd now exposes the memory.usage_in_bytes cgroup
2295 attribute and shows it for each service in the "systemctl
2296 status" output, if available.
2298 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
2299 immediate reboot is triggered. This useful if shutdown is
2300 hung and is unable to complete, to expedite the
2301 operation. Note that this kind of reboot will still unmount
2302 all file systems, and hence should not result in fsck being
2305 * A .device unit for an optical block device will now be
2306 considered active only when a medium is in the drive. Also,
2307 mount units are now bound to their backing devices thus
2308 triggering automatic unmounting when devices become
2309 unavailable. With this in place systemd will now
2310 automatically unmount left-over mounts when a CD-ROM is
2311 ejected or an USB stick is yanked from the system.
2313 * networkd-wait-online now has support for waiting for
2314 specific interfaces only (with globbing), and for giving up
2315 after a configurable timeout.
2317 * networkd now exits when idle. It will be automatically
2318 restarted as soon as interfaces show up, are removed or
2319 change state. networkd will stay around as long as there is
2320 at least one DHCP state machine or similar around, that keep
2323 * networkd may now configure IPv6 link-local addressing in
2324 addition to IPv4 link-local addressing.
2326 * The IPv6 "token" for use in SLAAC may now be configured for
2327 each .network interface in networkd.
2329 * Routes configured with networkd may now be assigned a scope
2332 * networkd's [Match] sections now support globbing and lists
2333 of multiple space-separated matches per item.
2335 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
2336 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
2337 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
2338 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
2339 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
2340 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
2341 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
2342 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
2343 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
2344 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
2345 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
2346 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
2347 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
2348 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
2349 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
2350 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
2351 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
2352 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
2353 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
2354 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
2355 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
2356 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
2357 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
2358 Hoffmann, Zbigniew Jędrzejewski-Szmek
2360 — Berlin, 2015-02-16
2364 * When querying unit file enablement status (for example via
2365 "systemctl is-enabled"), a new state "indirect" is now known
2366 which indicates that a unit might not be enabled itself, but
2367 another unit listed in its Also= setting might be.
2369 * Similar to the various existing ConditionXYZ= settings for
2370 units, there are now matching AssertXYZ= settings. While
2371 failing conditions cause a unit to be skipped, but its job
2372 to succeed, failing assertions declared like this will cause
2373 a unit start operation and its job to fail.
2375 * hostnamed now knows a new chassis type "embedded".
2377 * systemctl gained a new "edit" command. When used on a unit
2378 file, this allows extending unit files with .d/ drop-in
2379 configuration snippets or editing the full file (after
2380 copying it from /usr/lib to /etc). This will invoke the
2381 user's editor (as configured with $EDITOR), and reload the
2382 modified configuration after editing.
2384 * "systemctl status" now shows the suggested enablement state
2385 for a unit, as declared in the (usually vendor-supplied)
2386 system preset files.
2388 * nss-myhostname will now resolve the single-label host name
2389 "gateway" to the locally configured default IP routing
2390 gateways, ordered by their metrics. This assigns a stable
2391 name to the used gateways, regardless which ones are
2392 currently configured. Note that the name will only be
2393 resolved after all other name sources (if nss-myhostname is
2394 configured properly) and should hence not negatively impact
2395 systems that use the single-label host name "gateway" in
2398 * systemd-inhibit now allows filtering by mode when listing
2401 * Scope and service units gained a new "Delegate" boolean
2402 property, which, when set, allows processes running inside the
2403 unit to further partition resources. This is primarily
2404 useful for systemd user instances as well as container
2407 * journald will now pick up audit messages directly from
2408 the kernel, and log them like any other log message. The
2409 audit fields are split up and fully indexed. This means that
2410 journalctl in many ways is now a (nicer!) alternative to
2411 ausearch, the traditional audit client. Note that this
2412 implements only a minimal audit client. If you want the
2413 special audit modes like reboot-on-log-overflow, please use
2414 the traditional auditd instead, which can be used in
2415 parallel to journald.
2417 * The ConditionSecurity= unit file option now understands the
2418 special string "audit" to check whether auditing is
2421 * journalctl gained two new commands --vacuum-size= and
2422 --vacuum-time= to delete old journal files until the
2423 remaining ones take up no more than the specified size on disk,
2424 or are not older than the specified time.
2426 * A new, native PPPoE library has been added to sd-network,
2427 systemd's library of light-weight networking protocols. This
2428 library will be used in a future version of networkd to
2429 enable PPPoE communication without an external pppd daemon.
2431 * The busctl tool now understands a new "capture" verb that
2432 works similar to "monitor", but writes a packet capture
2433 trace to STDOUT that can be redirected to a file which is
2434 compatible with libcap's capture file format. This can then
2435 be loaded in Wireshark and similar tools to inspect bus
2438 * The busctl tool now understands a new "tree" verb that shows
2439 the object trees of a specific service on the bus, or of all
2442 * The busctl tool now understands a new "introspect" verb that
2443 shows all interfaces and members of objects on the bus,
2444 including their signature and values. This is particularly
2445 useful to get more information about bus objects shown by
2446 the new "busctl tree" command.
2448 * The busctl tool now understands new verbs "call",
2449 "set-property" and "get-property" for invoking bus method
2450 calls, setting and getting bus object properties in a
2453 * busctl gained a new --augment-creds= argument that controls
2454 whether the tool shall augment credential information it
2455 gets from the bus with data from /proc, in a possibly
2458 * nspawn's --link-journal= switch gained two new values
2459 "try-guest" and "try-host" that work like "guest" and
2460 "host", but do not fail if the host has no persistent
2461 journalling enabled. -j is now equivalent to
2462 --link-journal=try-guest.
2464 * macvlan network devices created by nspawn will now have
2465 stable MAC addresses.
2467 * A new SmackProcessLabel= unit setting has been added, which
2468 controls the SMACK security label processes forked off by
2469 the respective unit shall use.
2471 * If compiled with --enable-xkbcommon, systemd-localed will
2472 verify x11 keymap settings by compiling the given keymap. It
2473 will spew out warnings if the compilation fails. This
2474 requires libxkbcommon to be installed.
2476 * When a coredump is collected, a larger number of metadata
2477 fields is now collected and included in the journal records
2478 created for it. More specifically, control group membership,
2479 environment variables, memory maps, working directory,
2480 chroot directory, /proc/$PID/status, and a list of open file
2481 descriptors is now stored in the log entry.
2483 * The udev hwdb now contains DPI information for mice. For
2486 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
2488 * All systemd programs that read standalone configuration
2489 files in /etc now also support a corresponding series of
2490 .conf.d configuration directories in /etc/, /run/,
2491 /usr/local/lib/, /usr/lib/, and (if configured with
2492 --enable-split-usr) /lib/. In particular, the following
2493 configuration files now have corresponding configuration
2494 directories: system.conf user.conf, logind.conf,
2495 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
2496 resolved.conf, timesyncd.conf, journal-remote.conf, and
2497 journal-upload.conf. Note that distributions should use the
2498 configuration directories in /usr/lib/; the directories in
2499 /etc/ are reserved for the system administrator.
2501 * systemd-rfkill will no longer take the rfkill device name
2502 into account when storing rfkill state on disk, as the name
2503 might be dynamically assigned and not stable. Instead, the
2504 ID_PATH udev variable combined with the rfkill type (wlan,
2505 bluetooth, ...) is used.
2507 * A new service systemd-machine-id-commit.service has been
2508 added. When used on systems where /etc is read-only during
2509 boot, and /etc/machine-id is not initialized (but an empty
2510 file), this service will copy the temporary machine ID
2511 created as replacement into /etc after the system is fully
2512 booted up. This is useful for systems that are freshly
2513 installed with a non-initialized machine ID, but should get
2514 a fixed machine ID for subsequent boots.
2516 * networkd's .netdev files now provide a large set of
2517 configuration parameters for VXLAN devices. Similarly, the
2518 bridge port cost parameter is now configurable in .network
2519 files. There's also new support for configuring IP source
2520 routing. networkd .link files gained support for a new
2521 OriginalName= match that is useful to match against the
2522 original interface name the kernel assigned. .network files
2523 may include MTU= and MACAddress= fields for altering the MTU
2524 and MAC address while being connected to a specific network
2527 * The LUKS logic gained supported for configuring
2528 UUID-specific key files. There's also new support for naming
2529 LUKS device from the kernel command line, using the new
2530 luks.name= argument.
2532 * Timer units may now be transiently created via the bus API
2533 (this was previously already available for scope and service
2534 units). In addition it is now possible to create multiple
2535 transient units at the same time with a single bus call. The
2536 "systemd-run" tool has been updated to make use of this for
2537 running commands on a specified time, in at(1)-style.
2539 * tmpfiles gained support for "t" lines, for assigning
2540 extended attributes to files. Among other uses this may be
2541 used to assign SMACK labels to files.
2543 Contributions from: Alin Rauta, Alison Chaiken, Andrej
2544 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
2545 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
2546 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
2547 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
2548 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
2549 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
2550 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
2551 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
2552 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
2553 Michael Biebl, Michael Chapman, Michael Marineau, Michal
2554 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
2555 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
2556 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
2557 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
2558 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
2559 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
2560 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
2562 — Berlin, 2014-12-10
2566 * journalctl gained the new options -t/--identifier= to match
2567 on the syslog identifier (aka "tag"), as well as --utc to
2568 show log timestamps in the UTC timezone. journalctl now also
2569 accepts -n/--lines=all to disable line capping in a pager.
2571 * journalctl gained a new switch, --flush, that synchronously
2572 flushes logs from /run/log/journal to /var/log/journal if
2573 persistent storage is enabled. systemd-journal-flush.service
2574 now waits until the operation is complete.
2576 * Services can notify the manager before they start a reload
2577 (by sending RELOADING=1) or shutdown (by sending
2578 STOPPING=1). This allows the manager to track and show the
2579 internal state of daemons and closes a race condition when
2580 the process is still running but has closed its D-Bus
2583 * Services with Type=oneshot do not have to have any ExecStart
2586 * User units are now loaded also from
2587 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
2588 /run/systemd/user directory that was already previously
2589 supported, but is under the control of the user.
2591 * Job timeouts (i.e. time-outs on the time a job that is
2592 queued stays in the run queue) can now optionally result in
2593 immediate reboot or power-off actions (JobTimeoutAction= and
2594 JobTimeoutRebootArgument=). This is useful on ".target"
2595 units, to limit the maximum time a target remains
2596 undispatched in the run queue, and to trigger an emergency
2597 operation in such a case. This is now used by default to
2598 turn off the system if boot-up (as defined by everything in
2599 basic.target) hangs and does not complete for at least
2600 15min. Also, if power-off or reboot hang for at least 30min
2601 an immediate power-off/reboot operation is triggered. This
2602 functionality is particularly useful to increase reliability
2603 on embedded devices, but also on laptops which might
2604 accidentally get powered on when carried in a backpack and
2605 whose boot stays stuck in a hard disk encryption passphrase
2608 * systemd-logind can be configured to also handle lid switch
2609 events even when the machine is docked or multiple displays
2610 are attached (HandleLidSwitchDocked= option).
2612 * A helper binary and a service have been added which can be
2613 used to resume from hibernation in the initramfs. A
2614 generator will parse the resume= option on the kernel
2615 command line to trigger resume.
2617 * A user console daemon systemd-consoled has been
2618 added. Currently, it is a preview, and will so far open a
2619 single terminal on each session of the user marked as
2620 Desktop=systemd-console.
2622 * Route metrics can be specified for DHCP routes added by
2625 * The SELinux context of socket-activated services can be set
2626 from the information provided by the networking stack
2627 (SELinuxContextFromNet= option).
2629 * Userspace firmware loading support has been removed and
2630 the minimum supported kernel version is thus bumped to 3.7.
2632 * Timeout for udev workers has been increased from 1 to 3
2633 minutes, but a warning will be printed after 1 minute to
2634 help diagnose kernel modules that take a long time to load.
2636 * Udev rules can now remove tags on devices with TAG-="foobar".
2638 * systemd's readahead implementation has been removed. In many
2639 circumstances it didn't give expected benefits even for
2640 rotational disk drives and was becoming less relevant in the
2641 age of SSDs. As none of the developers has been using
2642 rotating media anymore, and nobody stepped up to actively
2643 maintain this component of systemd it has now been removed.
2645 * Swap units can use Options= to specify discard options.
2646 Discard options specified for swaps in /etc/fstab are now
2649 * Docker containers are now detected as a separate type of
2652 * The Password Agent protocol gained support for queries where
2653 the user input is shown, useful e.g. for user names.
2654 systemd-ask-password gained a new --echo option to turn that
2657 * The default sysctl.d/ snippets will now set:
2659 net.core.default_qdisc = fq_codel
2661 This selects Fair Queuing Controlled Delay as the default
2662 queuing discipline for network interfaces. fq_codel helps
2663 fight the network bufferbloat problem. It is believed to be
2664 a good default with no tuning required for most workloads.
2665 Downstream distributions may override this choice. On 10Gbit
2666 servers that do not do forwarding, "fq" may perform better.
2667 Systems without a good clocksource should use "pfifo_fast".
2669 * If kdbus is enabled during build a new option BusPolicy= is
2670 available for service units, that allows locking all service
2671 processes into a stricter bus policy, in order to limit
2672 access to various bus services, or even hide most of them
2673 from the service's view entirely.
2675 * networkctl will now show the .network and .link file
2676 networkd has applied to a specific interface.
2678 * sd-login gained a new API call sd_session_get_desktop() to
2679 query which desktop environment has been selected for a
2682 * UNIX utmp support is now compile-time optional to support
2683 legacy-free systems.
2685 * systemctl gained two new commands "add-wants" and
2686 "add-requires" for pulling in units from specific targets
2689 * If the word "rescue" is specified on the kernel command line
2690 the system will now boot into rescue mode (aka
2691 rescue.target), which was previously available only by
2692 specifying "1" or "systemd.unit=rescue.target" on the kernel
2693 command line. This new kernel command line option nicely
2694 mirrors the already existing "emergency" kernel command line
2697 * New kernel command line options mount.usr=, mount.usrflags=,
2698 mount.usrfstype= have been added that match root=, rootflags=,
2699 rootfstype= but allow mounting a specific file system to
2702 * The $NOTIFY_SOCKET is now also passed to control processes of
2703 services, not only the main process.
2705 * This version reenables support for fsck's -l switch. This
2706 means at least version v2.25 of util-linux is required for
2707 operation, otherwise dead-locks on device nodes may
2708 occur. Again: you need to update util-linux to at least
2709 v2.25 when updating systemd to v217.
2711 * The "multi-seat-x" tool has been removed from systemd, as
2712 its functionality has been integrated into X servers 1.16,
2713 and the tool is hence redundant. It is recommended to update
2714 display managers invoking this tool to simply invoke X
2715 directly from now on, again.
2717 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
2718 message flag has been added for all of systemd's PolicyKit
2719 authenticated method calls has been added. In particular
2720 this now allows optional interactive authorization via
2721 PolicyKit for many of PID1's privileged operations such as
2722 unit file enabling and disabling.
2724 * "udevadm hwdb --update" learnt a new switch "--usr" for
2725 placing the rebuilt hardware database in /usr instead of
2726 /etc. When used only hardware database entries stored in
2727 /usr will be used, and any user database entries in /etc are
2728 ignored. This functionality is useful for vendors to ship a
2729 pre-built database on systems where local configuration is
2730 unnecessary or unlikely.
2732 * Calendar time specifications in .timer units now also
2733 understand the strings "semi-annually", "quarterly" and
2734 "minutely" as shortcuts (in addition to the preexisting
2735 "anually", "hourly", ...).
2737 * systemd-tmpfiles will now correctly create files in /dev
2738 at boot which are marked for creation only at boot. It is
2739 recommended to always create static device nodes with 'c!'
2740 and 'b!', so that they are created only at boot and not
2741 overwritten at runtime.
2743 * When the watchdog logic is used for a service (WatchdogSec=)
2744 and the watchdog timeout is hit the service will now be
2745 terminated with SIGABRT (instead of just SIGTERM), in order
2746 to make sure a proper coredump and backtrace is
2747 generated. This ensures that hanging services will result in
2748 similar coredump/backtrace behaviour as services that hit a
2751 Contributions from: Andreas Henriksson, Andrei Borzenkov,
2752 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
2753 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
2754 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
2755 Herrmann, David Sommerseth, David Strauss, Emil Renner
2756 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
2757 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
2758 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
2759 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
2760 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
2761 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
2762 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
2763 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
2764 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
2765 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
2766 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
2767 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
2768 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
2769 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
2770 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
2771 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
2774 — Berlin, 2014-10-28
2778 * timedated no longer reads NTP implementation unit names from
2779 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
2780 implementations should add a
2782 Conflicts=systemd-timesyncd.service
2784 to their unit files to take over and replace systemd's NTP
2785 default functionality.
2787 * systemd-sysusers gained a new line type "r" for configuring
2788 which UID/GID ranges to allocate system users/groups
2789 from. Lines of type "u" may now add an additional column
2790 that specifies the home directory for the system user to be
2791 created. Also, systemd-sysusers may now optionally read user
2792 information from STDIN instead of a file. This is useful for
2793 invoking it from RPM preinst scriptlets that need to create
2794 users before the first RPM file is installed since these
2795 files might need to be owned by them. A new
2796 %sysusers_create_inline RPM macro has been introduced to do
2797 just that. systemd-sysusers now updates the shadow files as
2798 well as the user/group databases, which should enhance
2799 compatibility with certain tools like grpck.
2801 * A number of bus APIs of PID 1 now optionally consult
2802 PolicyKit to permit access for otherwise unprivileged
2803 clients under certain conditions. Note that this currently
2804 doesn't support interactive authentication yet, but this is
2805 expected to be added eventually, too.
2807 * /etc/machine-info now has new fields for configuring the
2808 deployment environment of the machine, as well as the
2809 location of the machine. hostnamectl has been updated with
2810 new command to update these fields.
2812 * systemd-timesyncd has been updated to automatically acquire
2813 NTP server information from systemd-networkd, which might
2814 have been discovered via DHCP.
2816 * systemd-resolved now includes a caching DNS stub resolver
2817 and a complete LLMNR name resolution implementation. A new
2818 NSS module "nss-resolve" has been added which can be used
2819 instead of glibc's own "nss-dns" to resolve hostnames via
2820 systemd-resolved. Hostnames, addresses and arbitrary RRs may
2821 be resolved via systemd-resolved D-Bus APIs. In contrast to
2822 the glibc internal resolver systemd-resolved is aware of
2823 multi-homed system, and keeps DNS server and caches separate
2824 and per-interface. Queries are sent simultaneously on all
2825 interfaces that have DNS servers configured, in order to
2826 properly handle VPNs and local LANs which might resolve
2827 separate sets of domain names. systemd-resolved may acquire
2828 DNS server information from systemd-networkd automatically,
2829 which in turn might have discovered them via DHCP. A tool
2830 "systemd-resolve-host" has been added that may be used to
2831 query the DNS logic in resolved. systemd-resolved implements
2832 IDNA and automatically uses IDNA or UTF-8 encoding depending
2833 on whether classic DNS or LLMNR is used as transport. In the
2834 next releases we intend to add a DNSSEC and mDNS/DNS-SD
2835 implementation to systemd-resolved.
2837 * A new NSS module nss-mymachines has been added, that
2838 automatically resolves the names of all local registered
2839 containers to their respective IP addresses.
2841 * A new client tool "networkctl" for systemd-networkd has been
2842 added. It currently is entirely passive and will query
2843 networking configuration from udev, rtnetlink and networkd,
2844 and present it to the user in a very friendly
2845 way. Eventually, we hope to extend it to become a full
2846 control utility for networkd.
2848 * .socket units gained a new DeferAcceptSec= setting that
2849 controls the kernels' TCP_DEFER_ACCEPT sockopt for
2850 TCP. Similarly, support for controlling TCP keep-alive
2851 settings has been added (KeepAliveTimeSec=,
2852 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
2853 turning off Nagle's algorithm on TCP has been added
2856 * logind learned a new session type "web", for use in projects
2857 like Cockpit which register web clients as PAM sessions.
2859 * timer units with at least one OnCalendar= setting will now
2860 be started only after timer-sync.target has been
2861 reached. This way they will not elapse before the system
2862 clock has been corrected by a local NTP client or
2863 similar. This is particular useful on RTC-less embedded
2864 machines, that come up with an invalid system clock.
2866 * systemd-nspawn's --network-veth= switch should now result in
2867 stable MAC addresses for both the outer and the inner side
2870 * systemd-nspawn gained a new --volatile= switch for running
2871 container instances with /etc or /var unpopulated.
2873 * The kdbus client code has been updated to use the new Linux
2874 3.17 memfd subsystem instead of the old kdbus-specific one.
2876 * systemd-networkd's DHCP client and server now support
2877 FORCERENEW. There are also new configuration options to
2878 configure the vendor client identifier and broadcast mode
2881 * systemd will no longer inform the kernel about the current
2882 timezone, as this is necessarily incorrect and racy as the
2883 kernel has no understanding of DST and similar
2884 concepts. This hence means FAT timestamps will be always
2885 considered UTC, similar to what Android is already
2886 doing. Also, when the RTC is configured to the local time
2887 (rather than UTC) systemd will never synchronize back to it,
2888 as this might confuse Windows at a later boot.
2890 * systemd-analyze gained a new command "verify" for offline
2891 validation of unit files.
2893 * systemd-networkd gained support for a couple of additional
2894 settings for bonding networking setups. Also, the metric for
2895 statically configured routes may now be configured. For
2896 network interfaces where this is appropriate the peer IP
2897 address may now be configured.
2899 * systemd-networkd's DHCP client will no longer request
2900 broadcasting by default, as this tripped up some networks.
2901 For hardware where broadcast is required the feature should
2902 be switched back on using RequestBroadcast=yes.
2904 * systemd-networkd will now set up IPv4LL addresses (when
2905 enabled) even if DHCP is configured successfully.
2907 * udev will now default to respect network device names given
2908 by the kernel when the kernel indicates that these are
2909 predictable. This behavior can be tweaked by changing
2910 NamePolicy= in the relevant .link file.
2912 * A new library systemd-terminal has been added that
2913 implements full TTY stream parsing and rendering. This
2914 library is supposed to be used later on for implementing a
2915 full userspace VT subsystem, replacing the current kernel
2918 * A new tool systemd-journal-upload has been added to push
2919 journal data to a remote system running
2920 systemd-journal-remote.
2922 * journald will no longer forward all local data to another
2923 running syslog daemon. This change has been made because
2924 rsyslog (which appears to be the most commonly used syslog
2925 implementation these days) no longer makes use of this, and
2926 instead pulls the data out of the journal on its own. Since
2927 forwarding the messages to a non-existent syslog server is
2928 more expensive than we assumed we have now turned this
2929 off. If you run a syslog server that is not a recent rsyslog
2930 version, you have to turn this option on again
2931 (ForwardToSyslog= in journald.conf).
2933 * journald now optionally supports the LZ4 compressor for
2934 larger journal fields. This compressor should perform much
2935 better than XZ which was the previous default.
2937 * machinectl now shows the IP addresses of local containers,
2938 if it knows them, plus the interface name of the container.
2940 * A new tool "systemd-escape" has been added that makes it
2941 easy to escape strings to build unit names and similar.
2943 * sd_notify() messages may now include a new ERRNO= field
2944 which is parsed and collected by systemd and shown among the
2945 "systemctl status" output for a service.
2947 * A new component "systemd-firstboot" has been added that
2948 queries the most basic systemd information (timezone,
2949 hostname, root password) interactively on first
2950 boot. Alternatively it may also be used to provision these
2951 things offline on OS images installed into directories.
2953 * The default sysctl.d/ snippets will now set
2955 net.ipv4.conf.default.promote_secondaries=1
2957 This has the benefit of no flushing secondary IP addresses
2958 when primary addresses are removed.
2960 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
2961 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
2962 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
2963 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
2964 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
2965 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
2966 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
2967 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
2968 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
2969 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
2970 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
2971 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
2972 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
2973 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
2974 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
2976 — Berlin, 2014-08-19
2980 * A new tool systemd-sysusers has been added. This tool
2981 creates system users and groups in /etc/passwd and
2982 /etc/group, based on static declarative system user/group
2983 definitions in /usr/lib/sysusers.d/. This is useful to
2984 enable factory resets and volatile systems that boot up with
2985 an empty /etc directory, and thus need system users and
2986 groups created during early boot. systemd now also ships
2987 with two default sysusers.d/ files for the most basic
2988 users and groups systemd and the core operating system
2991 * A new tmpfiles snippet has been added that rebuilds the
2992 essential files in /etc on boot, should they be missing.
2994 * A directive for ensuring automatic clean-up of
2995 /var/cache/man/ has been removed from the default
2996 configuration. This line should now be shipped by the man
2997 implementation. The necessary change has been made to the
2998 man-db implementation. Note that you need to update your man
2999 implementation to one that ships this line, otherwise no
3000 automatic clean-up of /var/cache/man will take place.
3002 * A new condition ConditionNeedsUpdate= has been added that
3003 may conditionalize services to only run when /etc or /var
3004 are "older" than the vendor operating system resources in
3005 /usr. This is useful for reconstructing or updating /etc
3006 after an offline update of /usr or a factory reset, on the
3007 next reboot. Services that want to run once after such an
3008 update or reset should use this condition and order
3009 themselves before the new systemd-update-done.service, which
3010 will mark the two directories as fully updated. A number of
3011 service files have been added making use of this, to rebuild
3012 the udev hardware database, the journald message catalog and
3013 dynamic loader cache (ldconfig). The systemd-sysusers tool
3014 described above also makes use of this now. With this in
3015 place it is now possible to start up a minimal operating
3016 system with /etc empty cleanly. For more information on the
3017 concepts involved see this recent blog story:
3019 http://0pointer.de/blog/projects/stateless.html
3021 * A new system group "input" has been introduced, and all
3022 input device nodes get this group assigned. This is useful
3023 for system-level software to get access to input devices. It
3024 complements what is already done for "audio" and "video".
3026 * systemd-networkd learnt minimal DHCPv4 server support in
3027 addition to the existing DHCPv4 client support. It also
3028 learnt DHCPv6 client and IPv6 Router Solicitation client
3029 support. The DHCPv4 client gained support for static routes
3030 passed in from the server. Note that the [DHCPv4] section
3031 known in older systemd-networkd versions has been renamed to
3032 [DHCP] and is now also used by the DHCPv6 client. Existing
3033 .network files using settings of this section should be
3034 updated, though compatibility is maintained. Optionally, the
3035 client hostname may now be sent to the DHCP server.
3037 * networkd gained support for vxlan virtual networks as well
3038 as tun/tap and dummy devices.
3040 * networkd gained support for automatic allocation of address
3041 ranges for interfaces from a system-wide pool of
3042 addresses. This is useful for dynamically managing a large
3043 number of interfaces with a single network configuration
3044 file. In particular this is useful to easily assign
3045 appropriate IP addresses to the veth links of a large number
3046 of nspawn instances.
3048 * RPM macros for processing sysusers, sysctl and binfmt
3049 drop-in snippets at package installation time have been
3052 * The /etc/os-release file should now be placed in
3053 /usr/lib/os-release. The old location is automatically
3054 created as symlink. /usr/lib is the more appropriate
3055 location of this file, since it shall actually describe the
3056 vendor operating system shipped in /usr, and not the
3057 configuration stored in /etc.
3059 * .mount units gained a new boolean SloppyOptions= setting
3060 that maps to mount(8)'s -s option which enables permissive
3061 parsing of unknown mount options.
3063 * tmpfiles learnt a new "L+" directive which creates a symlink
3064 but (unlike "L") deletes a pre-existing file first, should
3065 it already exist and not already be the correct
3066 symlink. Similarly, "b+", "c+" and "p+" directives have been
3067 added as well, which create block and character devices, as
3068 well as fifos in the filesystem, possibly removing any
3069 pre-existing files of different types.
3071 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
3072 'argument' field (which so far specified the source to
3073 symlink/copy the files from) is now optional. If omitted the
3074 same file os copied from /usr/share/factory/ suffixed by the
3075 full destination path. This is useful for populating /etc
3076 with essential files, by copying them from vendor defaults
3077 shipped in /usr/share/factory/etc.
3079 * A new command "systemctl preset-all" has been added that
3080 applies the service preset settings to all installed unit
3081 files. A new switch --preset-mode= has been added that
3082 controls whether only enable or only disable operations
3085 * A new command "systemctl is-system-running" has been added
3086 that allows checking the overall state of the system, for
3087 example whether it is fully up and running.
3089 * When the system boots up with an empty /etc, the equivalent
3090 to "systemctl preset-all" is executed during early boot, to
3091 make sure all default services are enabled after a factory
3094 * systemd now contains a minimal preset file that enables the
3095 most basic services systemd ships by default.
3097 * Unit files' [Install] section gained a new DefaultInstance=
3098 field for defining the default instance to create if a
3099 template unit is enabled with no instance specified.
3101 * A new passive target cryptsetup-pre.target has been added
3102 that may be used by services that need to make they run and
3103 finish before the first LUKS cryptographic device is set up.
3105 * The /dev/loop-control and /dev/btrfs-control device nodes
3106 are now owned by the "disk" group by default, opening up
3107 access to this group.
3109 * systemd-coredump will now automatically generate a
3110 stack trace of all core dumps taking place on the system,
3111 based on elfutils' libdw library. This stack trace is logged
3114 * systemd-coredump may now optionally store coredumps directly
3115 on disk (in /var/lib/systemd/coredump, possibly compressed),
3116 instead of storing them unconditionally in the journal. This
3117 mode is the new default. A new configuration file
3118 /etc/systemd/coredump.conf has been added to configure this
3119 and other parameters of systemd-coredump.
3121 * coredumpctl gained a new "info" verb to show details about a
3122 specific coredump. A new switch "-1" has also been added
3123 that makes sure to only show information about the most
3124 recent entry instead of all entries. Also, as the tool is
3125 generally useful now the "systemd-" prefix of the binary
3126 name has been removed. Distributions that want to maintain
3127 compatibility with the old name should add a symlink from
3128 the old name to the new name.
3130 * journald's SplitMode= now defaults to "uid". This makes sure
3131 that unprivileged users can access their own coredumps with
3132 coredumpctl without restrictions.
3134 * New kernel command line options "systemd.wants=" (for
3135 pulling an additional unit during boot), "systemd.mask="
3136 (for masking a specific unit for the boot), and
3137 "systemd.debug-shell" (for enabling the debug shell on tty9)
3138 have been added. This is implemented in the new generator
3139 "systemd-debug-generator".
3141 * systemd-nspawn will now by default filter a couple of
3142 syscalls for containers, among them those required for
3143 kernel module loading, direct x86 IO port access, swap
3144 management, and kexec. Most importantly though
3145 open_by_handle_at() is now prohibited for containers,
3146 closing a hole similar to a recently discussed vulnerability
3147 in docker regarding access to files on file hierarchies the
3148 container should normally not have access to. Note that, for
3149 nspawn, we generally make no security claims anyway (and
3150 this is explicitly documented in the man page), so this is
3151 just a fix for one of the most obvious problems.
3153 * A new man page file-hierarchy(7) has been added that
3154 contains a minimized, modernized version of the file system
3155 layout systemd expects, similar in style to the FHS
3156 specification or hier(5). A new tool systemd-path(1) has
3157 been added to query many of these paths for the local
3160 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
3161 longer done. Since the directory now has a per-user size
3162 limit, and is cleaned on logout this appears unnecessary,
3163 in particular since this now brings the lifecycle of this
3164 directory closer in line with how IPC objects are handled.
3166 * systemd.pc now exports a number of additional directories,
3167 including $libdir (which is useful to identify the library
3168 path for the primary architecture of the system), and a
3169 couple of drop-in directories.
3171 * udev's predictable network interface names now use the dev_port
3172 sysfs attribute, introduced in linux 3.15 instead of dev_id to
3173 distinguish between ports of the same PCI function. dev_id should
3174 only be used for ports using the same HW address, hence the need
3177 * machined has been updated to export the OS version of a
3178 container (read from /etc/os-release and
3179 /usr/lib/os-release) on the bus. This is now shown in
3180 "machinectl status" for a machine.
3182 * A new service setting RestartForceExitStatus= has been
3183 added. If configured to a set of exit signals or process
3184 return values, the service will be restarted when the main
3185 daemon process exits with any of them, regardless of the
3188 * systemctl's -H switch for connecting to remote systemd
3189 machines has been extended so that it may be used to
3190 directly connect to a specific container on the
3191 host. "systemctl -H root@foobar:waldi" will now connect as
3192 user "root" to host "foobar", and then proceed directly to
3193 the container named "waldi". Note that currently you have to
3194 authenticate as user "root" for this to work, as entering
3195 containers is a privileged operation.
3197 Contributions from: Andreas Henriksson, Benjamin Steinwender,
3198 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
3199 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
3200 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
3201 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
3202 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
3203 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
3204 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
3205 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
3206 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
3207 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
3208 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
3210 — Berlin, 2014-07-03
3214 * As an experimental feature, udev now tries to lock the
3215 disk device node (flock(LOCK_SH|LOCK_NB)) while it
3216 executes events for the disk or any of its partitions.
3217 Applications like partitioning programs can lock the
3218 disk device node (flock(LOCK_EX)) and claim temporary
3219 device ownership that way; udev will entirely skip all event
3220 handling for this disk and its partitions. If the disk
3221 was opened for writing, the close will trigger a partition
3222 table rescan in udev's "watch" facility, and if needed
3223 synthesize "change" events for the disk and all its partitions.
3224 This is now unconditionally enabled, and if it turns out to
3225 cause major problems, we might turn it on only for specific
3226 devices, or might need to disable it entirely. Device Mapper
3227 devices are excluded from this logic.
3229 * We temporarily dropped the "-l" switch for fsck invocations,
3230 since they collide with the flock() logic above. util-linux
3231 upstream has been changed already to avoid this conflict,
3232 and we will readd "-l" as soon as util-linux with this
3233 change has been released.
3235 * The dependency on libattr has been removed. Since a long
3236 time, the extended attribute calls have moved to glibc, and
3237 libattr is thus unnecessary.
3239 * Virtualization detection works without privileges now. This
3240 means the systemd-detect-virt binary no longer requires
3241 CAP_SYS_PTRACE file capabilities, and our daemons can run
3242 with fewer privileges.
3244 * systemd-networkd now runs under its own "systemd-network"
3245 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
3246 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
3247 loses the ability to write to files owned by root this way.
3249 * Similarly, systemd-resolved now runs under its own
3250 "systemd-resolve" user with no capabilities remaining.
3252 * Similarly, systemd-bus-proxyd now runs under its own
3253 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
3255 * systemd-networkd gained support for setting up "veth"
3256 virtual Ethernet devices for container connectivity, as well
3257 as GRE and VTI tunnels.
3259 * systemd-networkd will no longer automatically attempt to
3260 manually load kernel modules necessary for certain tunnel
3261 transports. Instead, it is assumed the kernel loads them
3262 automatically when required. This only works correctly on
3263 very new kernels. On older kernels, please consider adding
3264 the kernel modules to /etc/modules-load.d/ as a work-around.
3266 * The resolv.conf file systemd-resolved generates has been
3267 moved to /run/systemd/resolve/. If you have a symlink from
3268 /etc/resolv.conf, it might be necessary to correct it.
3270 * Two new service settings, ProtectHome= and ProtectSystem=,
3271 have been added. When enabled, they will make the user data
3272 (such as /home) inaccessible or read-only and the system
3273 (such as /usr) read-only, for specific services. This allows
3274 very light-weight per-service sandboxing to avoid
3275 modifications of user data or system files from
3276 services. These two new switches have been enabled for all
3277 of systemd's long-running services, where appropriate.
3279 * Socket units gained new SocketUser= and SocketGroup=
3280 settings to set the owner user and group of AF_UNIX sockets
3281 and FIFOs in the file system.
3283 * Socket units gained a new RemoveOnStop= setting. If enabled,
3284 all FIFOS and sockets in the file system will be removed
3285 when the specific socket unit is stopped.
3287 * Socket units gained a new Symlinks= setting. It takes a list
3288 of symlinks to create to file system sockets or FIFOs
3289 created by the specific Unix sockets. This is useful to
3290 manage symlinks to socket nodes with the same life-cycle as
3293 * The /dev/log socket and /dev/initctl FIFO have been moved to
3294 /run, and have been replaced by symlinks. This allows
3295 connecting to these facilities even if PrivateDevices=yes is
3296 used for a service (which makes /dev/log itself unavailable,
3297 but /run is left). This also has the benefit of ensuring
3298 that /dev only contains device nodes, directories and
3299 symlinks, and nothing else.
3301 * sd-daemon gained two new calls sd_pid_notify() and
3302 sd_pid_notifyf(). They are similar to sd_notify() and
3303 sd_notifyf(), but allow overriding of the source PID of
3304 notification messages if permissions permit this. This is
3305 useful to send notify messages on behalf of a different
3306 process (for example, the parent process). The
3307 systemd-notify tool has been updated to make use of this
3308 when sending messages (so that notification messages now
3309 originate from the shell script invoking systemd-notify and
3310 not the systemd-notify process itself. This should minimize
3311 a race where systemd fails to associate notification
3312 messages to services when the originating process already
3315 * A new "on-abnormal" setting for Restart= has been added. If
3316 set, it will result in automatic restarts on all "abnormal"
3317 reasons for a process to exit, which includes unclean
3318 signals, core dumps, timeouts and watchdog timeouts, but
3319 does not include clean and unclean exit codes or clean
3320 signals. Restart=on-abnormal is an alternative for
3321 Restart=on-failure for services that shall be able to
3322 terminate and avoid restarts on certain errors, by
3323 indicating so with an unclean exit code. Restart=on-failure
3324 or Restart=on-abnormal is now the recommended setting for
3325 all long-running services.
3327 * If the InaccessibleDirectories= service setting points to a
3328 mount point (or if there are any submounts contained within
3329 it), it is now attempted to completely unmount it, to make
3330 the file systems truly unavailable for the respective
3333 * The ReadOnlyDirectories= service setting and
3334 systemd-nspawn's --read-only parameter are now recursively
3335 applied to all submounts, too.
3337 * Mount units may now be created transiently via the bus APIs.
3339 * The support for SysV and LSB init scripts has been removed
3340 from the systemd daemon itself. Instead, it is now
3341 implemented as a generator that creates native systemd units
3342 from these scripts when needed. This enables us to remove a
3343 substantial amount of legacy code from PID 1, following the
3344 fact that many distributions only ship a very small number
3345 of LSB/SysV init scripts nowadays.
3347 * Privileged Xen (dom0) domains are not considered
3348 virtualization anymore by the virtualization detection
3349 logic. After all, they generally have unrestricted access to
3350 the hardware and usually are used to manage the unprivileged
3353 * systemd-tmpfiles gained a new "C" line type, for copying
3354 files or entire directories.
3356 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
3357 lines. So far, they have been non-globbing versions of the
3358 latter, and have thus been redundant. In future, it is
3359 recommended to only use "z". "m" has hence been removed
3360 from the documentation, even though it stays supported.
3362 * A tmpfiles snippet to recreate the most basic structure in
3363 /var has been added. This is enough to create the /var/run →
3364 /run symlink and create a couple of structural
3365 directories. This allows systems to boot up with an empty or
3366 volatile /var. Of course, while with this change, the core OS
3367 now is capable with dealing with a volatile /var, not all
3368 user services are ready for it. However, we hope that sooner
3369 or later, many service daemons will be changed upstream so
3370 that they are able to automatically create their necessary
3371 directories in /var at boot, should they be missing. This is
3372 the first step to allow state-less systems that only require
3373 the vendor image for /usr to boot.
3375 * systemd-nspawn has gained a new --tmpfs= switch to mount an
3376 empty tmpfs instance to a specific directory. This is
3377 particularly useful for making use of the automatic
3378 reconstruction of /var (see above), by passing --tmpfs=/var.
3380 * Access modes specified in tmpfiles snippets may now be
3381 prefixed with "~", which indicates that they shall be masked
3382 by whether the existing file or directory is currently
3383 writable, readable or executable at all. Also, if specified,
3384 the sgid/suid/sticky bits will be masked for all
3387 * A new passive target unit "network-pre.target" has been
3388 added which is useful for services that shall run before any
3389 network is configured, for example firewall scripts.
3391 * The "floppy" group that previously owned the /dev/fd*
3392 devices is no longer used. The "disk" group is now used
3393 instead. Distributions should probably deprecate usage of
3396 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
3397 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
3398 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
3399 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
3400 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
3401 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
3404 — Berlin, 2014-06-11
3408 * A new "systemd-timesyncd" daemon has been added for
3409 synchronizing the system clock across the network. It
3410 implements an SNTP client. In contrast to NTP
3411 implementations such as chrony or the NTP reference server,
3412 this only implements a client side, and does not bother with
3413 the full NTP complexity, focusing only on querying time from
3414 one remote server and synchronizing the local clock to
3415 it. Unless you intend to serve NTP to networked clients or
3416 want to connect to local hardware clocks, this simple NTP
3417 client should be more than appropriate for most
3418 installations. The daemon runs with minimal privileges, and
3419 has been hooked up with networkd to only operate when
3420 network connectivity is available. The daemon saves the
3421 current clock to disk every time a new NTP sync has been
3422 acquired, and uses this to possibly correct the system clock
3423 early at bootup, in order to accommodate for systems that
3424 lack an RTC such as the Raspberry Pi and embedded devices,
3425 and to make sure that time monotonically progresses on these
3426 systems, even if it is not always correct. To make use of
3427 this daemon, a new system user and group "systemd-timesync"
3428 needs to be created on installation of systemd.
3430 * The queue "seqnum" interface of libudev has been disabled, as
3431 it was generally incompatible with device namespacing as
3432 sequence numbers of devices go "missing" if the devices are
3433 part of a different namespace.
3435 * "systemctl list-timers" and "systemctl list-sockets" gained
3436 a --recursive switch for showing units of these types also
3437 for all local containers, similar in style to the already
3438 supported --recursive switch for "systemctl list-units".
3440 * A new RebootArgument= setting has been added for service
3441 units, which may be used to specify a kernel reboot argument
3442 to use when triggering reboots with StartLimitAction=.
3444 * A new FailureAction= setting has been added for service
3445 units which may be used to specify an operation to trigger
3446 when a service fails. This works similarly to
3447 StartLimitAction=, but unlike it, controls what is done
3448 immediately rather than only after several attempts to
3449 restart the service in question.
3451 * hostnamed got updated to also expose the kernel name,
3452 release, and version on the bus. This is useful for
3453 executing commands like hostnamectl with the -H switch.
3454 systemd-analyze makes use of this to properly display
3455 details when running non-locally.
3457 * The bootchart tool can now show cgroup information in the
3458 graphs it generates.
3460 * The CFS CPU quota cgroup attribute is now exposed for
3461 services. The new CPUQuota= switch has been added for this
3462 which takes a percentage value. Setting this will have the
3463 result that a service may never get more CPU time than the
3464 specified percentage, even if the machine is otherwise idle.
3466 * systemd-networkd learned IPIP and SIT tunnel support.
3468 * LSB init scripts exposing a dependency on $network will now
3469 get a dependency on network-online.target rather than simply
3470 network.target. This should bring LSB handling closer to
3471 what it was on SysV systems.
3473 * A new fsck.repair= kernel option has been added to control
3474 how fsck shall deal with unclean file systems at boot.
3476 * The (.ini) configuration file parser will now silently
3477 ignore sections whose name begins with "X-". This may be
3478 used to maintain application-specific extension sections in unit
3481 * machined gained a new API to query the IP addresses of
3482 registered containers. "machinectl status" has been updated
3483 to show these addresses in its output.
3485 * A new call sd_uid_get_display() has been added to the
3486 sd-login APIs for querying the "primary" session of a
3487 user. The "primary" session of the user is elected from the
3488 user's sessions and generally a graphical session is
3489 preferred over a text one.
3491 * A minimal systemd-resolved daemon has been added. It
3492 currently simply acts as a companion to systemd-networkd and
3493 manages resolv.conf based on per-interface DNS
3494 configuration, possibly supplied via DHCP. In the long run
3495 we hope to extend this into a local DNSSEC enabled DNS and
3498 * The systemd-networkd-wait-online tool is now enabled by
3499 default. It will delay network-online.target until a network
3500 connection has been configured. The tool primarily integrates
3501 with networkd, but will also make a best effort to make sense
3502 of network configuration performed in some other way.
3504 * Two new service options StartupCPUShares= and
3505 StartupBlockIOWeight= have been added that work similarly to
3506 CPUShares= and BlockIOWeight= however only apply during
3507 system startup. This is useful to prioritize certain services
3508 differently during bootup than during normal runtime.
3510 * hostnamed has been changed to prefer the statically
3511 configured hostname in /etc/hostname (unless set to
3512 'localhost' or empty) over any dynamic one supplied by
3513 dhcp. With this change, the rules for picking the hostname
3514 match more closely the rules of other configuration settings
3515 where the local administrator's configuration in /etc always
3516 overrides any other settings.
3518 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
3519 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
3520 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
3521 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
3522 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
3523 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
3524 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
3525 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
3526 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
3527 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
3528 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
3529 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
3530 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
3531 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
3532 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
3533 Lindskog, WaLyong Cho, Will Woods, Zbigniew
3536 — Beijing, 2014-05-28
3540 * When restoring the screen brightness at boot, stay away from
3541 the darkest setting or from the lowest 5% of the available
3542 range, depending on which is the larger value of both. This
3543 should effectively protect the user from rebooting into a
3544 black screen, should the brightness have been set to minimum
3547 * sd-login gained a new sd_machine_get_class() call to
3548 determine the class ("vm" or "container") of a machine
3549 registered with machined.
3551 * sd-login gained new calls
3552 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
3553 to query the identity of the peer of a local AF_UNIX
3554 connection. They operate similarly to their sd_pid_get_xyz()
3557 * PID 1 will now maintain a system-wide system state engine
3558 with the states "starting", "running", "degraded",
3559 "maintenance", "stopping". These states are bound to system
3560 startup, normal runtime, runtime with at least one failed
3561 service, rescue/emergency mode and system shutdown. This
3562 state is shown in the "systemctl status" output when no unit
3563 name is passed. It is useful to determine system state, in
3564 particularly when doing so for many systems or containers at
3567 * A new command "list-machines" has been added to "systemctl"
3568 that lists all local OS containers and shows their system
3569 state (see above), if systemd runs inside of them.
3571 * systemctl gained a new "-r" switch to recursively enumerate
3572 units on all local containers, when used with the
3573 "list-unit" command (which is the default one that is
3574 executed when no parameters are specified).
3576 * The GPT automatic partition discovery logic will now honour
3577 two GPT partition flags: one may be set on a partition to
3578 cause it to be mounted read-only, and the other may be set
3579 on a partition to ignore it during automatic discovery.
3581 * Two new GPT type UUIDs have been added for automatic root
3582 partition discovery, for 32-bit and 64-bit ARM. This is not
3583 particularly useful for discovering the root directory on
3584 these architectures during bare-metal boots (since UEFI is
3585 not common there), but still very useful to allow booting of
3586 ARM disk images in nspawn with the -i option.
3588 * MAC addresses of interfaces created with nspawn's
3589 --network-interface= switch will now be generated from the
3590 machine name, and thus be stable between multiple invocations
3593 * logind will now automatically remove all IPC objects owned
3594 by a user if she or he fully logs out. This makes sure that
3595 users who are logged out cannot continue to consume IPC
3596 resources. This covers SysV memory, semaphores and message
3597 queues as well as POSIX shared memory and message
3598 queues. Traditionally, SysV and POSIX IPC had no life-cycle
3599 limits. With this functionality, that is corrected. This may
3600 be turned off by using the RemoveIPC= switch of logind.conf.
3602 * The systemd-machine-id-setup and tmpfiles tools gained a
3603 --root= switch to operate on a specific root directory,
3606 * journald can now forward logged messages to the TTYs of all
3607 logged in users ("wall"). This is the default for all
3608 emergency messages now.
3610 * A new tool systemd-journal-remote has been added to stream
3611 journal log messages across the network.
3613 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
3614 controller trees are mounted into it. Note that the
3615 directories mounted beneath it are not read-only. This is a
3616 security measure and is particularly useful because glibc
3617 actually includes a search logic to pick any tmpfs it can
3618 find to implement shm_open() if /dev/shm is not available
3619 (which it might very well be in namespaced setups).
3621 * machinectl gained a new "poweroff" command to cleanly power
3622 down a local OS container.
3624 * The PrivateDevices= unit file setting will now also drop the
3625 CAP_MKNOD capability from the capability bound set, and
3626 imply DevicePolicy=closed.
3628 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
3629 comprehensively on all long-running systemd services where
3630 this is appropriate.
3632 * systemd-udevd will now run in a disassociated mount
3633 namespace. To mount directories from udev rules, make sure to
3634 pull in mount units via SYSTEMD_WANTS properties.
3636 * The kdbus support gained support for uploading policy into
3637 the kernel. sd-bus gained support for creating "monitoring"
3638 connections that can eavesdrop into all bus communication
3639 for debugging purposes.
3641 * Timestamps may now be specified in seconds since the UNIX
3642 epoch Jan 1st, 1970 by specifying "@" followed by the value
3645 * Native tcpwrap support in systemd has been removed. tcpwrap
3646 is old code, not really maintained anymore and has serious
3647 shortcomings, and better options such as firewalls
3648 exist. For setups that require tcpwrap usage, please
3649 consider invoking your socket-activated service via tcpd,
3650 like on traditional inetd.
3652 * A new system.conf configuration option
3653 DefaultTimerAccuracySec= has been added that controls the
3654 default AccuracySec= setting of .timer units.
3656 * Timer units gained a new WakeSystem= switch. If enabled,
3657 timers configured this way will cause the system to resume
3658 from system suspend (if the system supports that, which most
3661 * Timer units gained a new Persistent= switch. If enabled,
3662 timers configured this way will save to disk when they have
3663 been last triggered. This information is then used on next
3664 reboot to possible execute overdue timer events, that
3665 could not take place because the system was powered off.
3666 This enables simple anacron-like behaviour for timer units.
3668 * systemctl's "list-timers" will now also list the time a
3669 timer unit was last triggered in addition to the next time
3670 it will be triggered.
3672 * systemd-networkd will now assign predictable IPv4LL
3673 addresses to its local interfaces.
3675 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
3676 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
3677 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
3678 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
3679 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
3680 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
3681 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
3682 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
3685 — Berlin, 2014-03-25
3689 * A new unit file setting RestrictAddressFamilies= has been
3690 added to restrict which socket address families unit
3691 processes gain access to. This takes address family names
3692 like "AF_INET" or "AF_UNIX", and is useful to minimize the
3693 attack surface of services via exotic protocol stacks. This
3694 is built on seccomp system call filters.
3696 * Two new unit file settings RuntimeDirectory= and
3697 RuntimeDirectoryMode= have been added that may be used to
3698 manage a per-daemon runtime directories below /run. This is
3699 an alternative for setting up directory permissions with
3700 tmpfiles snippets, and has the advantage that the runtime
3701 directory's lifetime is bound to the daemon runtime and that
3702 the daemon starts up with an empty directory each time. This
3703 is particularly useful when writing services that drop
3704 privileges using the User= or Group= setting.
3706 * The DeviceAllow= unit setting now supports globbing for
3707 matching against device group names.
3709 * The systemd configuration file system.conf gained new
3710 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
3711 DefaultMemoryAccounting= to globally turn on/off accounting
3712 for specific resources (cgroups) for all units. These
3713 settings may still be overridden individually in each unit
3716 * systemd-gpt-auto-generator is now able to discover /srv and
3717 root partitions in addition to /home and swap partitions. It
3718 also supports LUKS-encrypted partitions now. With this in
3719 place, automatic discovery of partitions to mount following
3720 the Discoverable Partitions Specification
3721 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
3722 is now a lot more complete. This allows booting without
3723 /etc/fstab and without root= on the kernel command line on
3724 systems prepared appropriately.
3726 * systemd-nspawn gained a new --image= switch which allows
3727 booting up disk images and Linux installations on any block
3728 device that follow the Discoverable Partitions Specification
3729 (see above). This means that installations made with
3730 appropriately updated installers may now be started and
3731 deployed using container managers, completely
3732 unmodified. (We hope that libvirt-lxc will add support for
3733 this feature soon, too.)
3735 * systemd-nspawn gained a new --network-macvlan= setting to
3736 set up a private macvlan interface for the
3737 container. Similarly, systemd-networkd gained a new
3738 Kind=macvlan setting in .netdev files.
3740 * systemd-networkd now supports configuring local addresses
3743 * A new tool systemd-network-wait-online has been added to
3744 synchronously wait for network connectivity using
3747 * The sd-bus.h bus API gained a new sd_bus_track object for
3748 tracking the life-cycle of bus peers. Note that sd-bus.h is
3749 still not a public API though (unless you specify
3750 --enable-kdbus on the configure command line, which however
3751 voids your warranty and you get no API stability guarantee).
3753 * The $XDG_RUNTIME_DIR runtime directories for each user are
3754 now individual tmpfs instances, which has the benefit of
3755 introducing separate pools for each user, with individual
3756 size limits, and thus making sure that unprivileged clients
3757 can no longer negatively impact the system or other users by
3758 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
3759 RuntimeDirectorySize= has been introduced that allows
3760 controlling the default size limit for all users. It
3761 defaults to 10% of the available physical memory. This is no
3762 replacement for quotas on tmpfs though (which the kernel
3763 still does not support), as /dev/shm and /tmp are still
3764 shared resources used by both the system and unprivileged
3767 * logind will now automatically turn off automatic suspending
3768 on laptop lid close when more than one display is
3769 connected. This was previously expected to be implemented
3770 individually in desktop environments (such as GNOME),
3771 however has been added to logind now, in order to fix a
3772 boot-time race where a desktop environment might not have
3773 been started yet and thus not been able to take an inhibitor
3774 lock at the time where logind already suspends the system
3775 due to a closed lid.
3777 * logind will now wait at least 30s after each system
3778 suspend/resume cycle, and 3min after system boot before
3779 suspending the system due to a closed laptop lid. This
3780 should give USB docking stations and similar enough time to
3781 be probed and configured after system resume and boot in
3782 order to then act as suspend blocker.
3784 * systemd-run gained a new --property= setting which allows
3785 initialization of resource control properties (and others)
3786 for the created scope or service unit. Example: "systemd-run
3787 --property=BlockIOWeight=10 updatedb" may be used to run
3788 updatedb at a low block IO scheduling weight.
3790 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
3791 now also work in --scope mode.
3793 * When systemd is compiled with kdbus support, basic support
3794 for enforced policies is now in place. (Note that enabling
3795 kdbus still voids your warranty and no API compatibility
3798 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
3799 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
3800 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
3801 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
3802 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
3803 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
3804 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
3805 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
3806 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
3807 Zbigniew Jędrzejewski-Szmek
3809 — Berlin, 2014-03-12
3813 * systemd will now relabel /dev after loading the SMACK policy
3814 according to SMACK rules.
3816 * A new unit file option AppArmorProfile= has been added to
3817 set the AppArmor profile for the processes of a unit.
3819 * A new condition check ConditionArchitecture= has been added
3820 to conditionalize units based on the system architecture, as
3821 reported by uname()'s "machine" field.
3823 * systemd-networkd now supports matching on the system
3824 virtualization, architecture, kernel command line, host name
3827 * logind is now a lot more aggressive when suspending the
3828 machine due to a closed laptop lid. Instead of acting only
3829 on the lid close action, it will continuously watch the lid
3830 status and act on it. This is useful for laptops where the
3831 power button is on the outside of the chassis so that it can
3832 be reached without opening the lid (such as the Lenovo
3833 Yoga). On those machines, logind will now immediately
3834 re-suspend the machine if the power button has been
3835 accidentally pressed while the laptop was suspended and in a
3836 backpack or similar.
3838 * logind will now watch SW_DOCK switches and inhibit reaction
3839 to the lid switch if it is pressed. This means that logind
3840 will not suspend the machine anymore if the lid is closed
3841 and the system is docked, if the laptop supports SW_DOCK
3842 notifications via the input layer. Note that ACPI docking
3843 stations do not generate this currently. Also note that this
3844 logic is usually not fully sufficient and Desktop
3845 Environments should take a lid switch inhibitor lock when an
3846 external display is connected, as systemd will not watch
3849 * nspawn will now make use of the devices cgroup controller by
3850 default, and only permit creation of and access to the usual
3851 API device nodes like /dev/null or /dev/random, as well as
3852 access to (but not creation of) the pty devices.
3854 * We will now ship a default .network file for
3855 systemd-networkd that automatically configures DHCP for
3856 network interfaces created by nspawn's --network-veth or
3857 --network-bridge= switches.
3859 * systemd will now understand the usual M, K, G, T suffixes
3860 according to SI conventions (i.e. to the base 1000) when
3861 referring to throughput and hardware metrics. It will stay
3862 with IEC conventions (i.e. to the base 1024) for software
3863 metrics, according to what is customary according to
3864 Wikipedia. We explicitly document which base applies for
3865 each configuration option.
3867 * The DeviceAllow= setting in unit files now supports a syntax
3868 to whitelist an entire group of devices node majors at once,
3869 based on the /proc/devices listing. For example, with the
3870 string "char-pts", it is now possible to whitelist all
3871 current and future pseudo-TTYs at once.
3873 * sd-event learned a new "post" event source. Event sources of
3874 this type are triggered by the dispatching of any event
3875 source of a type that is not "post". This is useful for
3876 implementing clean-up and check event sources that are
3877 triggered by other work being done in the program.
3879 * systemd-networkd is no longer statically enabled, but uses
3880 the usual [Install] sections so that it can be
3881 enabled/disabled using systemctl. It still is enabled by
3884 * When creating a veth interface pair with systemd-nspawn, the
3885 host side will now be prefixed with "vb-" if
3886 --network-bridge= is used, and with "ve-" if --network-veth
3887 is used. This way, it is easy to distinguish these cases on
3888 the host, for example to apply different configuration to
3889 them with systemd-networkd.
3891 * The compatibility libraries for libsystemd-journal.so,
3892 libsystem-id128.so, libsystemd-login.so and
3893 libsystemd-daemon.so do not make use of IFUNC
3894 anymore. Instead, we now build libsystemd.so multiple times
3895 under these alternative names. This means that the footprint
3896 is drastically increased, but given that these are
3897 transitional compatibility libraries, this should not matter
3898 much. This change has been made necessary to support the ARM
3899 platform for these compatibility libraries, as the ARM
3900 toolchain is not really at the same level as the toolchain
3901 for other architectures like x86 and does not support
3902 IFUNC. Please make sure to use --enable-compat-libs only
3903 during a transitional period!
3905 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
3906 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
3907 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
3908 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
3909 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
3910 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
3911 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
3912 Zbigniew Jędrzejewski-Szmek
3914 — Berlin, 2014-02-24
3918 * A new component "systemd-networkd" has been added that can
3919 be used to configure local network interfaces statically or
3920 via DHCP. It is capable of bringing up bridges, VLANs, and
3921 bonding. Currently, no hook-ups for interactive network
3922 configuration are provided. Use this for your initrd,
3923 container, embedded, or server setup if you need a simple,
3924 yet powerful, network configuration solution. This
3925 configuration subsystem is quite nifty, as it allows wildcard
3926 hotplug matching in interfaces. For example, with a single
3927 configuration snippet, you can configure that all Ethernet
3928 interfaces showing up are automatically added to a bridge,
3929 or similar. It supports link-sensing and more.
3931 * A new tool "systemd-socket-proxyd" has been added which can
3932 act as a bidirectional proxy for TCP sockets. This is
3933 useful for adding socket activation support to services that
3934 do not actually support socket activation, including virtual
3935 machines and the like.
3937 * Add a new tool to save/restore rfkill state on
3940 * Save/restore state of keyboard backlights in addition to
3941 display backlights on shutdown/boot.
3943 * udev learned a new SECLABEL{} construct to label device
3944 nodes with a specific security label when they appear. For
3945 now, only SECLABEL{selinux} is supported, but the syntax is
3946 prepared for additional security frameworks.
3948 * udev gained a new scheme to configure link-level attributes
3949 from files in /etc/systemd/network/*.link. These files can
3950 match against MAC address, device path, driver name and type,
3951 and will apply attributes like the naming policy, link speed,
3952 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
3953 address assignment policy (randomized, ...).
3955 * The configuration of network interface naming rules for
3956 "permanent interface names" has changed: a new NamePolicy=
3957 setting in the [Link] section of .link files determines the
3958 priority of possible naming schemes (onboard, slot, MAC,
3959 path). The default value of this setting is determined by
3960 /usr/lib/net/links/99-default.link. Old
3961 80-net-name-slot.rules udev configuration file has been
3962 removed, so local configuration overriding this file should
3963 be adapted to override 99-default.link instead.
3965 * When the User= switch is used in a unit file, also
3966 initialize $SHELL= based on the user database entry.
3968 * systemd no longer depends on libdbus. All communication is
3969 now done with sd-bus, systemd's low-level bus library
3972 * kdbus support has been added to PID 1 itself. When kdbus is
3973 enabled, this causes PID 1 to set up the system bus and
3974 enable support for a new ".busname" unit type that
3975 encapsulates bus name activation on kdbus. It works a little
3976 bit like ".socket" units, except for bus names. A new
3977 generator has been added that converts classic dbus1 service
3978 activation files automatically into native systemd .busname
3981 * sd-bus: add a light-weight vtable implementation that allows
3982 defining objects on the bus with a simple static const
3983 vtable array of its methods, signals and properties.
3985 * systemd will not generate or install static dbus
3986 introspection data anymore to /usr/share/dbus-1/interfaces,
3987 as the precise format of these files is unclear, and
3988 nothing makes use of it.
3990 * A proxy daemon is now provided to proxy clients connecting
3991 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
3992 compatibility with classic D-Bus.
3994 * A bus driver implementation has been added that supports the
3995 classic D-Bus bus driver calls on kdbus, also for
3996 compatibility purposes.
3998 * A new API "sd-event.h" has been added that implements a
3999 minimal event loop API built around epoll. It provides a
4000 couple of features that direct epoll usage is lacking:
4001 prioritization of events, scales to large numbers of timer
4002 events, per-event timer slack (accuracy), system-wide
4003 coalescing of timer events, exit handlers, watchdog
4004 supervision support using systemd's sd_notify() API, child
4007 * A new API "sd-rntl.h" has been added that provides an API
4008 around the route netlink interface of the kernel, similar in
4009 style to "sd-bus.h".
4011 * A new API "sd-dhcp-client.h" has been added that provides a
4012 small DHCPv4 client-side implementation. This is used by
4015 * There is a new kernel command line option
4016 "systemd.restore_state=0|1". When set to "0", none of the
4017 systemd tools will restore saved runtime state to hardware
4018 devices. More specifically, the rfkill and backlight states
4021 * The FsckPassNo= compatibility option in mount/service units
4022 has been removed. The fstab generator will now add the
4023 necessary dependencies automatically, and does not require
4024 PID1's support for that anymore.
4026 * journalctl gained a new switch, --list-boots, that lists
4027 recent boots with their times and boot IDs.
4029 * The various tools like systemctl, loginctl, timedatectl,
4030 busctl, systemd-run, ... have gained a new switch "-M" to
4031 connect to a specific, local OS container (as direct
4032 connection, without requiring SSH). This works on any
4033 container that is registered with machined, such as those
4034 created by libvirt-lxc or nspawn.
4036 * systemd-run and systemd-analyze also gained support for "-H"
4037 to connect to remote hosts via SSH. This is particularly
4038 useful for systemd-run because it enables queuing of jobs
4039 onto remote systems.
4041 * machinectl gained a new command "login" to open a getty
4042 login in any local container. This works with any container
4043 that is registered with machined (such as those created by
4044 libvirt-lxc or nspawn), and which runs systemd inside.
4046 * machinectl gained a new "reboot" command that may be used to
4047 trigger a reboot on a specific container that is registered
4048 with machined. This works on any container that runs an init
4049 system of some kind.
4051 * systemctl gained a new "list-timers" command to print a nice
4052 listing of installed timer units with the times they elapse
4055 * Alternative reboot() parameters may now be specified on the
4056 "systemctl reboot" command line and are passed to the
4057 reboot() system call.
4059 * systemctl gained a new --job-mode= switch to configure the
4060 mode to queue a job with. This is a more generic version of
4061 --fail, --irreversible, and --ignore-dependencies, which are
4062 still available but not advertised anymore.
4064 * /etc/systemd/system.conf gained new settings to configure
4065 various default timeouts of units, as well as the default
4066 start limit interval and burst. These may still be overridden
4069 * PID1 will now export on the bus profile data of the security
4070 policy upload process (such as the SELinux policy upload to
4073 * journald: when forwarding logs to the console, include
4074 timestamps (following the setting in
4075 /sys/module/printk/parameters/time).
4077 * OnCalendar= in timer units now understands the special
4078 strings "yearly" and "annually". (Both are equivalent)
4080 * The accuracy of timer units is now configurable with the new
4081 AccuracySec= setting. It defaults to 1min.
4083 * A new dependency type JoinsNamespaceOf= has been added that
4084 allows running two services within the same /tmp and network
4085 namespace, if PrivateNetwork= or PrivateTmp= are used.
4087 * A new command "cat" has been added to systemctl. It outputs
4088 the original unit file of a unit, and concatenates the
4089 contents of additional "drop-in" unit file snippets, so that
4090 the full configuration is shown.
4092 * systemctl now supports globbing on the various "list-xyz"
4093 commands, like "list-units" or "list-sockets", as well as on
4094 those commands which take multiple unit names.
4096 * journalctl's --unit= switch gained support for globbing.
4098 * All systemd daemons now make use of the watchdog logic so
4099 that systemd automatically notices when they hang.
4101 * If the $container_ttys environment variable is set,
4102 getty-generator will automatically spawn a getty for each
4103 listed tty. This is useful for container managers to request
4104 login gettys to be spawned on as many ttys as needed.
4106 * %h, %s, %U specifier support is not available anymore when
4107 used in unit files for PID 1. This is because NSS calls are
4108 not safe from PID 1. They stay available for --user
4109 instances of systemd, and as special case for the root user.
4111 * loginctl gained a new "--no-legend" switch to turn off output
4114 * The "sd-login.h" API gained three new calls:
4115 sd_session_is_remote(), sd_session_get_remote_user(),
4116 sd_session_get_remote_host() to query information about
4119 * The udev hardware database now also carries vendor/product
4120 information of SDIO devices.
4122 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
4123 determine whether watchdog notifications are requested by
4126 * Socket-activated per-connection services now include a
4127 short description of the connection parameters in the
4130 * tmpfiles gained a new "--boot" option. When this is not used,
4131 only lines where the command character is not suffixed with
4132 "!" are executed. When this option is specified, those
4133 options are executed too. This partitions tmpfiles
4134 directives into those that can be safely executed at any
4135 time, and those which should be run only at boot (for
4136 example, a line that creates /run/nologin).
4138 * A new API "sd-resolve.h" has been added which provides a simple
4139 asynchronous wrapper around glibc NSS host name resolution
4140 calls, such as getaddrinfo(). In contrast to glibc's
4141 getaddrinfo_a(), it does not use signals. In contrast to most
4142 other asynchronous name resolution libraries, this one does
4143 not reimplement DNS, but reuses NSS, so that alternate
4144 host name resolution systems continue to work, such as mDNS,
4145 LDAP, etc. This API is based on libasyncns, but it has been
4146 cleaned up for inclusion in systemd.
4148 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
4149 "sd-daemon.h" are no longer found in individual libraries
4150 libsystemd-journal.so, libsystemd-login.so,
4151 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
4152 merged them into a single library, libsystemd.so, which
4153 provides all symbols. The reason for this is cyclic
4154 dependencies, as these libraries tend to use each other's
4155 symbols. So far, we have managed to workaround that by linking
4156 a copy of a good part of our code into each of these
4157 libraries again and again, which, however, makes certain
4158 things hard to do, like sharing static variables. Also, it
4159 substantially increases footprint. With this change, there
4160 is only one library for the basic APIs systemd
4161 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
4162 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
4163 library as well, however are subject to the --enable-kdbus
4164 switch (see below). Note that "sd-dhcp-client.h" is not part
4165 of this library (this is because it only consumes, never
4166 provides, services of/to other APIs). To make the transition
4167 easy from the separate libraries to the unified one, we
4168 provide the --enable-compat-libs compile-time switch which
4169 will generate stub libraries that are compatible with the
4170 old ones but redirect all calls to the new one.
4172 * All of the kdbus logic and the new APIs "sd-bus.h",
4173 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
4174 and "sd-utf8.h" are compile-time optional via the
4175 "--enable-kdbus" switch, and they are not compiled in by
4176 default. To make use of kdbus, you have to explicitly enable
4177 the switch. Note however, that neither the kernel nor the
4178 userspace API for all of this is considered stable yet. We
4179 want to maintain the freedom to still change the APIs for
4180 now. By specifying this build-time switch, you acknowledge
4181 that you are aware of the instability of the current
4184 * Also, note that while kdbus is pretty much complete,
4185 it lacks one thing: proper policy support. This means you
4186 can build a fully working system with all features; however,
4187 it will be highly insecure. Policy support will be added in
4188 one of the next releases, at the same time that we will
4189 declare the APIs stable.
4191 * When the kernel command line argument "kdbus" is specified,
4192 systemd will automatically load the kdbus.ko kernel module. At
4193 this stage of development, it is only useful for testing kdbus
4194 and should not be used in production. Note: if "--enable-kdbus"
4195 is specified, and the kdbus.ko kernel module is available, and
4196 "kdbus" is added to the kernel command line, the entire system
4197 runs with kdbus instead of dbus-daemon, with the above mentioned
4198 problem of missing the system policy enforcement. Also a future
4199 version of kdbus.ko or a newer systemd will not be compatible with
4200 each other, and will unlikely be able to boot the machine if only
4201 one of them is updated.
4203 * systemctl gained a new "import-environment" command which
4204 uploads the caller's environment (or parts thereof) into the
4205 service manager so that it is inherited by services started
4206 by the manager. This is useful to upload variables like
4207 $DISPLAY into the user service manager.
4209 * A new PrivateDevices= switch has been added to service units
4210 which allows running a service with a namespaced /dev
4211 directory that does not contain any device nodes for
4212 physical devices. More specifically, it only includes devices
4213 such as /dev/null, /dev/urandom, and /dev/zero which are API
4216 * logind has been extended to support behaviour like VT
4217 switching on seats that do not support a VT. This makes
4218 multi-session available on seats that are not the first seat
4219 (seat0), and on systems where kernel support for VTs has
4220 been disabled at compile-time.
4222 * If a process holds a delay lock for system sleep or shutdown
4223 and fails to release it in time, we will now log its
4224 identity. This makes it easier to identify processes that
4225 cause slow suspends or power-offs.
4227 * When parsing /etc/crypttab, support for a new key-slot=
4228 option as supported by Debian is added. It allows indicating
4229 which LUKS slot to use on disk, speeding up key loading.
4231 * The sd_journald_sendv() API call has been checked and
4232 officially declared to be async-signal-safe so that it may
4233 be invoked from signal handlers for logging purposes.
4235 * Boot-time status output is now enabled automatically after a
4236 short timeout if boot does not progress, in order to give
4237 the user an indication what she or he is waiting for.
4239 * The boot-time output has been improved to show how much time
4240 remains until jobs expire.
4242 * The KillMode= switch in service units gained a new possible
4243 value "mixed". If set, and the unit is shut down, then the
4244 initial SIGTERM signal is sent only to the main daemon
4245 process, while the following SIGKILL signal is sent to
4246 all remaining processes of the service.
4248 * When a scope unit is registered, a new property "Controller"
4249 may be set. If set to a valid bus name, systemd will send a
4250 RequestStop() signal to this name when it would like to shut
4251 down the scope. This may be used to hook manager logic into
4252 the shutdown logic of scope units. Also, scope units may now
4253 be put in a special "abandoned" state, in which case the
4254 manager process which created them takes no further
4255 responsibilities for it.
4257 * When reading unit files, systemd will now verify
4258 the access mode of these files, and warn about certain
4259 suspicious combinations. This has been added to make it
4260 easier to track down packaging bugs where unit files are
4261 marked executable or world-writable.
4263 * systemd-nspawn gained a new "--setenv=" switch to set
4264 container-wide environment variables. The similar option in
4265 systemd-activate was renamed from "--environment=" to
4266 "--setenv=" for consistency.
4268 * systemd-nspawn has been updated to create a new kdbus domain
4269 for each container that is invoked, thus allowing each
4270 container to have its own set of system and user buses,
4271 independent of the host.
4273 * systemd-nspawn gained a new --drop-capability= switch to run
4274 the container with less capabilities than the default. Both
4275 --drop-capability= and --capability= now take the special
4276 string "all" for dropping or keeping all capabilities.
4278 * systemd-nspawn gained new switches for executing containers
4279 with specific SELinux labels set.
4281 * systemd-nspawn gained a new --quiet switch to not generate
4282 any additional output but the container's own console
4285 * systemd-nspawn gained a new --share-system switch to run a
4286 container without PID namespacing enabled.
4288 * systemd-nspawn gained a new --register= switch to control
4289 whether the container is registered with systemd-machined or
4290 not. This is useful for containers that do not run full
4291 OS images, but only specific apps.
4293 * systemd-nspawn gained a new --keep-unit which may be used
4294 when invoked as the only program from a service unit, and
4295 results in registration of the unit service itself in
4296 systemd-machined, instead of a newly opened scope unit.
4298 * systemd-nspawn gained a new --network-interface= switch for
4299 moving arbitrary interfaces to the container. The new
4300 --network-veth switch creates a virtual Ethernet connection
4301 between host and container. The new --network-bridge=
4302 switch then allows assigning the host side of this virtual
4303 Ethernet connection to a bridge device.
4305 * systemd-nspawn gained a new --personality= switch for
4306 setting the kernel personality for the container. This is
4307 useful when running a 32-bit container on a 64-bit host. A
4308 similar option Personality= is now also available for service
4311 * logind will now also track a "Desktop" identifier for each
4312 session which encodes the desktop environment of it. This is
4313 useful for desktop environments that want to identify
4314 multiple running sessions of itself easily.
4316 * A new SELinuxContext= setting for service units has been
4317 added that allows setting a specific SELinux execution
4318 context for a service.
4320 * Most systemd client tools will now honour $SYSTEMD_LESS for
4321 settings of the "less" pager. By default, these tools will
4322 override $LESS to allow certain operations to work, such as
4323 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
4324 influence this logic.
4326 * systemd's "seccomp" hook-up has been changed to make use of
4327 the libseccomp library instead of using its own
4328 implementation. This has benefits for portability among
4331 * For usage together with SystemCallFilter=, a new
4332 SystemCallErrorNumber= setting has been introduced that
4333 allows configuration of a system error number to be returned
4334 on filtered system calls, instead of immediately killing the
4335 process. Also, SystemCallArchitectures= has been added to
4336 limit access to system calls of a particular architecture
4337 (in order to turn off support for unused secondary
4338 architectures). There is also a global
4339 SystemCallArchitectures= setting in system.conf now to turn
4340 off support for non-native system calls system-wide.
4342 * systemd requires a kernel with a working name_to_handle_at(),
4343 please see the kernel config requirements in the README file.
4345 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
4346 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
4347 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
4348 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
4349 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
4350 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
4351 Elia Pinto, Florian Weimer, George McCollister, Goffredo
4352 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
4353 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
4354 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
4355 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
4356 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
4357 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
4358 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
4359 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
4360 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
4361 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
4362 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
4363 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
4364 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
4365 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
4366 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
4367 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
4368 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
4370 — Berlin, 2014-02-20
4374 * logind has gained support for facilitating privileged input
4375 and drm device access for unprivileged clients. This work is
4376 useful to allow Wayland display servers (and similar
4377 programs, such as kmscon) to run under the user's ID and
4378 access input and drm devices which are normally
4379 protected. When this is used (and the kernel is new enough)
4380 logind will "mute" IO on the file descriptors passed to
4381 Wayland as long as it is in the background and "unmute" it
4382 if it returns into the foreground. This allows secure
4383 session switching without allowing background sessions to
4384 eavesdrop on input and display data. This also introduces
4385 session switching support if VT support is turned off in the
4386 kernel, and on seats that are not seat0.
4388 * A new kernel command line option luks.options= is understood
4389 now which allows specifying LUKS options for usage for LUKS
4390 encrypted partitions specified with luks.uuid=.
4392 * tmpfiles.d(5) snippets may now use specifier expansion in
4393 path names. More specifically %m, %b, %H, %v, are now
4394 replaced by the local machine id, boot id, hostname, and
4395 kernel version number.
4397 * A new tmpfiles.d(5) command "m" has been introduced which
4398 may be used to change the owner/group/access mode of a file
4399 or directory if it exists, but do nothing if it does not.
4401 * This release removes high-level support for the
4402 MemorySoftLimit= cgroup setting. The underlying kernel
4403 cgroup attribute memory.soft_limit= is currently badly
4404 designed and likely to be removed from the kernel API in its
4405 current form, hence we should not expose it for now.
4407 * The memory.use_hierarchy cgroup attribute is now enabled for
4408 all cgroups systemd creates in the memory cgroup
4409 hierarchy. This option is likely to be come the built-in
4410 default in the kernel anyway, and the non-hierarchical mode
4411 never made much sense in the intrinsically hierarchical
4414 * A new field _SYSTEMD_SLICE= is logged along with all journal
4415 messages containing the slice a message was generated
4416 from. This is useful to allow easy per-customer filtering of
4417 logs among other things.
4419 * systemd-journald will no longer adjust the group of journal
4420 files it creates to the "systemd-journal" group. Instead we
4421 rely on the journal directory to be owned by the
4422 "systemd-journal" group, and its setgid bit set, so that the
4423 kernel file system layer will automatically enforce that
4424 journal files inherit this group assignment. The reason for
4425 this change is that we cannot allow NSS look-ups from
4426 journald which would be necessary to resolve
4427 "systemd-journal" to a numeric GID, because this might
4428 create deadlocks if NSS involves synchronous queries to
4429 other daemons (such as nscd, or sssd) which in turn are
4430 logging clients of journald and might block on it, which
4431 would then dead lock. A tmpfiles.d(5) snippet included in
4432 systemd will make sure the setgid bit and group are
4433 properly set on the journal directory if it exists on every
4434 boot. However, we recommend adjusting it manually after
4435 upgrades too (or from RPM scriptlets), so that the change is
4436 not delayed until next reboot.
4438 * Backlight and random seed files in /var/lib/ have moved into
4439 the /var/lib/systemd/ directory, in order to centralize all
4440 systemd generated files in one directory.
4442 * Boot time performance measurements (as displayed by
4443 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
4444 performance information if that's available to determine how
4445 much time BIOS and boot loader initialization required. With
4446 a sufficiently new BIOS you hence no longer need to boot
4447 with Gummiboot to get access to such information.
4449 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
4450 Cristian Rodríguez, Dave Reisner, David Herrmann, David
4451 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
4452 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
4453 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
4454 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
4455 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
4457 — Berlin, 2013-10-02
4461 * The Restart= option for services now understands a new
4462 on-watchdog setting, which will restart the service
4463 automatically if the service stops sending out watchdog keep
4464 alive messages (as configured with WatchdogSec=).
4466 * The getty generator (which is responsible for bringing up a
4467 getty on configured serial consoles) will no longer only
4468 start a getty on the primary kernel console but on all
4469 others, too. This makes the order in which console= is
4470 specified on the kernel command line less important.
4472 * libsystemd-logind gained a new sd_session_get_vt() call to
4473 retrieve the VT number of a session.
4475 * If the option "tries=0" is set for an entry of /etc/crypttab
4476 its passphrase is queried indefinitely instead of any
4477 maximum number of tries.
4479 * If a service with a configure PID file terminates its PID
4480 file will now be removed automatically if it still exists
4481 afterwards. This should put an end to stale PID files.
4483 * systemd-run will now also take relative binary path names
4484 for execution and no longer insists on absolute paths.
4486 * InaccessibleDirectories= and ReadOnlyDirectories= now take
4487 paths that are optionally prefixed with "-" to indicate that
4488 it should not be considered a failure if they do not exist.
4490 * journalctl -o (and similar commands) now understands a new
4491 output mode "short-precise", it is similar to "short" but
4492 shows timestamps with usec accuracy.
4494 * The option "discard" (as known from Debian) is now
4495 synonymous to "allow-discards" in /etc/crypttab. In fact,
4496 "discard" is preferred now (since it is easier to remember
4499 * Some licensing clean-ups were made, so that more code is now
4500 LGPL-2.1 licensed than before.
4502 * A minimal tool to save/restore the display backlight
4503 brightness across reboots has been added. It will store the
4504 backlight setting as late as possible at shutdown, and
4505 restore it as early as possible during reboot.
4507 * A logic to automatically discover and enable home and swap
4508 partitions on GPT disks has been added. With this in place
4509 /etc/fstab becomes optional for many setups as systemd can
4510 discover certain partitions located on the root disk
4511 automatically. Home partitions are recognized under their
4512 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
4513 partitions are recognized under their GPT type ID
4514 0657fd6da4ab43c484e50933c84b4f4f.
4516 * systemd will no longer pass any environment from the kernel
4517 or initrd to system services. If you want to set an
4518 environment for all services, do so via the kernel command
4519 line systemd.setenv= assignment.
4521 * The systemd-sysctl tool no longer natively reads the file
4522 /etc/sysctl.conf. If desired, the file should be symlinked
4523 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
4524 legacy support by a symlink rather than built-in code, it
4525 also makes the otherwise hidden order of application of the
4526 different files visible. (Note that this partly reverts to a
4527 pre-198 application order of sysctl knobs!)
4529 * The "systemctl set-log-level" and "systemctl dump" commands
4530 have been moved to systemd-analyze.
4532 * systemd-run learned the new --remain-after-exit switch,
4533 which causes the scope unit not to be cleaned up
4534 automatically after the process terminated.
4536 * tmpfiles learned a new --exclude-prefix= switch to exclude
4537 certain paths from operation.
4539 * journald will now automatically flush all messages to disk
4540 as soon as a message at the log level CRIT, ALERT or EMERG
4543 Contributions from: Andrew Cook, Brandon Philips, Christian
4544 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
4545 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
4546 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
4547 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
4548 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
4549 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
4550 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
4551 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
4552 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
4553 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
4554 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
4555 William Giokas, Zbigniew Jędrzejewski-Szmek
4557 — Berlin, 2013-09-13
4561 * The documentation has been updated to cover the various new
4562 concepts introduced with 205.
4564 * Unit files now understand the new %v specifier which
4565 resolves to the kernel version string as returned by "uname
4568 * systemctl now supports filtering the unit list output by
4569 load state, active state and sub state, using the new
4572 * "systemctl status" will now show the results of the
4573 condition checks (like ConditionPathExists= and similar) of
4574 the last start attempts of the unit. They are also logged to
4577 * "journalctl -b" may now be used to look for boot output of a
4578 specific boot. Try "journalctl -b -1" for the previous boot,
4579 but the syntax is substantially more powerful.
4581 * "journalctl --show-cursor" has been added which prints the
4582 cursor string the last shown log line. This may then be used
4583 with the new "journalctl --after-cursor=" switch to continue
4584 browsing logs from that point on.
4586 * "journalctl --force" may now be used to force regeneration
4589 * Creation of "dead" device nodes has been moved from udev
4590 into kmod and tmpfiles. Previously, udev would read the kmod
4591 databases to pre-generate dead device nodes based on meta
4592 information contained in kernel modules, so that these would
4593 be auto-loaded on access rather then at boot. As this
4594 does not really have much to do with the exposing actual
4595 kernel devices to userspace this has always been slightly
4596 alien in the udev codebase. Following the new scheme kmod
4597 will now generate a runtime snippet for tmpfiles from the
4598 module meta information and it now is tmpfiles' job to the
4599 create the nodes. This also allows overriding access and
4600 other parameters for the nodes using the usual tmpfiles
4601 facilities. As side effect this allows us to remove the
4602 CAP_SYS_MKNOD capability bit from udevd entirely.
4604 * logind's device ACLs may now be applied to these "dead"
4605 devices nodes too, thus finally allowing managed access to
4606 devices such as /dev/snd/sequencer without loading the
4607 backing module right-away.
4609 * A new RPM macro has been added that may be used to apply
4610 tmpfiles configuration during package installation.
4612 * systemd-detect-virt and ConditionVirtualization= now can
4613 detect User-Mode-Linux machines (UML).
4615 * journald will now implicitly log the effective capabilities
4616 set of processes in the message metadata.
4618 * systemd-cryptsetup has gained support for TrueCrypt volumes.
4620 * The initrd interface has been simplified (more specifically,
4621 support for passing performance data via environment
4622 variables and fsck results via files in /run has been
4623 removed). These features were non-essential, and are
4624 nowadays available in a much nicer way by having systemd in
4625 the initrd serialize its state and have the hosts systemd
4626 deserialize it again.
4628 * The udev "keymap" data files and tools to apply keyboard
4629 specific mappings of scan to key codes, and force-release
4630 scan code lists have been entirely replaced by a udev
4631 "keyboard" builtin and a hwdb data file.
4633 * systemd will now honour the kernel's "quiet" command line
4634 argument also during late shutdown, resulting in a
4635 completely silent shutdown when used.
4637 * There's now an option to control the SO_REUSEPORT socket
4638 option in .socket units.
4640 * Instance units will now automatically get a per-template
4641 subslice of system.slice unless something else is explicitly
4642 configured. For example, instances of sshd@.service will now
4643 implicitly be placed in system-sshd.slice rather than
4644 system.slice as before.
4646 * Test coverage support may now be enabled at build time.
4648 Contributions from: Dave Reisner, Frederic Crozat, Harald
4649 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
4650 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
4651 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
4652 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
4653 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
4654 Giokas, Zbigniew Jędrzejewski-Szmek
4656 — Berlin, 2013-07-23
4660 * Two new unit types have been introduced:
4662 Scope units are very similar to service units, however, are
4663 created out of pre-existing processes — instead of PID 1
4664 forking off the processes. By using scope units it is
4665 possible for system services and applications to group their
4666 own child processes (worker processes) in a powerful way
4667 which then maybe used to organize them, or kill them
4668 together, or apply resource limits on them.
4670 Slice units may be used to partition system resources in an
4671 hierarchical fashion and then assign other units to them. By
4672 default there are now three slices: system.slice (for all
4673 system services), user.slice (for all user sessions),
4674 machine.slice (for VMs and containers).
4676 Slices and scopes have been introduced primarily in
4677 context of the work to move cgroup handling to a
4678 single-writer scheme, where only PID 1
4679 creates/removes/manages cgroups.
4681 * There's a new concept of "transient" units. In contrast to
4682 normal units these units are created via an API at runtime,
4683 not from configuration from disk. More specifically this
4684 means it is now possible to run arbitrary programs as
4685 independent services, with all execution parameters passed
4686 in via bus APIs rather than read from disk. Transient units
4687 make systemd substantially more dynamic then it ever was,
4688 and useful as a general batch manager.
4690 * logind has been updated to make use of scope and slice units
4691 for managing user sessions. As a user logs in he will get
4692 his own private slice unit, to which all sessions are added
4693 as scope units. We also added support for automatically
4694 adding an instance of user@.service for the user into the
4695 slice. Effectively logind will no longer create cgroup
4696 hierarchies on its own now, it will defer entirely to PID 1
4697 for this by means of scope, service and slice units. Since
4698 user sessions this way become entities managed by PID 1
4699 the output of "systemctl" is now a lot more comprehensive.
4701 * A new mini-daemon "systemd-machined" has been added which
4702 may be used by virtualization managers to register local
4703 VMs/containers. nspawn has been updated accordingly, and
4704 libvirt will be updated shortly. machined will collect a bit
4705 of meta information about the VMs/containers, and assign
4706 them their own scope unit (see above). The collected
4707 meta-data is then made available via the "machinectl" tool,
4708 and exposed in "ps" and similar tools. machined/machinectl
4709 is compile-time optional.
4711 * As discussed earlier, the low-level cgroup configuration
4712 options ControlGroup=, ControlGroupModify=,
4713 ControlGroupPersistent=, ControlGroupAttribute= have been
4714 removed. Please use high-level attribute settings instead as
4715 well as slice units.
4717 * A new bus call SetUnitProperties() has been added to alter
4718 various runtime parameters of a unit. This is primarily
4719 useful to alter cgroup parameters dynamically in a nice way,
4720 but will be extended later on to make more properties
4721 modifiable at runtime. systemctl gained a new set-properties
4722 command that wraps this call.
4724 * A new tool "systemd-run" has been added which can be used to
4725 run arbitrary command lines as transient services or scopes,
4726 while configuring a number of settings via the command
4727 line. This tool is currently very basic, however already
4728 very useful. We plan to extend this tool to even allow
4729 queuing of execution jobs with time triggers from the
4730 command line, similar in fashion to "at".
4732 * nspawn will now inform the user explicitly that kernels with
4733 audit enabled break containers, and suggest the user to turn
4736 * Support for detecting the IMA and AppArmor security
4737 frameworks with ConditionSecurity= has been added.
4739 * journalctl gained a new "-k" switch for showing only kernel
4740 messages, mimicking dmesg output; in addition to "--user"
4741 and "--system" switches for showing only user's own logs
4744 * systemd-delta can now show information about drop-in
4745 snippets extending unit files.
4747 * libsystemd-bus has been substantially updated but is still
4748 not available as public API.
4750 * systemd will now look for the "debug" argument on the kernel
4751 command line and enable debug logging, similar to what
4752 "systemd.log_level=debug" already did before.
4754 * "systemctl set-default", "systemctl get-default" has been
4755 added to configure the default.target symlink, which
4756 controls what to boot into by default.
4758 * "systemctl set-log-level" has been added as a convenient
4759 way to raise and lower systemd logging threshold.
4761 * "systemd-analyze plot" will now show the time the various
4762 generators needed for execution, as well as information
4763 about the unit file loading.
4765 * libsystemd-journal gained a new sd_journal_open_files() call
4766 for opening specific journal files. journactl also gained a
4767 new switch to expose this new functionality. Previously we
4768 only supported opening all files from a directory, or all
4769 files from the system, as opening individual files only is
4770 racy due to journal file rotation.
4772 * systemd gained the new DefaultEnvironment= setting in
4773 /etc/systemd/system.conf to set environment variables for
4776 * If a privileged process logs a journal message with the
4777 OBJECT_PID= field set, then journald will automatically
4778 augment this with additional OBJECT_UID=, OBJECT_GID=,
4779 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
4780 system services want to log events about specific client
4781 processes. journactl/systemctl has been updated to make use
4782 of this information if all log messages regarding a specific
4785 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
4786 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
4787 Reisner, David Coppa, David King, David Strauss, Eelco
4788 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
4789 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
4790 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
4791 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
4792 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
4793 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
4794 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
4795 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
4796 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
4797 Łukasz Stelmach, 장동준
4801 * The Python bindings gained some minimal support for the APIs
4802 exposed by libsystemd-logind.
4804 * ConditionSecurity= gained support for detecting SMACK. Since
4805 this condition already supports SELinux and AppArmor we only
4806 miss IMA for this. Patches welcome!
4808 Contributions from: Karol Lewandowski, Lennart Poettering,
4809 Zbigniew Jędrzejewski-Szmek
4813 * systemd-nspawn will now create /etc/resolv.conf if
4814 necessary, before bind-mounting the host's file onto it.
4816 * systemd-nspawn will now store meta information about a
4817 container on the container's cgroup as extended attribute
4818 fields, including the root directory.
4820 * The cgroup hierarchy has been reworked in many ways. All
4821 objects any of the components systemd creates in the cgroup
4822 tree are now suffixed. More specifically, user sessions are
4823 now placed in cgroups suffixed with ".session", users in
4824 cgroups suffixed with ".user", and nspawn containers in
4825 cgroups suffixed with ".nspawn". Furthermore, all cgroup
4826 names are now escaped in a simple scheme to avoid collision
4827 of userspace object names with kernel filenames. This work
4828 is preparation for making these objects relocatable in the
4829 cgroup tree, in order to allow easy resource partitioning of
4830 these objects without causing naming conflicts.
4832 * systemctl list-dependencies gained the new switches
4833 --plain, --reverse, --after and --before.
4835 * systemd-inhibit now shows the process name of processes that
4836 have taken an inhibitor lock.
4838 * nss-myhostname will now also resolve "localhost"
4839 implicitly. This makes /etc/hosts an optional file and
4840 nicely handles that on IPv6 ::1 maps to both "localhost" and
4843 * libsystemd-logind.so gained a new call
4844 sd_get_machine_names() to enumerate running containers and
4845 VMs (currently only supported by very new libvirt and
4846 nspawn). sd_login_monitor can now be used to watch
4847 VMs/containers coming and going.
4849 * .include is not allowed recursively anymore, and only in
4850 unit files. Usually it is better to use drop-in snippets in
4851 .d/*.conf anyway, as introduced with systemd 198.
4853 * systemd-analyze gained a new "critical-chain" command that
4854 determines the slowest chain of units run during system
4855 boot-up. It is very useful for tracking down where
4856 optimizing boot time is the most beneficial.
4858 * systemd will no longer allow manipulating service paths in
4859 the name=systemd:/system cgroup tree using ControlGroup= in
4860 units. (But is still fine with it in all other dirs.)
4862 * There's a new systemd-nspawn@.service service file that may
4863 be used to easily run nspawn containers as system
4864 services. With the container's root directory in
4865 /var/lib/container/foobar it is now sufficient to run
4866 "systemctl start systemd-nspawn@foobar.service" to boot it.
4868 * systemd-cgls gained a new parameter "--machine" to list only
4869 the processes within a certain container.
4871 * ConditionSecurity= now can check for "apparmor". We still
4872 are lacking checks for SMACK and IMA for this condition
4873 check though. Patches welcome!
4875 * A new configuration file /etc/systemd/sleep.conf has been
4876 added that may be used to configure which kernel operation
4877 systemd is supposed to execute when "suspend", "hibernate"
4878 or "hybrid-sleep" is requested. This makes the new kernel
4879 "freeze" state accessible to the user.
4881 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
4882 the passed argument if applicable.
4884 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
4885 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
4886 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
4887 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
4888 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
4889 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
4890 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
4895 * The output of 'systemctl list-jobs' got some polishing. The
4896 '--type=' argument may now be passed more than once. A new
4897 command 'systemctl list-sockets' has been added which shows
4898 a list of kernel sockets systemd is listening on with the
4899 socket units they belong to, plus the units these socket
4902 * The experimental libsystemd-bus library got substantial
4903 updates to work in conjunction with the (also experimental)
4904 kdbus kernel project. It works well enough to exchange
4905 messages with some sophistication. Note that kdbus is not
4906 ready yet, and the library is mostly an elaborate test case
4907 for now, and not installable.
4909 * systemd gained a new unit 'systemd-static-nodes.service'
4910 that generates static device nodes earlier during boot, and
4911 can run in conjunction with udev.
4913 * libsystemd-login gained a new call sd_pid_get_user_unit()
4914 to retrieve the user systemd unit a process is running
4915 in. This is useful for systems where systemd is used as
4918 * systemd-nspawn now places all containers in the new /machine
4919 top-level cgroup directory in the name=systemd
4920 hierarchy. libvirt will soon do the same, so that we get a
4921 uniform separation of /system, /user and /machine for system
4922 services, user processes and containers/virtual
4923 machines. This new cgroup hierarchy is also useful to stick
4924 stable names to specific container instances, which can be
4925 recognized later this way (this name may be controlled
4926 via systemd-nspawn's new -M switch). libsystemd-login also
4927 gained a new call sd_pid_get_machine_name() to retrieve the
4928 name of the container/VM a specific process belongs to.
4930 * bootchart can now store its data in the journal.
4932 * libsystemd-journal gained a new call
4933 sd_journal_add_conjunction() for AND expressions to the
4934 matching logic. This can be used to express more complex
4935 logical expressions.
4937 * journactl can now take multiple --unit= and --user-unit=
4940 * The cryptsetup logic now understands the "luks.key=" kernel
4941 command line switch for specifying a file to read the
4942 decryption key from. Also, if a configured key file is not
4943 found the tool will now automatically fall back to prompting
4946 * Python systemd.journal module was updated to wrap recently
4947 added functions from libsystemd-journal. The interface was
4948 changed to bring the low level interface in s.j._Reader
4949 closer to the C API, and the high level interface in
4950 s.j.Reader was updated to wrap and convert all data about
4953 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
4954 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
4955 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
4956 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
4957 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
4958 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
4962 * journalctl --update-catalog now understands a new --root=
4963 option to operate on catalogs found in a different root
4966 * During shutdown after systemd has terminated all running
4967 services a final killing loop kills all remaining left-over
4968 processes. We will now print the name of these processes
4969 when we send SIGKILL to them, since this usually indicates a
4972 * If /etc/crypttab refers to password files stored on
4973 configured mount points automatic dependencies will now be
4974 generated to ensure the specific mount is established first
4975 before the key file is attempted to be read.
4977 * 'systemctl status' will now show information about the
4978 network sockets a socket unit is listening on.
4980 * 'systemctl status' will also shown information about any
4981 drop-in configuration file for units. (Drop-In configuration
4982 files in this context are files such as
4983 /etc/systemd/systemd/foobar.service.d/*.conf)
4985 * systemd-cgtop now optionally shows summed up CPU times of
4986 cgroups. Press '%' while running cgtop to switch between
4987 percentage and absolute mode. This is useful to determine
4988 which cgroups use up the most CPU time over the entire
4989 runtime of the system. systemd-cgtop has also been updated
4990 to be 'pipeable' for processing with further shell tools.
4992 * 'hostnamectl set-hostname' will now allow setting of FQDN
4995 * The formatting and parsing of time span values has been
4996 changed. The parser now understands fractional expressions
4997 such as "5.5h". The formatter will now output fractional
4998 expressions for all time spans under 1min, i.e. "5.123456s"
4999 rather than "5s 123ms 456us". For time spans under 1s
5000 millisecond values are shown, for those under 1ms
5001 microsecond values are shown. This should greatly improve
5002 all time-related output of systemd.
5004 * libsystemd-login and libsystemd-journal gained new
5005 functions for querying the poll() events mask and poll()
5006 timeout value for integration into arbitrary event
5009 * localectl gained the ability to list available X11 keymaps
5010 (models, layouts, variants, options).
5012 * 'systemd-analyze dot' gained the ability to filter for
5013 specific units via shell-style globs, to create smaller,
5014 more useful graphs. I.e. it is now possible to create simple
5015 graphs of all the dependencies between only target units, or
5016 of all units that Avahi has dependencies with.
5018 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
5019 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
5020 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
5021 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
5022 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
5023 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
5024 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
5028 * The boot-time readahead implementation for rotating media
5029 will now read the read-ahead data in multiple passes which
5030 consist of all read requests made in equidistant time
5031 intervals. This means instead of strictly reading read-ahead
5032 data in its physical order on disk we now try to find a
5033 middle ground between physical and access time order.
5035 * /etc/os-release files gained a new BUILD_ID= field for usage
5036 on operating systems that provide continuous builds of OS
5039 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
5040 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
5041 William Douglas, Zbigniew Jędrzejewski-Szmek
5045 * systemd-python gained an API exposing libsystemd-daemon.
5047 * The SMACK setup logic gained support for uploading CIPSO
5050 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
5051 ReadOnlyDirectories= and InaccessibleDirectories= has
5052 changed. The private /tmp and /var/tmp directories are now
5053 shared by all processes of a service (which means
5054 ExecStartPre= may now leave data in /tmp that ExecStart= of
5055 the same service can still access). When a service is
5056 stopped its temporary directories are immediately deleted
5057 (normal clean-up with tmpfiles is still done in addition to
5060 * By default, systemd will now set a couple of sysctl
5061 variables in the kernel: the safe sysrq options are turned
5062 on, IP route verification is turned on, and source routing
5063 disabled. The recently added hardlink and softlink
5064 protection of the kernel is turned on. These settings should
5065 be reasonably safe, and good defaults for all new systems.
5067 * The predictable network naming logic may now be turned off
5068 with a new kernel command line switch: net.ifnames=0.
5070 * A new libsystemd-bus module has been added that implements a
5071 pretty complete D-Bus client library. For details see:
5073 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
5075 * journald will now explicitly flush the journal files to disk
5076 at the latest 5min after each write. The file will then also
5077 be marked offline until the next write. This should increase
5078 reliability in case of a crash. The synchronization delay
5079 can be configured via SyncIntervalSec= in journald.conf.
5081 * There's a new remote-fs-setup.target unit that can be used
5082 to pull in specific services when at least one remote file
5083 system is to be mounted.
5085 * There are new targets timers.target and paths.target as
5086 canonical targets to pull user timer and path units in
5087 from. This complements sockets.target with a similar
5088 purpose for socket units.
5090 * libudev gained a new call udev_device_set_attribute_value()
5091 to set sysfs attributes of a device.
5093 * The udev daemon now sets the default number of worker
5094 processes executed in parallel based on the number of available
5095 CPUs instead of the amount of available RAM. This is supposed
5096 to provide a more reliable default and limit a too aggressive
5097 parallelism for setups with 1000s of devices connected.
5099 Contributions from: Auke Kok, Colin Walters, Cristian
5100 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
5101 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
5102 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
5103 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
5104 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
5105 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
5106 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
5107 Zbigniew Jędrzejewski-Szmek
5111 * Configuration of unit files may now be extended via drop-in
5112 files without having to edit/override the unit files
5113 themselves. More specifically, if the administrator wants to
5114 change one value for a service file foobar.service he can
5115 now do so by dropping in a configuration snippet into
5116 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
5117 will load all these snippets and apply them on top of the
5118 main unit configuration file, possibly extending or
5119 overriding its settings. Using these drop-in snippets is
5120 generally nicer than the two earlier options for changing
5121 unit files locally: copying the files from
5122 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
5123 them there; or creating a new file in /etc/systemd/system/
5124 that incorporates the original one via ".include". Drop-in
5125 snippets into these .d/ directories can be placed in any
5126 directory systemd looks for units in, and the usual
5127 overriding semantics between /usr/lib, /etc and /run apply
5130 * Most unit file settings which take lists of items can now be
5131 reset by assigning the empty string to them. For example,
5132 normally, settings such as Environment=FOO=BAR append a new
5133 environment variable assignment to the environment block,
5134 each time they are used. By assigning Environment= the empty
5135 string the environment block can be reset to empty. This is
5136 particularly useful with the .d/*.conf drop-in snippets
5137 mentioned above, since this adds the ability to reset list
5138 settings from vendor unit files via these drop-ins.
5140 * systemctl gained a new "list-dependencies" command for
5141 listing the dependencies of a unit recursively.
5143 * Inhibitors are now honored and listed by "systemctl
5144 suspend", "systemctl poweroff" (and similar) too, not only
5145 GNOME. These commands will also list active sessions by
5148 * Resource limits (as exposed by the various control group
5149 controllers) can now be controlled dynamically at runtime
5150 for all units. More specifically, you can now use a command
5151 like "systemctl set-cgroup-attr foobar.service cpu.shares
5152 2000" to alter the CPU shares a specific service gets. These
5153 settings are stored persistently on disk, and thus allow the
5154 administrator to easily adjust the resource usage of
5155 services with a few simple commands. This dynamic resource
5156 management logic is also available to other programs via the
5157 bus. Almost any kernel cgroup attribute and controller is
5160 * systemd-vconsole-setup will now copy all font settings to
5161 all allocated VTs, where it previously applied them only to
5164 * libsystemd-login gained the new sd_session_get_tty() API
5167 * This release drops support for a few legacy or
5168 distribution-specific LSB facility names when parsing init
5169 scripts: $x-display-manager, $mail-transfer-agent,
5170 $mail-transport-agent, $mail-transfer-agent, $smtp,
5171 $null. Also, the mail-transfer-agent.target unit backing
5172 this has been removed. Distributions which want to retain
5173 compatibility with this should carry the burden for
5174 supporting this themselves and patch support for these back
5175 in, if they really need to. Also, the facilities $syslog and
5176 $local_fs are now ignored, since systemd does not support
5177 early-boot LSB init scripts anymore, and these facilities
5178 are implied anyway for normal services. syslog.target has
5181 * There are new bus calls on PID1's Manager object for
5182 cancelling jobs, and removing snapshot units. Previously,
5183 both calls were only available on the Job and Snapshot
5186 * systemd-journal-gatewayd gained SSL support.
5188 * The various "environment" files, such as /etc/locale.conf
5189 now support continuation lines with a backslash ("\") as
5190 last character in the line, similarly in style (but different)
5191 to how this is supported in shells.
5193 * For normal user processes the _SYSTEMD_USER_UNIT= field is
5194 now implicitly appended to every log entry logged. systemctl
5195 has been updated to filter by this field when operating on a
5196 user systemd instance.
5198 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
5199 CAP_AUDIT_CONTROL capabilities to the capabilities set for
5200 the container. This makes it easier to boot unmodified
5201 Fedora systems in a container, which however still requires
5202 audit=0 to be passed on the kernel command line. Auditing in
5203 kernel and userspace is unfortunately still too broken in
5204 context of containers, hence we recommend compiling it out
5205 of the kernel or using audit=0. Hopefully this will be fixed
5206 one day for good in the kernel.
5208 * nspawn gained the new --bind= and --bind-ro= parameters to
5209 bind mount specific directories from the host into the
5212 * nspawn will now mount its own devpts file system instance
5213 into the container, in order not to leak pty devices from
5214 the host into the container.
5216 * systemd will now read the firmware boot time performance
5217 information from the EFI variables, if the used boot loader
5218 supports this, and takes it into account for boot performance
5219 analysis via "systemd-analyze". This is currently supported
5220 only in conjunction with Gummiboot, but could be supported
5221 by other boot loaders too. For details see:
5223 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
5225 * A new generator has been added that automatically mounts the
5226 EFI System Partition (ESP) to /boot, if that directory
5227 exists, is empty, and no other file system has been
5228 configured to be mounted there.
5230 * logind will now send out PrepareForSleep(false) out
5231 unconditionally, after coming back from suspend. This may be
5232 used by applications as asynchronous notification for
5233 system resume events.
5235 * "systemctl unlock-sessions" has been added, that allows
5236 unlocking the screens of all user sessions at once, similar
5237 to how "systemctl lock-sessions" already locked all users
5238 sessions. This is backed by a new D-Bus call UnlockSessions().
5240 * "loginctl seat-status" will now show the master device of a
5241 seat. (i.e. the device of a seat that needs to be around for
5242 the seat to be considered available, usually the graphics
5245 * tmpfiles gained a new "X" line type, that allows
5246 configuration of files and directories (with wildcards) that
5247 shall be excluded from automatic cleanup ("aging").
5249 * udev default rules set the device node permissions now only
5250 at "add" events, and do not change them any longer with a
5251 later "change" event.
5253 * The log messages for lid events and power/sleep keypresses
5254 now carry a message ID.
5256 * We now have a substantially larger unit test suite, but this
5257 continues to be work in progress.
5259 * udevadm hwdb gained a new --root= parameter to change the
5260 root directory to operate relative to.
5262 * logind will now issue a background sync() request to the kernel
5263 early at shutdown, so that dirty buffers are flushed to disk early
5264 instead of at the last moment, in order to optimize shutdown
5267 * A new bootctl tool has been added that is an interface for
5268 certain boot loader operations. This is currently a preview
5269 and is likely to be extended into a small mechanism daemon
5270 like timedated, localed, hostnamed, and can be used by
5271 graphical UIs to enumerate available boot options, and
5272 request boot into firmware operations.
5274 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
5275 the rest of the package. It also has been updated to work
5276 correctly in initrds.
5278 * Policykit previously has been runtime optional, and is now
5279 also compile time optional via a configure switch.
5281 * systemd-analyze has been reimplemented in C. Also "systemctl
5282 dot" has moved into systemd-analyze.
5284 * "systemctl status" with no further parameters will now print
5285 the status of all active or failed units.
5287 * Operations such as "systemctl start" can now be executed
5288 with a new mode "--irreversible" which may be used to queue
5289 operations that cannot accidentally be reversed by a later
5290 job queuing. This is by default used to make shutdown
5291 requests more robust.
5293 * The Python API of systemd now gained a new module for
5294 reading journal files.
5296 * A new tool kernel-install has been added that can install
5297 kernel images according to the Boot Loader Specification:
5299 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
5301 * Boot time console output has been improved to provide
5302 animated boot time output for hanging jobs.
5304 * A new tool systemd-activate has been added which can be used
5305 to test socket activation with, directly from the command
5306 line. This should make it much easier to test and debug
5307 socket activation in daemons.
5309 * journalctl gained a new "--reverse" (or -r) option to show
5310 journal output in reverse order (i.e. newest line first).
5312 * journalctl gained a new "--pager-end" (or -e) option to jump
5313 to immediately jump to the end of the journal in the
5314 pager. This is only supported in conjunction with "less".
5316 * journalctl gained a new "--user-unit=" option, that works
5317 similarly to "--unit=" but filters for user units rather than
5320 * A number of unit files to ease adoption of systemd in
5321 initrds has been added. This moves some minimal logic from
5322 the various initrd implementations into systemd proper.
5324 * The journal files are now owned by a new group
5325 "systemd-journal", which exists specifically to allow access
5326 to the journal, and nothing else. Previously, we used the
5327 "adm" group for that, which however possibly covers more
5328 than just journal/log file access. This new group is now
5329 already used by systemd-journal-gatewayd to ensure this
5330 daemon gets access to the journal files and as little else
5331 as possible. Note that "make install" will also set FS ACLs
5332 up for /var/log/journal to give "adm" and "wheel" read
5333 access to it, in addition to "systemd-journal" which owns
5334 the journal files. We recommend that packaging scripts also
5335 add read access to "adm" + "wheel" to /var/log/journal, and
5336 all existing/future journal files. To normal users and
5337 administrators little changes, however packagers need to
5338 ensure to create the "systemd-journal" system group at
5339 package installation time.
5341 * The systemd-journal-gatewayd now runs as unprivileged user
5342 systemd-journal-gateway:systemd-journal-gateway. Packaging
5343 scripts need to create these system user/group at
5346 * timedated now exposes a new boolean property CanNTP that
5347 indicates whether a local NTP service is available or not.
5349 * systemd-detect-virt will now also detect xen PVs
5351 * The pstore file system is now mounted by default, if it is
5354 * In addition to the SELinux and IMA policies we will now also
5355 load SMACK policies at early boot.
5357 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
5358 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
5359 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
5360 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
5361 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
5362 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
5363 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
5364 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
5365 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
5366 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
5367 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
5368 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
5369 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
5370 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
5374 * Timer units now support calendar time events in addition to
5375 monotonic time events. That means you can now trigger a unit
5376 based on a calendar time specification such as "Thu,Fri
5377 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
5378 or fifth day of any month of the year 2013, given that it is
5379 a thursday or friday. This brings timer event support
5380 considerably closer to cron's capabilities. For details on
5381 the supported calendar time specification language see
5384 * udev now supports a number of different naming policies for
5385 network interfaces for predictable names, and a combination
5386 of these policies is now the default. Please see this wiki
5387 document for details:
5389 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
5391 * Auke Kok's bootchart implementation has been added to the
5392 systemd tree. It is an optional component that can graph the
5393 boot in quite some detail. It is one of the best bootchart
5394 implementations around and minimal in its code and
5397 * nss-myhostname has been integrated into the systemd source
5398 tree. nss-myhostname guarantees that the local hostname
5399 always stays resolvable via NSS. It has been a weak
5400 requirement of systemd-hostnamed since a long time, and
5401 since its code is actually trivial we decided to just
5402 include it in systemd's source tree. It can be turned off
5403 with a configure switch.
5405 * The read-ahead logic is now capable of properly detecting
5406 whether a btrfs file system is on SSD or rotating media, in
5407 order to optimize the read-ahead scheme. Previously, it was
5408 only capable of detecting this on traditional file systems
5411 * In udev, additional device properties are now read from the
5412 IAB in addition to the OUI database. Also, Bluetooth company
5413 identities are attached to the devices as well.
5415 * In service files %U may be used as specifier that is
5416 replaced by the configured user name of the service.
5418 * nspawn may now be invoked without a controlling TTY. This
5419 makes it suitable for invocation as its own service. This
5420 may be used to set up a simple containerized server system
5421 using only core OS tools.
5423 * systemd and nspawn can now accept socket file descriptors
5424 when they are started for socket activation. This enables
5425 implementation of socket activated nspawn
5426 containers. i.e. think about autospawning an entire OS image
5427 when the first SSH or HTTP connection is received. We expect
5428 that similar functionality will also be added to libvirt-lxc
5431 * journalctl will now suppress ANSI color codes when
5432 presenting log data.
5434 * systemctl will no longer show control group information for
5435 a unit if the control group is empty anyway.
5437 * logind can now automatically suspend/hibernate/shutdown the
5440 * /etc/machine-info and hostnamed now also expose the chassis
5441 type of the system. This can be used to determine whether
5442 the local system is a laptop, desktop, handset or
5443 tablet. This information may either be configured by the
5444 user/vendor or is automatically determined from ACPI and DMI
5445 information if possible.
5447 * A number of PolicyKit actions are now bound together with
5448 "imply" rules. This should simplify creating UIs because
5449 many actions will now authenticate similar ones as well.
5451 * Unit files learnt a new condition ConditionACPower= which
5452 may be used to conditionalize a unit depending on whether an
5453 AC power source is connected or not, of whether the system
5454 is running on battery power.
5456 * systemctl gained a new "is-failed" verb that may be used in
5457 shell scripts and suchlike to check whether a specific unit
5458 is in the "failed" state.
5460 * The EnvironmentFile= setting in unit files now supports file
5461 globbing, and can hence be used to easily read a number of
5462 environment files at once.
5464 * systemd will no longer detect and recognize specific
5465 distributions. All distribution-specific #ifdeffery has been
5466 removed, systemd is now fully generic and
5467 distribution-agnostic. Effectively, not too much is lost as
5468 a lot of the code is still accessible via explicit configure
5469 switches. However, support for some distribution specific
5470 legacy configuration file formats has been dropped. We
5471 recommend distributions to simply adopt the configuration
5472 files everybody else uses now and convert the old
5473 configuration from packaging scripts. Most distributions
5474 already did that. If that's not possible or desirable,
5475 distributions are welcome to forward port the specific
5476 pieces of code locally from the git history.
5478 * When logging a message about a unit systemd will now always
5479 log the unit name in the message meta data.
5481 * localectl will now also discover system locale data that is
5482 not stored in locale archives, but directly unpacked.
5484 * logind will no longer unconditionally use framebuffer
5485 devices as seat masters, i.e. as devices that are required
5486 to be existing before a seat is considered preset. Instead,
5487 it will now look for all devices that are tagged as
5488 "seat-master" in udev. By default, framebuffer devices will
5489 be marked as such, but depending on local systems, other
5490 devices might be marked as well. This may be used to
5491 integrate graphics cards using closed source drivers (such
5492 as NVidia ones) more nicely into logind. Note however, that
5493 we recommend using the open source NVidia drivers instead,
5494 and no udev rules for the closed-source drivers will be
5495 shipped from us upstream.
5497 Contributions from: Adam Williamson, Alessandro Crismani, Auke
5498 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
5499 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
5500 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
5501 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
5502 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
5503 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
5504 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
5505 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
5506 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
5507 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
5508 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
5513 * udev gained support for loading additional device properties
5514 from an indexed database that is keyed by vendor/product IDs
5515 and similar device identifiers. For the beginning this
5516 "hwdb" is populated with data from the well-known PCI and
5517 USB database, but also includes PNP, ACPI and OID data. In
5518 the longer run this indexed database shall grow into
5519 becoming the one central database for non-essential
5520 userspace device metadata. Previously, data from the PCI/USB
5521 database was only attached to select devices, since the
5522 lookup was a relatively expensive operation due to O(n) time
5523 complexity (with n being the number of entries in the
5524 database). Since this is now O(1), we decided to add in this
5525 data for all devices where this is available, by
5526 default. Note that the indexed database needs to be rebuilt
5527 when new data files are installed. To achieve this you need
5528 to update your packaging scripts to invoke "udevadm hwdb
5529 --update" after installation of hwdb data files. For
5530 RPM-based distributions we introduced the new
5531 %udev_hwdb_update macro for this purpose.
5533 * The Journal gained support for the "Message Catalog", an
5534 indexed database to link up additional information with
5535 journal entries. For further details please check:
5537 http://www.freedesktop.org/wiki/Software/systemd/catalog
5539 The indexed message catalog database also needs to be
5540 rebuilt after installation of message catalog files. Use
5541 "journalctl --update-catalog" for this. For RPM-based
5542 distributions we introduced the %journal_catalog_update
5543 macro for this purpose.
5545 * The Python Journal bindings gained support for the standard
5546 Python logging framework.
5548 * The Journal API gained new functions for checking whether
5549 the underlying file system of a journal file is capable of
5550 properly reporting file change notifications, or whether
5551 applications that want to reflect journal changes "live"
5552 need to recheck journal files continuously in appropriate
5555 * It is now possible to set the "age" field for tmpfiles
5556 entries to 0, indicating that files matching this entry
5557 shall always be removed when the directories are cleaned up.
5559 * coredumpctl gained a new "gdb" verb which invokes gdb
5560 right-away on the selected coredump.
5562 * There's now support for "hybrid sleep" on kernels that
5563 support this, in addition to "suspend" and "hibernate". Use
5564 "systemctl hybrid-sleep" to make use of this.
5566 * logind's HandleSuspendKey= setting (and related settings)
5567 now gained support for a new "lock" setting to simply
5568 request the screen lock on all local sessions, instead of
5569 actually executing a suspend or hibernation.
5571 * systemd will now mount the EFI variables file system by
5574 * Socket units now gained support for configuration of the
5575 SMACK security label.
5577 * timedatectl will now output the time of the last and next
5578 daylight saving change.
5580 * We dropped support for various legacy and distro-specific
5581 concepts, such as insserv, early-boot SysV services
5582 (i.e. those for non-standard runlevels such as 'b' or 'S')
5583 or ArchLinux /etc/rc.conf support. We recommend the
5584 distributions who still need support this to either continue
5585 to maintain the necessary patches downstream, or find a
5586 different solution. (Talk to us if you have questions!)
5588 * Various systemd components will now bypass PolicyKit checks
5589 for root and otherwise handle properly if PolicyKit is not
5590 found to be around. This should fix most issues for
5591 PolicyKit-less systems. Quite frankly this should have been
5592 this way since day one. It is absolutely our intention to
5593 make systemd work fine on PolicyKit-less systems, and we
5594 consider it a bug if something does not work as it should if
5595 PolicyKit is not around.
5597 * For embedded systems it is now possible to build udev and
5598 systemd without blkid and/or kmod support.
5600 * "systemctl switch-root" is now capable of switching root
5601 more than once. I.e. in addition to transitions from the
5602 initrd to the host OS it is now possible to transition to
5603 further OS images from the host. This is useful to implement
5604 offline updating tools.
5606 * Various other additions have been made to the RPM macros
5607 shipped with systemd. Use %udev_rules_update() after
5608 installing new udev rules files. %_udevhwdbdir,
5609 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
5610 %_sysctldir are now available which resolve to the right
5611 directories for packages to place various data files in.
5613 * journalctl gained the new --full switch (in addition to
5614 --all, to disable ellipsation for long messages.
5616 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
5617 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
5618 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
5619 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
5620 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
5621 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
5622 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
5623 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
5624 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
5628 * journalctl gained new --since= and --until= switches to
5629 filter by time. It also now supports nice filtering for
5630 units via --unit=/-u.
5632 * Type=oneshot services may use ExecReload= and do the
5635 * The journal daemon now supports time-based rotation and
5636 vacuuming, in addition to the usual disk-space based
5639 * The journal will now index the available field values for
5640 each field name. This enables clients to show pretty drop
5641 downs of available match values when filtering. The bash
5642 completion of journalctl has been updated
5643 accordingly. journalctl gained a new switch -F to list all
5644 values a certain field takes in the journal database.
5646 * More service events are now written as structured messages
5647 to the journal, and made recognizable via message IDs.
5649 * The timedated, localed and hostnamed mini-services which
5650 previously only provided support for changing time, locale
5651 and hostname settings from graphical DEs such as GNOME now
5652 also have a minimal (but very useful) text-based client
5653 utility each. This is probably the nicest way to changing
5654 these settings from the command line now, especially since
5655 it lists available options and is fully integrated with bash
5658 * There's now a new tool "systemd-coredumpctl" to list and
5659 extract coredumps from the journal.
5661 * We now install a README each in /var/log/ and
5662 /etc/rc.d/init.d explaining where the system logs and init
5663 scripts went. This hopefully should help folks who go to
5664 that dirs and look into the otherwise now empty void and
5665 scratch their heads.
5667 * When user-services are invoked (by systemd --user) the
5668 $MANAGERPID env var is set to the PID of systemd.
5670 * SIGRTMIN+24 when sent to a --user instance will now result
5671 in immediate termination of systemd.
5673 * gatewayd received numerous feature additions such as a
5674 "follow" mode, for live syncing and filtering.
5676 * browse.html now allows filtering and showing detailed
5677 information on specific entries. Keyboard navigation and
5678 mouse screen support has been added.
5680 * gatewayd/journalctl now supports HTML5/JSON
5681 Server-Sent-Events as output.
5683 * The SysV init script compatibility logic will now
5684 heuristically determine whether a script supports the
5685 "reload" verb, and only then make this available as
5688 * "systemctl status --follow" has been removed, use "journalctl
5691 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
5692 have been removed since they are hardly useful to be
5695 * And I'd like to take the opportunity to specifically mention
5696 Zbigniew for his great contributions. Zbigniew, you rock!
5698 Contributions from: Andrew Eikum, Christian Hesse, Colin
5699 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
5700 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
5701 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
5702 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
5703 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
5704 Jędrzejewski-Szmek, Сковорода Никита Андреевич
5708 * If /etc/vconsole.conf is non-existent or empty we will no
5709 longer load any console font or key map at boot by
5710 default. Instead the kernel defaults will be left
5711 intact. This is definitely the right thing to do, as no
5712 configuration should mean no configuration, and hard-coding
5713 font names that are different on all archs is probably a bad
5714 idea. Also, the kernel default key map and font should be
5715 good enough for most cases anyway, and mostly identical to
5716 the userspace fonts/key maps we previously overloaded them
5717 with. If distributions want to continue to default to a
5718 non-kernel font or key map they should ship a default
5719 /etc/vconsole.conf with the appropriate contents.
5721 Contributions from: Colin Walters, Daniel J Walsh, Dave
5722 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
5723 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
5727 * journalctl gained a new --cursor= switch to show entries
5728 starting from the specified location in the journal.
5730 * We now enforce a size limit on journal entry fields exported
5731 with "-o json" in journalctl. Fields larger than 4K will be
5732 assigned null. This can be turned off with --all.
5734 * An (optional) journal gateway daemon is now available as
5735 "systemd-journal-gatewayd.service". This service provides
5736 access to the journal via HTTP and JSON. This functionality
5737 will be used to implement live log synchronization in both
5738 pull and push modes, but has various other users too, such
5739 as easy log access for debugging of embedded devices. Right
5740 now it is already useful to retrieve the journal via HTTP:
5742 # systemctl start systemd-journal-gatewayd.service
5743 # wget http://localhost:19531/entries
5745 This will download the journal contents in a
5746 /var/log/messages compatible format. The same as JSON:
5748 # curl -H"Accept: application/json" http://localhost:19531/entries
5750 This service is also accessible via a web browser where a
5751 single static HTML5 app is served that uses the JSON logic
5752 to enable the user to do some basic browsing of the
5753 journal. This will be extended later on. Here's an example
5754 screenshot of this app in its current state:
5756 http://0pointer.de/public/journal-gatewayd
5758 Contributions from: Kay Sievers, Lennart Poettering, Robert
5759 Milasan, Tom Gundersen
5763 * The bash completion logic is now available for journalctl
5766 * We do not mount the "cpuset" controller anymore together with
5767 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
5768 started if no parameters are assigned to it. "cpuset" hence
5769 broke code that assumed it could create "cpu" groups and
5772 * journalctl -f will now subscribe to terminal size changes,
5773 and line break accordingly.
5775 Contributions from: Dave Reisner, Kay Sievers, Lennart
5776 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
5780 * nspawn will now create a symlink /etc/localtime in the
5781 container environment, copying the host's timezone
5782 setting. Previously this has been done via a bind mount, but
5783 since symlinks cannot be bind mounted this has now been
5784 changed to create/update the appropriate symlink.
5786 * journalctl -n's line number argument is now optional, and
5787 will default to 10 if omitted.
5789 * journald will now log the maximum size the journal files may
5790 take up on disk. This is particularly useful if the default
5791 built-in logic of determining this parameter from the file
5792 system size is used. Use "systemctl status
5793 systemd-journald.service" to see this information.
5795 * The multi-seat X wrapper tool has been stripped down. As X
5796 is now capable of enumerating graphics devices via udev in a
5797 seat-aware way the wrapper is not strictly necessary
5798 anymore. A stripped down temporary stop-gap is still shipped
5799 until the upstream display managers have been updated to
5800 fully support the new X logic. Expect this wrapper to be
5801 removed entirely in one of the next releases.
5803 * HandleSleepKey= in logind.conf has been split up into
5804 HandleSuspendKey= and HandleHibernateKey=. The old setting
5805 is not available anymore. X11 and the kernel are
5806 distinguishing between these keys and we should too. This
5807 also means the inhibition lock for these keys has been split
5810 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
5811 Poettering, Lukas Nykryn, Václav Pavlín
5815 * Whenever a unit changes state we will now log this to the
5816 journal and show along the unit's own log output in
5819 * ConditionPathIsMountPoint= can now properly detect bind
5820 mount points too. (Previously, a bind mount of one file
5821 system to another place in the same file system could not be
5822 detected as mount, since they shared struct stat's st_dev
5825 * We will now mount the cgroup controllers cpu, cpuacct,
5826 cpuset and the controllers net_cls, net_prio together by
5829 * nspawn containers will now have a virtualized boot
5830 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
5831 over with a randomized ID at container initialization). This
5832 has the effect of making "journalctl -b" do the right thing
5835 * The JSON output journal serialization has been updated not
5836 to generate "endless" list objects anymore, but rather one
5837 JSON object per line. This is more in line how most JSON
5838 parsers expect JSON objects. The new output mode
5839 "json-pretty" has been added to provide similar output, but
5840 neatly aligned for readability by humans.
5842 * We dropped all explicit sync() invocations in the shutdown
5843 code. The kernel does this implicitly anyway in the kernel
5844 reboot() syscall. halt(8)'s -n option is now a compatibility
5847 * We now support virtualized reboot() in containers, as
5848 supported by newer kernels. We will fall back to exit() if
5849 CAP_SYS_REBOOT is not available to the container. Also,
5850 nspawn makes use of this now and will actually reboot the
5851 container if the containerized OS asks for that.
5853 * journalctl will only show local log output by default
5854 now. Use --merge (-m) to show remote log output, too.
5856 * libsystemd-journal gained the new sd_journal_get_usage()
5857 call to determine the current disk usage of all journal
5858 files. This is exposed in the new "journalctl --disk-usage"
5861 * journald gained a new configuration setting SplitMode= in
5862 journald.conf which may be used to control how user journals
5863 are split off. See journald.conf(5) for details.
5865 * A new condition type ConditionFileNotEmpty= has been added.
5867 * tmpfiles' "w" lines now support file globbing, to write
5868 multiple files at once.
5870 * We added Python bindings for the journal submission
5871 APIs. More Python APIs for a number of selected APIs will
5872 likely follow. Note that we intend to add native bindings
5873 only for the Python language, as we consider it common
5874 enough to deserve bindings shipped within systemd. There are
5875 various projects outside of systemd that provide bindings
5876 for languages such as PHP or Lua.
5878 * Many conditions will now resolve specifiers such as %i. In
5879 addition, PathChanged= and related directives of .path units
5880 now support specifiers as well.
5882 * There's now a new RPM macro definition for the system preset
5885 * journald will now warn if it ca not forward a message to the
5886 syslog daemon because its socket is full.
5888 * timedated will no longer write or process /etc/timezone,
5889 except on Debian. As we do not support late mounted /usr
5890 anymore /etc/localtime always being a symlink is now safe,
5891 and hence the information in /etc/timezone is not necessary
5894 * logind will now always reserve one VT for a text getty (VT6
5895 by default). Previously if more than 6 X sessions where
5896 started they took up all the VTs with auto-spawned gettys,
5897 so that no text gettys were available anymore.
5899 * udev will now automatically inform the btrfs kernel logic
5900 about btrfs RAID components showing up. This should make
5901 simple hotplug based btrfs RAID assembly work.
5903 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
5904 (but not for its children which will stay at the kernel
5905 default). This should allow setups with a lot more listening
5908 * systemd will now always pass the configured timezone to the
5909 kernel at boot. timedated will do the same when the timezone
5912 * logind's inhibition logic has been updated. By default,
5913 logind will now handle the lid switch, the power and sleep
5914 keys all the time, even in graphical sessions. If DEs want
5915 to handle these events on their own they should take the new
5916 handle-power-key, handle-sleep-key and handle-lid-switch
5917 inhibitors during their runtime. A simple way to achieve
5918 that is to invoke the DE wrapped in an invocation of:
5920 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
5922 * Access to unit operations is now checked via SELinux taking
5923 the unit file label and client process label into account.
5925 * systemd will now notify the administrator in the journal
5926 when he over-mounts a non-empty directory.
5928 * There are new specifiers that are resolved in unit files,
5929 for the host name (%H), the machine ID (%m) and the boot ID
5932 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
5933 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
5934 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
5935 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
5936 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
5937 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
5938 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
5942 * Support for reading structured kernel messages from
5943 /dev/kmsg has now been added and is enabled by default.
5945 * Support for reading kernel messages from /proc/kmsg has now
5946 been removed. If you want kernel messages in the journal
5947 make sure to run a recent kernel (>= 3.5) that supports
5948 reading structured messages from /dev/kmsg (see
5949 above). /proc/kmsg is now exclusive property of classic
5950 syslog daemons again.
5952 * The libudev API gained the new
5953 udev_device_new_from_device_id() call.
5955 * The logic for file system namespace (ReadOnlyDirectory=,
5956 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
5957 require pivot_root() anymore. This means fewer temporary
5958 directories are created below /tmp for this feature.
5960 * nspawn containers will now see and receive all submounts
5961 made on the host OS below the root file system of the
5964 * Forward Secure Sealing is now supported for Journal files,
5965 which provide cryptographical sealing of journal files so
5966 that attackers cannot alter log history anymore without this
5967 being detectable. Lennart will soon post a blog story about
5968 this explaining it in more detail.
5970 * There are two new service settings RestartPreventExitStatus=
5971 and SuccessExitStatus= which allow configuration of exit
5972 status (exit code or signal) which will be excepted from the
5973 restart logic, resp. consider successful.
5975 * journalctl gained the new --verify switch that can be used
5976 to check the integrity of the structure of journal files and
5977 (if Forward Secure Sealing is enabled) the contents of
5980 * nspawn containers will now be run with /dev/stdin, /dev/fd/
5981 and similar symlinks pre-created. This makes running shells
5982 as container init process a lot more fun.
5984 * The fstab support can now handle PARTUUID= and PARTLABEL=
5987 * A new ConditionHost= condition has been added to match
5988 against the hostname (with globs) and machine ID. This is
5989 useful for clusters where a single OS image is used to
5990 provision a large number of hosts which shall run slightly
5991 different sets of services.
5993 * Services which hit the restart limit will now be placed in a
5996 Contributions from: Bertram Poettering, Dave Reisner, Huang
5997 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
5998 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
6002 * When running in --user mode systemd will now become a
6003 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
6004 tree a lot more organized.
6006 * A new PartOf= unit dependency type has been introduced that
6007 may be used to group services in a natural way.
6009 * "systemctl enable" may now be used to enable instances of
6012 * journalctl now prints error log levels in red, and
6013 warning/notice log levels in bright white. It also supports
6014 filtering by log level now.
6016 * cgtop gained a new -n switch (similar to top), to configure
6017 the maximum number of iterations to run for. It also gained
6018 -b, to run in batch mode (accepting no input).
6020 * The suffix ".service" may now be omitted on most systemctl
6021 command lines involving service unit names.
6023 * There's a new bus call in logind to lock all sessions, as
6024 well as a loginctl verb for it "lock-sessions".
6026 * libsystemd-logind.so gained a new call sd_journal_perror()
6027 that works similar to libc perror() but logs to the journal
6028 and encodes structured information about the error number.
6030 * /etc/crypttab entries now understand the new keyfile-size=
6033 * shutdown(8) now can send a (configurable) wall message when
6034 a shutdown is cancelled.
6036 * The mount propagation mode for the root file system will now
6037 default to "shared", which is useful to make containers work
6038 nicely out-of-the-box so that they receive new mounts from
6039 the host. This can be undone locally by running "mount
6040 --make-rprivate /" if needed.
6042 * The prefdm.service file has been removed. Distributions
6043 should maintain this unit downstream if they intend to keep
6044 it around. However, we recommend writing normal unit files
6045 for display managers instead.
6047 * Since systemd is a crucial part of the OS we will now
6048 default to a number of compiler switches that improve
6049 security (hardening) such as read-only relocations, stack
6050 protection, and suchlike.
6052 * The TimeoutSec= setting for services is now split into
6053 TimeoutStartSec= and TimeoutStopSec= to allow configuration
6054 of individual time outs for the start and the stop phase of
6057 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
6058 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
6059 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
6060 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
6061 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
6062 Gundersen, Zbigniew Jędrzejewski-Szmek
6066 * The journal and id128 C APIs are now fully documented as man
6069 * Extra safety checks have been added when transitioning from
6070 the initial RAM disk to the main system to avoid accidental
6073 * /etc/crypttab entries now understand the new keyfile-offset=
6076 * systemctl -t can now be used to filter by unit load state.
6078 * The journal C API gained the new sd_journal_wait() call to
6079 make writing synchronous journal clients easier.
6081 * journalctl gained the new -D switch to show journals from a
6084 * journalctl now displays a special marker between log
6085 messages of two different boots.
6087 * The journal is now explicitly flushed to /var via a service
6088 systemd-journal-flush.service, rather than implicitly simply
6089 by seeing /var/log/journal to be writable.
6091 * journalctl (and the journal C APIs) can now match for much
6092 more complex expressions, with alternatives and
6095 * When transitioning from the initial RAM disk to the main
6096 system we will now kill all processes in a killing spree to
6097 ensure no processes stay around by accident.
6099 * Three new specifiers may be used in unit files: %u, %h, %s
6100 resolve to the user name, user home directory resp. user
6101 shell. This is useful for running systemd user instances.
6103 * We now automatically rotate journal files if their data
6104 object hash table gets a fill level > 75%. We also size the
6105 hash table based on the configured maximum file size. This
6106 together should lower hash collisions drastically and thus
6107 speed things up a bit.
6109 * journalctl gained the new "--header" switch to introspect
6110 header data of journal files.
6112 * A new setting SystemCallFilters= has been added to services
6113 which may be used to apply blacklists or whitelists to
6114 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
6116 * nspawn gained a new --link-journal= switch (and quicker: -j)
6117 to link the container journal with the host. This makes it
6118 very easy to centralize log viewing on the host for all
6119 guests while still keeping the journal files separated.
6121 * Many bugfixes and optimizations
6123 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
6124 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
6125 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
6130 * Several tools now understand kernel command line arguments,
6131 which are only read when run in an initial RAM disk. They
6132 usually follow closely their normal counterparts, but are
6135 * There's a new tool to analyze the readahead files that are
6136 automatically generated at boot. Use:
6138 /usr/lib/systemd/systemd-readahead analyze /.readahead
6140 * We now provide an early debug shell on tty9 if this enabled. Use:
6142 systemctl enable debug-shell.service
6144 * All plymouth related units have been moved into the Plymouth
6145 package. Please make sure to upgrade your Plymouth version
6148 * systemd-tmpfiles now supports getting passed the basename of
6149 a configuration file only, in which case it will look for it
6150 in all appropriate directories automatically.
6152 * udevadm info now takes a /dev or /sys path as argument, and
6153 does the right thing. Example:
6155 udevadm info /dev/sda
6156 udevadm info /sys/class/block/sda
6158 * systemctl now prints a warning if a unit is stopped but a
6159 unit that might trigger it continues to run. Example: a
6160 service is stopped but the socket that activates it is left
6163 * "systemctl status" will now mention if the log output was
6164 shortened due to rotation since a service has been started.
6166 * The journal API now exposes functions to determine the
6167 "cutoff" times due to rotation.
6169 * journald now understands SIGUSR1 and SIGUSR2 for triggering
6170 immediately flushing of runtime logs to /var if possible,
6171 resp. for triggering immediate rotation of the journal
6174 * It is now considered an error if a service is attempted to
6175 be stopped that is not loaded.
6177 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
6179 * systemd-analyze now supports Python 3
6181 * tmpfiles now supports cleaning up directories via aging
6182 where the first level dirs are always kept around but
6183 directories beneath it automatically aged. This is enabled
6184 by prefixing the age field with '~'.
6186 * Seat objects now expose CanGraphical, CanTTY properties
6187 which is required to deal with very fast bootups where the
6188 display manager might be running before the graphics drivers
6189 completed initialization.
6191 * Seat objects now expose a State property.
6193 * We now include RPM macros for service enabling/disabling
6194 based on the preset logic. We recommend RPM based
6195 distributions to make use of these macros if possible. This
6196 makes it simpler to reuse RPM spec files across
6199 * We now make sure that the collected systemd unit name is
6200 always valid when services log to the journal via
6203 * There's a new man page kernel-command-line(7) detailing all
6204 command line options we understand.
6206 * The fstab generator may now be disabled at boot by passing
6207 fstab=0 on the kernel command line.
6209 * A new kernel command line option modules-load= is now understood
6210 to load a specific kernel module statically, early at boot.
6212 * Unit names specified on the systemctl command line are now
6213 automatically escaped as needed. Also, if file system or
6214 device paths are specified they are automatically turned
6215 into the appropriate mount or device unit names. Example:
6217 systemctl status /home
6218 systemctl status /dev/sda
6220 * The SysVConsole= configuration option has been removed from
6221 system.conf parsing.
6223 * The SysV search path is no longer exported on the D-Bus
6226 * The Names= option has been removed from unit file parsing.
6228 * There's a new man page bootup(7) detailing the boot process.
6230 * Every unit and every generator we ship with systemd now
6231 comes with full documentation. The self-explanatory boot is
6234 * A couple of services gained "systemd-" prefixes in their
6235 name if they wrap systemd code, rather than only external
6236 code. Among them fsck@.service which is now
6237 systemd-fsck@.service.
6239 * The HaveWatchdog property has been removed from the D-Bus
6242 * systemd.confirm_spawn= on the kernel command line should now
6245 * There's a new man page crypttab(5) which details all options
6246 we actually understand.
6248 * systemd-nspawn gained a new --capability= switch to pass
6249 additional capabilities to the container.
6251 * timedated will now read known NTP implementation unit names
6252 from /usr/lib/systemd/ntp-units.d/*.list,
6253 systemd-timedated-ntp.target has been removed.
6255 * journalctl gained a new switch "-b" that lists log data of
6256 the current boot only.
6258 * The notify socket is in the abstract namespace again, in
6259 order to support daemons which chroot() at start-up.
6261 * There is a new Storage= configuration option for journald
6262 which allows configuration of where log data should go. This
6263 also provides a way to disable journal logging entirely, so
6264 that data collected is only forwarded to the console, the
6265 kernel log buffer or another syslog implementation.
6267 * Many bugfixes and optimizations
6269 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
6270 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
6271 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
6272 Shawn Landden, Tom Gundersen
6276 * "systemctl help <unit>" now shows the man page if one is
6279 * Several new man pages have been added.
6281 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
6282 MaxLevelConsole= can now be specified in
6283 journald.conf. These options allow reducing the amount of
6284 data stored on disk or forwarded by the log level.
6286 * TimerSlackNSec= can now be specified in system.conf for
6287 PID1. This allows system-wide power savings.
6289 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
6290 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
6295 * logind is now capable of (optionally) handling power and
6296 sleep keys as well as the lid switch.
6298 * journalctl now understands the syntax "journalctl
6299 /usr/bin/avahi-daemon" to get all log output of a specific
6302 * CapabilityBoundingSet= in system.conf now also influences
6303 the capability bound set of usermode helpers of the kernel.
6305 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
6306 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
6307 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
6308 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
6312 * Note that we skipped 139 releases here in order to set the
6313 new version to something that is greater than both udev's
6314 and systemd's most recent version number.
6316 * udev: all udev sources are merged into the systemd source tree now.
6317 All future udev development will happen in the systemd tree. It
6318 is still fully supported to use the udev daemon and tools without
6319 systemd running, like in initramfs or other init systems. Building
6320 udev though, will require the *build* of the systemd tree, but
6321 udev can be properly *run* without systemd.
6323 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
6324 should be used to create dead device nodes as workarounds for broken
6327 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
6328 no longer supported. udev_monitor_new_from_netlink() needs to be
6329 used to subscribe to events.
6331 * udev: when udevd is started by systemd, processes which are left
6332 behind by forking them off of udev rules, are unconditionally cleaned
6333 up and killed now after the event handling has finished. Services or
6334 daemons must be started as systemd services. Services can be
6335 pulled-in by udev to get started, but they can no longer be directly
6336 forked by udev rules.
6338 * udev: the daemon binary is called systemd-udevd now and installed
6339 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
6340 to adapt to that, create symlink, or rename the binary after building
6343 * libudev no longer provides these symbols:
6344 udev_monitor_from_socket()
6345 udev_queue_get_failed_list_entry()
6346 udev_get_{dev,sys,run}_path()
6347 The versions number was bumped and symbol versioning introduced.
6349 * systemd-loginctl and systemd-journalctl have been renamed
6350 to loginctl and journalctl to match systemctl.
6352 * The config files: /etc/systemd/systemd-logind.conf and
6353 /etc/systemd/systemd-journald.conf have been renamed to
6354 logind.conf and journald.conf. Package updates should rename
6355 the files to the new names on upgrade.
6357 * For almost all files the license is now LGPL2.1+, changed
6358 from the previous GPL2.0+. Exceptions are some minor stuff
6359 of udev (which will be changed to LGPL2.1 eventually, too),
6360 and the MIT licensed sd-daemon.[ch] library that is suitable
6361 to be used as drop-in files.
6363 * systemd and logind now handle system sleep states, in
6364 particular suspending and hibernating.
6366 * logind now implements a sleep/shutdown/idle inhibiting logic
6367 suitable for a variety of uses. Soonishly Lennart will blog
6368 about this in more detail.
6370 * var-run.mount and var-lock.mount are no longer provided
6371 (which previously bind mounted these directories to their new
6372 places). Distributions which have not converted these
6373 directories to symlinks should consider stealing these files
6374 from git history and add them downstream.
6376 * We introduced the Documentation= field for units and added
6377 this to all our shipped units. This is useful to make it
6378 easier to explore the boot and the purpose of the various
6381 * All smaller setup units (such as
6382 systemd-vconsole-setup.service) now detect properly if they
6383 are run in a container and are skipped when
6384 appropriate. This guarantees an entirely noise-free boot in
6385 Linux container environments such as systemd-nspawn.
6387 * A framework for implementing offline system updates is now
6388 integrated, for details see:
6389 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
6391 * A new service type Type=idle is available now which helps us
6392 avoiding ugly interleaving of getty output and boot status
6395 * There's now a system-wide CapabilityBoundingSet= option to
6396 globally reduce the set of capabilities for the
6397 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
6398 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
6399 even CAP_NET_ADMIN system-wide for secure systems.
6401 * There are now system-wide DefaultLimitXXX= options to
6402 globally change the defaults of the various resource limits
6403 for all units started by PID 1.
6405 * Harald Hoyer's systemd test suite has been integrated into
6406 systemd which allows easy testing of systemd builds in qemu
6407 and nspawn. (This is really awesome! Ask us for details!)
6409 * The fstab parser is now implemented as generator, not inside
6412 * systemctl will now warn you if .mount units generated from
6413 /etc/fstab are out of date due to changes in fstab that
6414 have not been read by systemd yet.
6416 * systemd is now suitable for usage in initrds. Dracut has
6417 already been updated to make use of this. With this in place
6418 initrds get a slight bit faster but primarily are much
6419 easier to introspect and debug since "systemctl status" in
6420 the host system can be used to introspect initrd services,
6421 and the journal from the initrd is kept around too.
6423 * systemd-delta has been added, a tool to explore differences
6424 between user/admin configuration and vendor defaults.
6426 * PrivateTmp= now affects both /tmp and /var/tmp.
6428 * Boot time status messages are now much prettier and feature
6429 proper english language. Booting up systemd has never been
6432 * Read-ahead pack files now include the inode number of all
6433 files to pre-cache. When the inode changes the pre-caching
6434 is not attempted. This should be nicer to deal with updated
6435 packages which might result in changes of read-ahead
6438 * We now temporaritly lower the kernel's read_ahead_kb variable
6439 when collecting read-ahead data to ensure the kernel's
6440 built-in read-ahead does not add noise to our measurements
6441 of necessary blocks to pre-cache.
6443 * There's now RequiresMountsFor= to add automatic dependencies
6444 for all mounts necessary for a specific file system path.
6446 * MountAuto= and SwapAuto= have been removed from
6447 system.conf. Mounting file systems at boot has to take place
6450 * nspawn now learned a new switch --uuid= to set the machine
6451 ID on the command line.
6453 * nspawn now learned the -b switch to automatically search
6456 * vt102 is now the default TERM for serial TTYs, upgraded from
6459 * systemd-logind now works on VT-less systems.
6461 * The build tree has been reorganized. The individual
6462 components now have directories of their own.
6464 * A new condition type ConditionPathIsReadWrite= is now available.
6466 * nspawn learned the new -C switch to create cgroups for the
6467 container in other hierarchies.
6469 * We now have support for hardware watchdogs, configurable in
6472 * The scheduled shutdown logic now has a public API.
6474 * We now mount /tmp as tmpfs by default, but this can be
6475 masked and /etc/fstab can override it.
6477 * Since udisks does not make use of /media anymore we are not
6478 mounting a tmpfs on it anymore.
6480 * journalctl gained a new --local switch to only interleave
6481 locally generated journal files.
6483 * We can now load the IMA policy at boot automatically.
6485 * The GTK tools have been split off into a systemd-ui.
6487 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
6488 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
6489 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
6490 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
6491 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
6492 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
6493 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
6494 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
6495 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
6500 * This is mostly a bugfix release
6502 * Support optional initialization of the machine ID from the
6503 KVM or container configured UUID.
6505 * Support immediate reboots with "systemctl reboot -ff"
6507 * Show /etc/os-release data in systemd-analyze output
6509 * Many bugfixes for the journal, including endianness fixes and
6510 ensuring that disk space enforcement works
6512 * sd-login.h is C++ compatible again
6514 * Extend the /etc/os-release format on request of the Debian
6517 * We now refuse non-UTF8 strings used in various configuration
6518 and unit files. This is done to ensure we do not pass invalid
6519 data over D-Bus or expose it elsewhere.
6521 * Register Mimo USB Screens as suitable for automatic seat
6524 * Read SELinux client context from journal clients in a race
6527 * Reorder configuration file lookup order. /etc now always
6528 overrides /run in order to allow the administrator to always
6529 and unconditionally override vendor-supplied or
6530 automatically generated data.
6532 * The various user visible bits of the journal now have man
6533 pages. We still lack man pages for the journal API calls
6536 * We now ship all man pages in HTML format again in the
6539 Contributions from: Dave Reisner, Dirk Eibach, Frederic
6540 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
6541 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
6546 * This is mostly a bugfix release
6548 * systems lacking /etc/os-release are no longer supported.
6550 * Various functionality updates to libsystemd-login.so
6552 * Track class of PAM logins to distinguish greeters from
6555 Contributions from: Kay Sievers, Lennart Poettering, Michael
6560 * This is an important bugfix release for v41.
6562 * Building man pages is now optional which should be useful
6563 for those building systemd from git but unwilling to install
6566 * Watchdog support for supervising services is now usable. In
6567 a future release support for hardware watchdogs
6568 (i.e. /dev/watchdog) will be added building on this.
6570 * Service start rate limiting is now configurable and can be
6571 turned off per service. When a start rate limit is hit a
6572 reboot can automatically be triggered.
6574 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
6576 Contributions from: Benjamin Franzke, Bill Nottingham,
6577 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
6578 Schmidt, Michał Górny, Piotr Drąg
6582 * The systemd binary is installed /usr/lib/systemd/systemd now;
6583 An existing /sbin/init symlink needs to be adapted with the
6586 * The code that loads kernel modules has been ported to invoke
6587 libkmod directly, instead of modprobe. This means we do not
6588 support systems with module-init-tools anymore.
6590 * Watchdog support is now already useful, but still not
6593 * A new kernel command line option systemd.setenv= is
6594 understood to set system wide environment variables
6595 dynamically at boot.
6597 * We now limit the set of capabilities of systemd-journald.
6599 * We now set SIGPIPE to ignore by default, since it only is
6600 useful in shell pipelines, and has little use in general
6601 code. This can be disabled with IgnoreSIPIPE=no in unit
6604 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
6605 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
6610 * This is mostly a bugfix release
6612 * We now expose the reason why a service failed in the
6613 "Result" D-Bus property.
6615 * Rudimentary service watchdog support (will be completed over
6616 the next few releases.)
6618 * When systemd forks off in order execute some service we will
6619 now immediately changes its argv[0] to reflect which process
6620 it will execute. This is useful to minimize the time window
6621 with a generic argv[0], which makes bootcharts more useful
6623 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
6624 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
6625 Mike Kazantsev, Ray Strode
6629 * This is mostly a test release, but incorporates many
6632 * New systemd-cgtop tool to show control groups by their
6635 * Linking against libacl for ACLs is optional again. If
6636 disabled, support tracking device access for active logins
6637 goes becomes unavailable, and so does access to the user
6638 journals by the respective users.
6640 * If a group "adm" exists, journal files are automatically
6641 owned by them, thus allow members of this group full access
6642 to the system journal as well as all user journals.
6644 * The journal now stores the SELinux context of the logging
6645 client for all entries.
6647 * Add C++ inclusion guards to all public headers
6649 * New output mode "cat" in the journal to print only text
6650 messages, without any meta data like date or time.
6652 * Include tiny X server wrapper as a temporary stop-gap to
6653 teach XOrg udev display enumeration. This is used by display
6654 managers such as gdm, and will go away as soon as XOrg
6655 learned native udev hotplugging for display devices.
6657 * Add new systemd-cat tool for executing arbitrary programs
6658 with STDERR/STDOUT connected to the journal. Can also act as
6659 BSD logger replacement, and does so by default.
6661 * Optionally store all locally generated coredumps in the
6662 journal along with meta data.
6664 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
6665 writing short strings to files (for usage for /sys), and for
6666 creating symlinks, character and block device nodes.
6668 * New unit file option ControlGroupPersistent= to make cgroups
6669 persistent, following the mechanisms outlined in
6670 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
6672 * Support multiple local RTCs in a sane way
6674 * No longer monopolize IO when replaying readahead data on
6675 rotating disks, since we might starve non-file-system IO to
6676 death, since fanotify() will not see accesses done by blkid,
6679 * Do not show kernel threads in systemd-cgls anymore, unless
6680 requested with new -k switch.
6682 Contributions from: Dan Horák, Kay Sievers, Lennart
6683 Poettering, Michal Schmidt
6687 * This is mostly a test release, but incorporates many
6690 * The git repository moved to:
6691 git://anongit.freedesktop.org/systemd/systemd
6692 ssh://git.freedesktop.org/git/systemd/systemd
6694 * First release with the journal
6695 http://0pointer.de/blog/projects/the-journal.html
6697 * The journal replaces both systemd-kmsg-syslogd and
6698 systemd-stdout-bridge.
6700 * New sd_pid_get_unit() API call in libsystemd-logind
6702 * Many systemadm clean-ups
6704 * Introduce remote-fs-pre.target which is ordered before all
6705 remote mounts and may be used to start services before all
6708 * Added Mageia support
6710 * Add bash completion for systemd-loginctl
6712 * Actively monitor PID file creation for daemons which exit in
6713 the parent process before having finished writing the PID
6714 file in the daemon process. Daemons which do this need to be
6715 fixed (i.e. PID file creation must have finished before the
6716 parent exits), but we now react a bit more gracefully to them.
6718 * Add colourful boot output, mimicking the well-known output
6719 of existing distributions.
6721 * New option PassCredentials= for socket units, for
6722 compatibility with a recent kernel ABI breakage.
6724 * /etc/rc.local is now hooked in via a generator binary, and
6725 thus will no longer act as synchronization point during
6728 * systemctl list-unit-files now supports --root=.
6730 * systemd-tmpfiles now understands two new commands: z, Z for
6731 relabelling files according to the SELinux database. This is
6732 useful to apply SELinux labels to specific files in /sys,
6735 * Output of SysV services is now forwarded to both the console
6736 and the journal by default, not only just the console.
6738 * New man pages for all APIs from libsystemd-login.
6740 * The build tree got reorganized and the build system is a
6741 lot more modular allowing embedded setups to specifically
6742 select the components of systemd they are interested in.
6744 * Support for Linux systems lacking the kernel VT subsystem is
6747 * configure's --with-rootdir= got renamed to
6748 --with-rootprefix= to follow the naming used by udev and
6751 * Unless specified otherwise we will now install to /usr instead
6752 of /usr/local by default.
6754 * Processes with '@' in argv[0][0] are now excluded from the
6755 final shut-down killing spree, following the logic explained
6757 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
6759 * All processes remaining in a service cgroup when we enter
6760 the START or START_PRE states are now killed with
6761 SIGKILL. That means it is no longer possible to spawn
6762 background processes from ExecStart= lines (which was never
6763 supported anyway, and bad style).
6765 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
6766 reloading of units together.
6768 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
6769 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
6770 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
6771 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
6772 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek