1 systemd System and Service Manager
3 CHANGES WITH 233 in spe
5 * DBus policy files are now installed into /usr rather than /etc. Make
6 sure your system has dbus >= 1.9.18 running before upgrading to this
7 version, or override the install path with --with-dbuspolicydir= .
9 * The shell invoked by debug-shell.service now defaults to /bin/sh in
10 all cases. If distributions want to use a different shell for this
11 purpose (for example Fedora's /sbin/sushell) they need to specify
12 this explicitly at configure time using --with-debug-shell=.
14 * The confirmation spawn prompt has been reworked to offer the
17 (c)ontinue, proceed without asking anymore
18 (D)ump, show the state of the unit
19 (f)ail, don't execute the command and pretend it failed
21 (i)nfo, show a short summary of the unit
22 (j)obs, show jobs that are in progress
23 (s)kip, don't execute the command and pretend it succeeded
24 (y)es, execute the command
26 The 'n' choice for the confirmation spawn prompt has been removed,
27 because its meaning was confusing.
29 * Services of Type=notify require a READY=1 notification to be sent
30 during startup. If no such message is sent, the service now fails,
31 even if the main process exited with a successful exit code.
33 * The option MulticastDNS= of network configuration files has got
34 actual implementation. With MulticastDNS=yes a host can resolve
35 names of remote hosts and to reply to mDNS's A and AAAA requests
40 * The new RemoveIPC= option can be used to remove IPC objects owned by
41 the user or group of a service when that service exits.
43 * The new ProtectKernelModules= option can be used to disable explicit
44 load and unload operations of kernel modules by a service. In
45 addition access to /usr/lib/modules is removed if this option is set.
47 * ProtectSystem= option gained a new value "strict", which causes the
48 whole file system tree with the exception of /dev, /proc, and /sys,
49 to be remounted read-only for a service.
51 * The new ProtectKernelTunables= option can be used to disable
52 modification of configuration files in /sys and /proc by a service.
53 Various directories and files are remounted read-only, so access is
54 restricted even if the file permissions would allow it.
56 * The new ProtectControlGroups= option can be used to disable write
57 access by a service to /sys/fs/cgroup.
59 * Various systemd services have been hardened with
60 ProtectKernelTunables=yes, ProtectControlGroups=yes,
61 RestrictAddressFamilies=.
63 * Support for dynamically creating users for the lifetime of a service
64 has been added. If DynamicUser=yes is specified, user and group IDs
65 will be allocated from the range 61184..65519 for the lifetime of the
66 service. They can be resolved using the new nss-systemd.so NSS
67 module. The module must be enabled in /etc/nsswitch.conf. Services
68 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
69 any resources allocated by the service will be cleaned up when the
70 service exits. They also have ProtectHome=read-only and
71 ProtectSystem=strict enabled, so they are not able to make any
72 permanent modifications to the system.
74 * The nss-systemd module also always resolves root and nobody, making
75 it possible to have no /etc/passwd or /etc/group files in minimal
76 container or chroot environments.
78 * Services may be started with their own user namespace using the new
79 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
80 under which the service is running are mapped. All other users are
83 * Support for the cgroup namespace has been added to systemd-nspawn. If
84 supported by kernel, the container system started by systemd-nspawn
85 will have its own view of the cgroup hierarchy. This new behaviour
86 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
88 * The new MemorySwapMax= option can be used to limit the maximum swap
89 usage under the unified cgroup hierarchy.
91 * Support for the CPU controller in the unified cgroup hierarchy has
92 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
93 options. This controller requires out-of-tree patches for the kernel
94 and the support is provisional.
96 * Mount and automount units may now be created transiently
97 (i.e. dynamically at runtime via the bus API, instead of requiring
98 unit files in the file system).
100 * systemd-mount is a new tool which may mount file systems – much like
101 mount(8), optionally pulling in additional dependencies through
102 transient .mount and .automount units. For example, this tool
103 automatically runs fsck on a backing block device before mounting,
104 and allows the automount logic to be used dynamically from the
105 command line for establishing mount points. This tool is particularly
106 useful when dealing with removable media, as it will ensure fsck is
107 run – if necessary – before the first access and that the file system
108 is quickly unmounted after each access by utilizing the automount
109 logic. This maximizes the chance that the file system on the
110 removable media stays in a clean state, and if it isn't in a clean
111 state is fixed automatically.
113 * LazyUnmount=yes option for mount units has been added to expose the
114 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
117 * /efi will be used as the mount point of the EFI boot partition, if
118 the directory is present, and the mount point was not configured
119 through other means (e.g. fstab). If /efi directory does not exist,
120 /boot will be used as before. This makes it easier to automatically
121 mount the EFI partition on systems where /boot is used for something
124 * When operating on GPT disk images for containers, systemd-nspawn will
125 now mount the ESP to /boot or /efi according to the same rules as PID
126 1 running on a host. This allows tools like "bootctl" to operate
127 correctly within such containers, in order to make container images
128 bootable on physical systems.
130 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
132 * Two new user session targets have been added to support running
133 graphical sessions under the systemd --user instance:
134 graphical-session.target and graphical-session-pre.target. See
135 systemd.special(7) for a description of how those targets should be
138 * The vconsole initialization code has been significantly reworked to
139 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
140 support unicode keymaps. Font and keymap configuration will now be
141 copied to all allocated virtual consoles.
143 * FreeBSD's bhyve virtualization is now detected.
145 * Information recorded in the journal for core dumps now includes the
146 contents of /proc/mountinfo and the command line of the process at
147 the top of the process hierarchy (which is usually the init process
150 * systemd-journal-gatewayd learned the --directory= option to serve
151 files from the specified location.
153 * journalctl --root=… can be used to peruse the journal in the
154 /var/log/ directories inside of a container tree. This is similar to
155 the existing --machine= option, but does not require the container to
158 * The hardware database has been extended to support
159 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
162 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
163 specify the click rate for mice which include a horizontal wheel with
164 a click rate that is different than the one for the vertical wheel.
166 * systemd-run gained a new --wait option that makes service execution
167 synchronous. (Specifically, the command will not return until the
168 specified service binary exited.)
170 * systemctl gained a new --wait option that causes the start command to
171 wait until the units being started have terminated again.
173 * A new journal output mode "short-full" has been added which displays
174 timestamps with abbreviated English day names and adds a timezone
175 suffix. Those timestamps include more information than the default
176 "short" output mode, and can be passed directly to journalctl's
177 --since= and --until= options.
179 * /etc/resolv.conf will be bind-mounted into containers started by
180 systemd-nspawn, if possible, so any changes to resolv.conf contents
181 are automatically propagated to the container.
183 * The number of instances for socket-activated services originating
184 from a single IP address can be limited with
185 MaxConnectionsPerSource=, extending the existing setting of
188 * systemd-networkd gained support for vcan ("Virtual CAN") interface
191 * .netdev and .network configuration can now be extended through
194 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
195 Segmentation Offload, Generic Receive Offload, Large Receive Offload
196 can be enabled and disabled using the new UDPSegmentationOffload=,
197 TCPSegmentationOffload=, GenericSegmentationOffload=,
198 GenericReceiveOffload=, LargeReceiveOffload= options in the
199 [Link] section of .link files.
201 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
202 Port VLAN ID can be configured for bridge devices using the new STP=,
203 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
204 section of .netdev files.
206 * The route table to which routes received over DHCP or RA should be
207 added can be configured with the new RouteTable= option in the [DHCP]
208 and [IPv6AcceptRA] sections of .network files.
210 * The Address Resolution Protocol can be disabled on links managed by
211 systemd-networkd using the ARP=no setting in the [Link] section of
214 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
215 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
216 encode information about the result and exit codes of the current
217 service runtime cycle.
219 * systemd-sysctl will now configure kernel parameters in the order
220 they occur in the configuration files. This matches what sysctl
221 has been traditionally doing.
223 * kernel-install "plugins" that are executed to perform various
224 tasks after a new kernel is added and before an old one is removed
225 can now return a special value to terminate the procedure and
226 prevent any later plugins from running.
228 * Journald's SplitMode=login setting has been deprecated. It has been
229 removed from documentation, and its use is discouraged. In a future
230 release it will be completely removed, and made equivalent to current
231 default of SplitMode=uid.
233 * Storage=both option setting in /etc/systemd/coredump.conf has been
234 removed. With fast LZ4 compression storing the core dump twice is not
237 * The --share-system systemd-nspawn option has been replaced with an
238 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
239 this functionality is discouraged. In addition the variables
240 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
241 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
242 individual namespaces.
244 * "machinectl list" now shows the IP address of running containers in
245 the output, as well as OS release information.
247 * "loginctl list" now shows the TTY of each session in the output.
249 * sd-bus gained new API calls sd_bus_track_set_recursive(),
250 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
251 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
252 tracking objects in a "recursive" mode, where a single client can be
253 counted multiple times, if it takes multiple references.
255 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
256 sd_bus_get_exit_on_disconnect(). They may be used to to make a
257 process using sd-bus automatically exit if the bus connection is
260 * Bus clients of the service manager may now "pin" loaded units into
261 memory, by taking an explicit reference on them. This is useful to
262 ensure the client can retrieve runtime data about the service even
263 after the service completed execution. Taking such a reference is
264 available only for privileged clients and should be helpful to watch
265 running services in a race-free manner, and in particular collect
266 information about exit statuses and results.
268 * The nss-resolve module has been changed to strictly return UNAVAIL
269 when communication via D-Bus with resolved failed, and NOTFOUND when
270 a lookup completed but was negative. This means it is now possible to
271 neatly configure fallbacks using nsswitch.conf result checking
272 expressions. Taking benefit of this, the new recommended
273 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
275 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
277 * A new setting CtrlAltDelBurstAction= has been added to
278 /etc/systemd/system.conf which may be used to configure the precise
279 behaviour if the user on the console presses Ctrl-Alt-Del more often
280 than 7 times in 2s. Previously this would unconditionally result in
281 an expedited, immediate reboot. With this new setting the precise
282 operation may be configured in more detail, and also turned off
285 * In .netdev files two new settings RemoteChecksumTx= and
286 RemoteChecksumRx= are now understood that permit configuring the
287 remote checksumming logic for VXLAN networks.
289 * The service manager learnt a new "invocation ID" concept for invoked
290 services. Each runtime cycle of a service will get a new invocation
291 ID (a 128bit random UUID) assigned that identifies the current
292 run of the service uniquely and globally. A new invocation ID
293 is generated each time a service starts up. The journal will store
294 the invocation ID of a service along with any logged messages, thus
295 making the invocation ID useful for matching the online runtime of a
296 service with the offline log data it generated in a safe way without
297 relying on synchronized timestamps. In many ways this new service
298 invocation ID concept is similar to the kernel's boot ID concept that
299 uniquely and globally identifies the runtime of each boot. The
300 invocation ID of a service is passed to the service itself via an
301 environment variable ($INVOCATION_ID). A new bus call
302 GetUnitByInvocationID() has been added that is similar to GetUnit()
303 but instead of retrieving the bus path for a unit by its name
304 retrieves it by its invocation ID. The returned path is valid only as
305 long as the passed invocation ID is current.
307 * systemd-resolved gained a new "DNSStubListener" setting in
308 resolved.conf. It either takes a boolean value or the special values
309 "udp" and "tcp", and configures whether to enable the stub DNS
310 listener on 127.0.0.53:53.
312 * IP addresses configured via networkd may now carry additional
313 configuration settings supported by the kernel. New options include:
314 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
315 PrefixRoute=, AutoJoin=.
317 * The PAM configuration fragment file for "user@.service" shipped with
318 systemd (i.e. the --user instance of systemd) has been stripped to
319 the minimum necessary to make the system boot. Previously, it
320 contained Fedora-specific stanzas that did not apply to other
321 distributions. It is expected that downstream distributions add
322 additional configuration lines, matching their needs to this file,
323 using it only as rough template of what systemd itself needs. Note
324 that this reduced fragment does not even include an invocation of
325 pam_limits which most distributions probably want to add, even though
326 systemd itself does not need it. (There's also the new build time
327 option --with-pamconfdir=no to disable installation of the PAM
330 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
331 capability is now also dropped from its set (in addition to
332 CAP_SYS_MKNOD as before).
334 * In service unit files it is now possible to connect a specific named
335 file descriptor with stdin/stdout/stdout of an executed service. The
336 name may be specified in matching .socket units using the
337 FileDescriptorName= setting.
339 * A number of journal settings may now be configured on the kernel
340 command line. Specifically, the following options are now understood:
341 systemd.journald.max_level_console=,
342 systemd.journald.max_level_store=,
343 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
344 systemd.journald.max_level_wall=.
346 * "systemctl is-enabled --full" will now show by which symlinks a unit
347 file is enabled in the unit dependency tree.
349 * Support for VeraCrypt encrypted partitions has been added to the
350 "cryptsetup" logic and /etc/crypttab.
352 * systemd-detect-virt gained support for a new --private-users switch
353 that checks whether the invoking processes are running inside a user
354 namespace. Similar, a new special value "private-users" for the
355 existing ConditionVirtualization= setting has been added, permitting
356 skipping of specific units in user namespace environments.
358 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
359 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
360 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
361 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
362 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
363 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
364 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
365 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
366 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
367 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
368 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
369 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
370 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
371 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
372 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
373 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
374 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
375 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
376 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
377 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
378 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
379 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
380 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
381 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
382 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
383 Jędrzejewski-Szmek, Zeal Jagannatha
385 — Santa Fe, 2016-11-03
389 * In service units the various ExecXYZ= settings have been extended
390 with an additional special character as first argument of the
391 assigned value: if the character '+' is used the specified command
392 line it will be run with full privileges, regardless of User=,
393 Group=, CapabilityBoundingSet= and similar options. The effect is
394 similar to the existing PermissionsStartOnly= option, but allows
395 configuration of this concept for each executed command line
398 * Services may now alter the service watchdog timeout at runtime by
399 sending a WATCHDOG_USEC= message via sd_notify().
401 * MemoryLimit= and related unit settings now optionally take percentage
402 specifications. The percentage is taken relative to the amount of
403 physical memory in the system (or in case of containers, the assigned
404 amount of memory). This allows scaling service resources neatly with
405 the amount of RAM available on the system. Similarly, systemd-logind's
406 RuntimeDirectorySize= option now also optionally takes percentage
409 * In similar fashion TasksMax= takes percentage values now, too. The
410 value is taken relative to the configured maximum number of processes
411 on the system. The per-service task maximum has been changed to 15%
412 using this functionality. (Effectively this is an increase of 512 →
413 4915 for service units, given the kernel's default pid_max setting.)
415 * Calendar time specifications in .timer units now understand a ".."
416 syntax for time ranges. Example: "4..7:10" may now be used for
417 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
420 * The InaccessableDirectories=, ReadOnlyDirectories= and
421 ReadWriteDirectories= unit file settings have been renamed to
422 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
423 applied to all kinds of file nodes, and not just directories, with
424 the exception of symlinks. Specifically these settings may now be
425 used on block and character device nodes, UNIX sockets and FIFOS as
426 well as regular files. The old names of these settings remain
427 available for compatibility.
429 * systemd will now log about all service processes it kills forcibly
430 (using SIGKILL) because they remained after the clean shutdown phase
431 of the service completed. This should help identifying services that
432 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
433 systemd-logind's configuration a similar log message is generated for
434 processes killed at the end of each session due to this setting.
436 * systemd will now set the $JOURNAL_STREAM environment variable for all
437 services whose stdout/stderr are connected to the Journal (which
438 effectively means by default: all services). The variable contains
439 the device and inode number of the file descriptor used for
440 stdout/stderr. This may be used by invoked programs to detect whether
441 their stdout/stderr is connected to the Journal, in which case they
442 can switch over to direct Journal communication, thus being able to
443 pass extended, structured metadata along with their log messages. As
444 one example, this is now used by glib's logging primitives.
446 * When using systemd's default tmp.mount unit for /tmp, the mount point
447 will now be established with the "nosuid" and "nodev" options. This
448 avoids privilege escalation attacks that put traps and exploits into
449 /tmp. However, this might cause problems if you e. g. put container
450 images or overlays into /tmp; if you need this, override tmp.mount's
451 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
454 * systemd now supports the "memory" cgroup controller also on
457 * The systemd-cgtop tool now optionally takes a control group path as
458 command line argument. If specified, the control group list shown is
459 limited to subgroups of that group.
461 * The SystemCallFilter= unit file setting gained support for
462 pre-defined, named system call filter sets. For example
463 SystemCallFilter=@clock is now an effective way to make all clock
464 changing-related system calls unavailable to a service. A number of
465 similar pre-defined groups are defined. Writing system call filters
466 for system services is simplified substantially with this new
467 concept. Accordingly, all of systemd's own, long-running services now
468 enable system call filtering based on this, by default.
470 * A new service setting MemoryDenyWriteExecute= has been added, taking
471 a boolean value. If turned on, a service may no longer create memory
472 mappings that are writable and executable at the same time. This
473 enhances security for services where this is enabled as it becomes
474 harder to dynamically write and then execute memory in exploited
475 service processes. This option has been enabled for all of systemd's
476 own long-running services.
478 * A new RestrictRealtime= service setting has been added, taking a
479 boolean argument. If set the service's processes may no longer
480 acquire realtime scheduling. This improves security as realtime
481 scheduling may otherwise be used to easily freeze the system.
483 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
484 value. This may be used for requesting that the system manager inside
485 of the container reports start-up completion to nspawn which then
486 propagates this notification further to the service manager
487 supervising nspawn itself. A related option NotifyReady= in .nspawn
488 files has been added too. This functionality allows ordering of the
489 start-up of multiple containers using the usual systemd ordering
492 * machinectl gained a new command "stop" that is an alias for
495 * systemd-resolved gained support for contacting DNS servers on
496 link-local IPv6 addresses.
498 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
499 its caches. A method call for requesting the same operation has been
500 added to the bus API too, and is made available via "systemd-resolve
503 * systemd-resolve gained a new --status switch. If passed a brief
504 summary of the used DNS configuration with per-interface information
507 * resolved.conf gained a new Cache= boolean option, defaulting to
508 on. If turned off local DNS caching is disabled. This comes with a
509 performance penalty in particular when DNSSEC is enabled. Note that
510 resolved disables its internal caching implicitly anyway, when the
511 configured DNS server is on a host-local IP address such as ::1 or
512 127.0.0.1, thus automatically avoiding double local caching.
514 * systemd-resolved now listens on the local IP address 127.0.0.53:53
515 for DNS requests. This improves compatibility with local programs
516 that do not use the libc NSS or systemd-resolved's bus APIs for name
517 resolution. This minimal DNS service is only available to local
518 programs and does not implement the full DNS protocol, but enough to
519 cover local DNS clients. A new, static resolv.conf file, listing just
520 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
521 now recommended to make /etc/resolv.conf a symlink to this file in
522 order to route all DNS lookups to systemd-resolved, regardless if
523 done via NSS, the bus API or raw DNS packets. Note that this local
524 DNS service is not as fully featured as the libc NSS or
525 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
526 used to deliver link-local address information (as this implies
527 sending a local interface index along), LLMNR/mDNS support via this
528 interface is severely restricted. It is thus strongly recommended for
529 all applications to use the libc NSS API or native systemd-resolved
532 * systemd-networkd's bridge support learned a new setting
533 VLANFiltering= for controlling VLAN filtering. Moreover a new section
534 in .network files has been added for configuring VLAN bridging in
535 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
537 * systemd-networkd's IPv6 Router Advertisement code now makes use of
538 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
539 now be acquired without relying on DHCPv6. Two new options
540 UseDomains= and UseDNS= have been added to configure this behaviour.
542 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
543 renamed IPv6AcceptRA=, without altering its behaviour. The old
544 setting name remains available for compatibility reasons.
546 * The systemd-networkd VTI/VTI6 tunneling support gained new options
547 Key=, InputKey= and OutputKey=.
549 * systemd-networkd gained support for VRF ("Virtual Routing Function")
550 interface configuration.
552 * "systemctl edit" may now be used to create new unit files by
553 specifying the --force switch.
555 * sd-event gained a new function sd_event_get_iteration() for
556 requesting the current iteration counter of the event loop. It starts
557 at zero and is increased by one with each event loop iteration.
559 * A new rpm macro %systemd_ordering is provided by the macros.systemd
560 file. It can be used in lieu of %systemd_requires in packages which
561 don't use any systemd functionality and are intended to be installed
562 in minimal containers without systemd present. This macro provides
563 ordering dependencies to ensure that if the package is installed in
564 the same rpm transaction as systemd, systemd will be installed before
565 the scriptlets for the package are executed, allowing unit presets
568 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
569 been added to simplify packaging of generators.
571 * The os-release file gained VERSION_CODENAME field for the
572 distribution nickname (e.g. VERSION_CODENAME=woody).
574 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
575 can be set to disable parsing of metadata and the creation
576 of persistent symlinks for that device.
578 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
579 to make them available to logged-in users has been reverted.
581 * Much of the common code of the various systemd components is now
582 built into an internal shared library libsystemd-shared-231.so
583 (incorporating the systemd version number in the name, to be updated
584 with future releases) that the components link to. This should
585 decrease systemd footprint both in memory during runtime and on
586 disk. Note that the shared library is not for public use, and is
587 neither API not ABI stable, but is likely to change with every new
588 released update. Packagers need to make sure that binaries
589 linking to libsystemd-shared.so are updated in step with the
592 * Configuration for "mkosi" is now part of the systemd
593 repository. mkosi is a tool to easily build legacy-free OS images,
594 and is available on github: https://github.com/systemd/mkosi. If
595 "mkosi" is invoked in the build tree a new raw OS image is generated
596 incorporating the systemd sources currently being worked on and a
597 clean, fresh distribution installation. The generated OS image may be
598 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
599 UEFI PC. This functionality is particularly useful to easily test
600 local changes made to systemd in a pristine, defined environment. See
603 * configure learned the --with-support-url= option to specify the
604 distribution's bugtracker.
606 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
607 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
608 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
609 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
610 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
611 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
612 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
613 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
614 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
615 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
616 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
617 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
618 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
619 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
620 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
621 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
622 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
623 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
624 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
630 * DNSSEC is now turned on by default in systemd-resolved (in
631 "allow-downgrade" mode), but may be turned off during compile time by
632 passing "--with-default-dnssec=no" to "configure" (and of course,
633 during runtime with DNSSEC= in resolved.conf). We recommend
634 downstreams to leave this on at least during development cycles and
635 report any issues with the DNSSEC logic upstream. We are very
636 interested in collecting feedback about the DNSSEC validator and its
637 limitations in the wild. Note however, that DNSSEC support is
638 probably nothing downstreams should turn on in stable distros just
639 yet, as it might create incompatibilities with a few DNS servers and
640 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
641 automatically whenever we detect such incompatible setups, but there
642 might be systems we do not cover yet. Hence: please help us testing
643 the DNSSEC code, leave this on where you can, report back, but then
644 again don't consider turning this on in your stable, LTS or
645 production release just yet. (Note that you have to enable
646 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
647 and its DNSSEC mode for host name resolution from local
650 * systemd-resolve conveniently resolves DANE records with the --tlsa
651 option and OPENPGPKEY records with the --openpgp option. It also
652 supports dumping raw DNS record data via the new --raw= switch.
654 * systemd-logind will now by default terminate user processes that are
655 part of the user session scope unit (session-XX.scope) when the user
656 logs out. This behavior is controlled by the KillUserProcesses=
657 setting in logind.conf, and the previous default of "no" is now
658 changed to "yes". This means that user sessions will be properly
659 cleaned up after, but additional steps are necessary to allow
660 intentionally long-running processes to survive logout.
662 While the user is logged in at least once, user@.service is running,
663 and any service that should survive the end of any individual login
664 session can be started at a user service or scope using systemd-run.
665 systemd-run(1) man page has been extended with an example which shows
666 how to run screen in a scope unit underneath user@.service. The same
667 command works for tmux.
669 After the user logs out of all sessions, user@.service will be
670 terminated too, by default, unless the user has "lingering" enabled.
671 To effectively allow users to run long-term tasks even if they are
672 logged out, lingering must be enabled for them. See loginctl(1) for
673 details. The default polkit policy was modified to allow users to
674 set lingering for themselves without authentication.
676 Previous defaults can be restored at compile time by the
677 --without-kill-user-processes option to "configure".
679 * systemd-logind gained new configuration settings SessionsMax= and
680 InhibitorsMax=, both with a default of 8192. It will not register new
681 user sessions or inhibitors above this limit.
683 * systemd-logind will now reload configuration on SIGHUP.
685 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
686 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
687 enable. Also, support for the "io" cgroup controller in the unified
688 hierarchy has been added, so that the "memory", "pids" and "io" are
689 now the controllers that are supported on the unified hierarchy.
691 WARNING: it is not possible to use previous systemd versions with
692 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
693 is necessary to also update systemd in the initramfs if using the
694 unified hierarchy. An updated SELinux policy is also required.
696 * LLDP support has been extended, and both passive (receive-only) and
697 active (sender) modes are supported. Passive mode ("routers-only") is
698 enabled by default in systemd-networkd. Active LLDP mode is enabled
699 by default for containers on the internal network. The "networkctl
700 lldp" command may be used to list information gathered. "networkctl
701 status" will also show basic LLDP information on connected peers now.
703 * The IAID and DUID unique identifier sent in DHCP requests may now be
704 configured for the system and each .network file managed by
705 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
707 * systemd-networkd gained support for configuring proxy ARP support for
708 each interface, via the ProxyArp= setting in .network files. It also
709 gained support for configuring the multicast querier feature of
710 bridge devices, via the new MulticastQuerier= setting in .netdev
711 files. Similarly, snooping on the IGMP traffic can be controlled
712 via the new setting MulticastSnooping=.
714 A new setting PreferredLifetime= has been added for addresses
715 configured in .network file to configure the lifetime intended for an
718 The systemd-networkd DHCP server gained the option EmitRouter=, which
719 defaults to yes, to configure whether the DHCP Option 3 (Router)
722 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
723 systemd-socket-activate and installed into /usr/bin. It is now fully
726 * systemd-journald now uses separate threads to flush changes to disk
727 when closing journal files, thus reducing impact of slow disk I/O on
730 * The sd-journal API gained two new calls
731 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
732 can be used to open journal files using file descriptors instead of
733 file or directory paths. sd_journal_open_container() has been
734 deprecated, sd_journal_open_directory_fd() should be used instead
735 with the flag SD_JOURNAL_OS_ROOT.
737 * journalctl learned a new output mode "-o short-unix" that outputs log
738 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
739 UTC). It also gained support for a new --no-hostname setting to
740 suppress the hostname column in the family of "short" output modes.
742 * systemd-ask-password now optionally skips printing of the password to
743 stdout with --no-output which can be useful in scripts.
745 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
746 (devices tagged with ID_MAKER_TOOL) are now tagged with
747 "uaccess" and are available to logged in users.
749 * The DeviceAllow= unit setting now supports specifiers (with "%").
751 * "systemctl show" gained a new --value switch, which allows print a
752 only the contents of a specific unit property, without also printing
753 the property's name. Similar support was added to "show*" verbs
754 of loginctl and machinectl that output "key=value" lists.
756 * A new unit type "generated" was added for files dynamically generated
757 by generator tools. Similarly, a new unit type "transient" is used
758 for unit files created using the runtime API. "systemctl enable" will
759 refuse to operate on such files.
761 * A new command "systemctl revert" has been added that may be used to
762 revert to the vendor version of a unit file, in case local changes
763 have been made by adding drop-ins or overriding the unit file.
765 * "machinectl clean" gained a new verb to automatically remove all or
766 just hidden container images.
768 * systemd-tmpfiles gained support for a new line type "e" for emptying
769 directories, if they exist, without creating them if they don't.
771 * systemd-nspawn gained support for automatically patching the UID/GIDs
772 of the owners and the ACLs of all files and directories in a
773 container tree to match the UID/GID user namespacing range selected
774 for the container invocation. This mode is enabled via the new
775 --private-users-chown switch. It also gained support for
776 automatically choosing a free, previously unused UID/GID range when
777 starting a container, via the new --private-users=pick setting (which
778 implies --private-users-chown). Together, these options for the first
779 time make user namespacing for nspawn containers fully automatic and
780 thus deployable. The systemd-nspawn@.service template unit file has
781 been changed to use this functionality by default.
783 * systemd-nspawn gained a new --network-zone= switch, that allows
784 creating ad-hoc virtual Ethernet links between multiple containers,
785 that only exist as long as at least one container referencing them is
786 running. This allows easy connecting of multiple containers with a
787 common link that implements an Ethernet broadcast domain. Each of
788 these network "zones" may be named relatively freely by the user, and
789 may be referenced by any number of containers, but each container may
790 only reference one of these "zones". On the lower level, this is
791 implemented by an automatically managed bridge network interface for
792 each zone, that is created when the first container referencing its
793 zone is created and removed when the last one referencing its zone
796 * The default start timeout may now be configured on the kernel command
797 line via systemd.default_timeout_start_sec=. It was already
798 configurable via the DefaultTimeoutStartSec= option in
799 /etc/systemd/system.conf.
801 * Socket units gained a new TriggerLimitIntervalSec= and
802 TriggerLimitBurst= setting to configure a limit on the activation
803 rate of the socket unit.
805 * The LimitNICE= setting now optionally takes normal UNIX nice values
806 in addition to the raw integer limit value. If the specified
807 parameter is prefixed with "+" or "-" and is in the range -20..19 the
808 value is understood as UNIX nice value. If not prefixed like this it
809 is understood as raw RLIMIT_NICE limit.
811 * Note that the effect of the PrivateDevices= unit file setting changed
812 slightly with this release: the per-device /dev file system will be
813 mounted read-only from this version on, and will have "noexec"
814 set. This (minor) change of behavior might cause some (exceptional)
815 legacy software to break, when PrivateDevices=yes is set for its
816 service. Please leave PrivateDevices= off if you run into problems
819 * systemd-bootchart has been split out to a separate repository:
820 https://github.com/systemd/systemd-bootchart
822 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
823 merged into the kernel in its current form.
825 * The compatibility libraries libsystemd-daemon.so,
826 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
827 which have been deprecated since systemd-209 have been removed along
828 with the corresponding pkg-config files. All symbols provided by
829 those libraries are provided by libsystemd.so.
831 * The Capabilities= unit file setting has been removed (it is ignored
832 for backwards compatibility). AmbientCapabilities= and
833 CapabilityBoundingSet= should be used instead.
835 * A new special target has been added, initrd-root-device.target,
836 which creates a synchronization point for dependencies of the root
837 device in early userspace. Initramfs builders must ensure that this
838 target is now included in early userspace.
840 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
841 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
842 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
843 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
844 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
845 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
846 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
847 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
848 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
849 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
850 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
851 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
852 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
853 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
854 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
855 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
856 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
857 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
858 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
859 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
860 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
861 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
862 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
863 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
864 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
867 — Fairfax, 2016-05-21
871 * The systemd-resolved DNS resolver service has gained a substantial
872 set of new features, most prominently it may now act as a DNSSEC
873 validating stub resolver. DNSSEC mode is currently turned off by
874 default, but is expected to be turned on by default in one of the
875 next releases. For now, we invite everybody to test the DNSSEC logic
876 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
877 service also gained a full set of D-Bus interfaces, including calls
878 to configure DNS and DNSSEC settings per link (for use by external
879 network management software). systemd-resolved and systemd-networkd
880 now distinguish between "search" and "routing" domains. The former
881 are used to qualify single-label names, the latter are used purely
882 for routing lookups within certain domains to specific links.
883 resolved now also synthesizes RRs for all entries from /etc/hosts.
885 * The systemd-resolve tool (which is a client utility for
886 systemd-resolved) has been improved considerably and is now fully
887 supported and documented. Hence it has moved from /usr/lib/systemd to
890 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
893 * The coredump collection logic has been reworked: when a coredump is
894 collected it is now written to disk, compressed and processed
895 (including stacktrace extraction) from a new instantiated service
896 systemd-coredump@.service, instead of directly from the
897 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
898 processing large coredumps can take up a substantial amount of
899 resources and time, and this previously happened entirely outside of
900 systemd's service supervision. With the new logic the core_pattern
901 hook only does minimal metadata collection before passing off control
902 to the new instantiated service, which is configured with a time
903 limit, a nice level and other settings to minimize negative impact on
904 the rest of the system. Also note that the new logic will honour the
905 RLIMIT_CORE setting of the crashed process, which now allows users
906 and processes to turn off coredumping for their processes by setting
909 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
910 and all forked processes by default. Previously, PID 1 would leave
911 the setting at "0" for all processes, as set by the kernel. Note that
912 the resource limit traditionally has no effect on the generated
913 coredumps on the system if the /proc/sys/kernel/core_pattern hook
914 logic is used. Since the limit is now honoured (see above) its
915 default has been changed so that the coredumping logic is enabled by
916 default for all processes, while allowing specific opt-out.
918 * When the stacktrace is extracted from processes of system users, this
919 is now done as "systemd-coredump" user, in order to sandbox this
920 potentially security sensitive parsing operation. (Note that when
921 processing coredumps of normal users this is done under the user ID
922 of process that crashed, as before.) Packagers should take notice
923 that it is now necessary to create the "systemd-coredump" system user
924 and group at package installation time.
926 * The systemd-activate socket activation testing tool gained support
927 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
928 and --seqpacket switches. It also has been extended to support both
929 new-style and inetd-style file descriptor passing. Use the new
930 --inetd switch to request inetd-style file descriptor passing.
932 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
933 variable, which takes a boolean value. If set to false, ANSI color
934 output is disabled in the tools even when run on a terminal that
937 * The VXLAN support in networkd now supports two new settings
938 DestinationPort= and PortRange=.
940 * A new systemd.machine_id= kernel command line switch has been added,
941 that may be used to set the machine ID in /etc/machine-id if it is
942 not initialized yet. This command line option has no effect if the
943 file is already initialized.
945 * systemd-nspawn gained a new --as-pid2 switch that invokes any
946 specified command line as PID 2 rather than PID 1 in the
947 container. In this mode PID 1 is a minimal stub init process that
948 implements the special POSIX and Linux semantics of PID 1 regarding
949 signal and child process management. Note that this stub init process
950 is implemented in nspawn itself and requires no support from the
951 container image. This new logic is useful to support running
952 arbitrary commands in the container, as normal processes are
953 generally not prepared to run as PID 1.
955 * systemd-nspawn gained a new --chdir= switch for setting the current
956 working directory for the process started in the container.
958 * "journalctl /dev/sda" will now output all kernel log messages for
959 specified device from the current boot, in addition to all devices
960 that are parents of it. This should make log output about devices
961 pretty useful, as long as kernel drivers attach enough metadata to
962 the log messages. (The usual SATA drivers do.)
964 * The sd-journal API gained two new calls
965 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
966 that report whether log data from /run or /var has been found.
968 * journalctl gained a new switch "--fields" that prints all journal
969 record field names currently in use in the journal. This is backed
970 by two new sd-journal API calls sd_journal_enumerate_fields() and
971 sd_journal_restart_fields().
973 * Most configurable timeouts in systemd now expect an argument of
974 "infinity" to turn them off, instead of "0" as before. The semantics
975 from now on is that a timeout of "0" means "now", and "infinity"
976 means "never". To maintain backwards compatibility, "0" continues to
977 turn off previously existing timeout settings.
979 * "systemctl reload-or-try-restart" has been renamed to "systemctl
980 try-reload-or-restart" to clarify what it actually does: the "try"
981 logic applies to both reloading and restarting, not just restarting.
982 The old name continues to be accepted for compatibility.
984 * On boot-up, when PID 1 detects that the system clock is behind the
985 release date of the systemd version in use, the clock is now set
986 to the latter. Previously, this was already done in timesyncd, in order
987 to avoid running with clocks set to the various clock epochs such as
988 1902, 1938 or 1970. With this change the logic is now done in PID 1
989 in addition to timesyncd during early boot-up, so that it is enforced
990 before the first process is spawned by systemd. Note that the logic
991 in timesyncd remains, as it is more comprehensive and ensures
992 clock monotonicity by maintaining a persistent timestamp file in
993 /var. Since /var is generally not available in earliest boot or the
994 initrd, this part of the logic remains in timesyncd, and is not done
997 * Support for tweaking details in net_cls.class_id through the
998 NetClass= configuration directive has been removed, as the kernel
999 people have decided to deprecate that controller in cgroup v2.
1000 Userspace tools such as nftables are moving over to setting rules
1001 that are specific to the full cgroup path of a task, which obsoletes
1002 these controllers anyway. The NetClass= directive is kept around for
1003 legacy compatibility reasons. For a more in-depth description of the
1004 kernel change, please refer to the respective upstream commit:
1006 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
1008 * A new service setting RuntimeMaxSec= has been added that may be used
1009 to specify a maximum runtime for a service. If the timeout is hit, the
1010 service is terminated and put into a failure state.
1012 * A new service setting AmbientCapabilities= has been added. It allows
1013 configuration of additional Linux process capabilities that are
1014 passed to the activated processes. This is only available on very
1017 * The process resource limit settings in service units may now be used
1018 to configure hard and soft limits individually.
1020 * The various libsystemd APIs such as sd-bus or sd-event now publicly
1021 expose support for gcc's __attribute__((cleanup())) C extension.
1022 Specifically, for many object destructor functions alternative
1023 versions have been added that have names suffixed with "p" and take a
1024 pointer to a pointer to the object to destroy, instead of just a
1025 pointer to the object itself. This is useful because these destructor
1026 functions may be used directly as parameters to the cleanup
1027 construct. Internally, systemd has been a heavy user of this GCC
1028 extension for a long time, and with this change similar support is
1029 now available to consumers of the library outside of systemd. Note
1030 that by using this extension in your sources compatibility with old
1031 and strictly ANSI compatible C compilers is lost. However, all gcc or
1032 LLVM versions of recent years support this extension.
1034 * Timer units gained support for a new setting RandomizedDelaySec= that
1035 allows configuring some additional randomized delay to the configured
1036 time. This is useful to spread out timer events to avoid load peaks in
1037 clusters or larger setups.
1039 * Calendar time specifications now support sub-second accuracy.
1041 * Socket units now support listening on SCTP and UDP-lite protocol
1044 * The sd-event API now comes with a full set of man pages.
1046 * Older versions of systemd contained experimental support for
1047 compressing journal files and coredumps with the LZ4 compressor that
1048 was not compatible with the lz4 binary (due to API limitations of the
1049 lz4 library). This support has been removed; only support for files
1050 compatible with the lz4 binary remains. This LZ4 logic is now
1051 officially supported and no longer considered experimental.
1053 * The dkr image import logic has been removed again from importd. dkr's
1054 micro-services focus doesn't fit into the machine image focus of
1055 importd, and quickly got out of date with the upstream dkr API.
1057 * Creation of the /run/lock/lockdev/ directory was dropped from
1058 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
1059 been available for many years. If you still need this, you need to
1060 create your own tmpfiles.d config file with:
1062 d /run/lock/lockdev 0775 root lock -
1064 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
1065 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
1066 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
1067 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
1068 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
1069 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
1070 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
1071 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
1072 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
1073 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
1074 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
1075 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
1076 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
1077 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
1078 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
1079 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
1080 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
1081 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
1082 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
1084 — Berlin, 2016-02-11
1088 * A number of properties previously only settable in unit
1089 files are now also available as properties to set when
1090 creating transient units programmatically via the bus, as it
1091 is exposed with systemd-run's --property=
1092 setting. Specifically, these are: SyslogIdentifier=,
1093 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
1094 EnvironmentFile=, ReadWriteDirectories=,
1095 ReadOnlyDirectories=, InaccessibleDirectories=,
1096 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
1098 * When creating transient services via the bus API it is now
1099 possible to pass in a set of file descriptors to use as
1100 STDIN/STDOUT/STDERR for the invoked process.
1102 * Slice units may now be created transiently via the bus APIs,
1103 similar to the way service and scope units may already be
1104 created transiently.
1106 * Wherever systemd expects a calendar timestamp specification
1107 (like in journalctl's --since= and --until= switches) UTC
1108 timestamps are now supported. Timestamps suffixed with "UTC"
1109 are now considered to be in Universal Time Coordinated
1110 instead of the local timezone. Also, timestamps may now
1111 optionally be specified with sub-second accuracy. Both of
1112 these additions also apply to recurring calendar event
1113 specification, such as OnCalendar= in timer units.
1115 * journalctl gained a new "--sync" switch that asks the
1116 journal daemon to write all so far unwritten log messages to
1117 disk and sync the files, before returning.
1119 * systemd-tmpfiles learned two new line types "q" and "Q" that
1120 operate like "v", but also set up a basic btrfs quota
1121 hierarchy when used on a btrfs file system with quota
1124 * tmpfiles' "v", "q" and "Q" will now create a plain directory
1125 instead of a subvolume (even on a btrfs file system) if the
1126 root directory is a plain directory, and not a
1127 subvolume. This should simplify things with certain chroot()
1128 environments which are not aware of the concept of btrfs
1131 * systemd-detect-virt gained a new --chroot switch to detect
1132 whether execution takes place in a chroot() environment.
1134 * CPUAffinity= now takes CPU index ranges in addition to
1137 * The various memory-related resource limit settings (such as
1138 LimitAS=) now understand the usual K, M, G, ... suffixes to
1139 the base of 1024 (IEC). Similar, the time-related resource
1140 limit settings understand the usual min, h, day, ...
1143 * There's a new system.conf setting DefaultTasksMax= to
1144 control the default TasksMax= setting for services and
1145 scopes running on the system. (TasksMax= is the primary
1146 setting that exposes the "pids" cgroup controller on systemd
1147 and was introduced in the previous systemd release.) The
1148 setting now defaults to 512, which means services that are
1149 not explicitly configured otherwise will only be able to
1150 create 512 processes or threads at maximum, from this
1151 version on. Note that this means that thread- or
1152 process-heavy services might need to be reconfigured to set
1153 TasksMax= to a higher value. It is sufficient to set
1154 TasksMax= in these specific unit files to a higher value, or
1155 even "infinity". Similar, there's now a logind.conf setting
1156 UserTasksMax= that defaults to 4096 and limits the total
1157 number of processes or tasks each user may own
1158 concurrently. nspawn containers also have the TasksMax=
1159 value set by default now, to 8192. Note that all of this
1160 only has an effect if the "pids" cgroup controller is
1161 enabled in the kernel. The general benefit of these changes
1162 should be a more robust and safer system, that provides a
1163 certain amount of per-service fork() bomb protection.
1165 * systemd-nspawn gained the new --network-veth-extra= switch
1166 to define additional and arbitrarily-named virtual Ethernet
1167 links between the host and the container.
1169 * A new service execution setting PassEnvironment= has been
1170 added that allows importing select environment variables
1171 from PID1's environment block into the environment block of
1174 * Timer units gained support for a new RemainAfterElapse=
1175 setting which takes a boolean argument. It defaults to on,
1176 exposing behaviour unchanged to previous releases. If set to
1177 off, timer units are unloaded after they elapsed if they
1178 cannot elapse again. This is particularly useful for
1179 transient timer units, which shall not stay around longer
1180 than until they first elapse.
1182 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
1183 default now (the kernel default is 16). This is beneficial
1184 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
1185 allows substantially larger numbers of queued
1186 datagrams. This should increase the capability of systemd to
1187 parallelize boot-up, as logging and sd_notify() are unlikely
1188 to stall execution anymore. If you need to change the value
1189 from the new defaults, use the usual sysctl.d/ snippets.
1191 * The compression framing format used by the journal or
1192 coredump processing has changed to be in line with what the
1193 official LZ4 tools generate. LZ4 compression support in
1194 systemd was considered unsupported previously, as the format
1195 was not compatible with the normal tools. With this release
1196 this has changed now, and it is hence safe for downstream
1197 distributions to turn it on. While not compressing as well
1198 as the XZ, LZ4 is substantially faster, which makes
1199 it a good default choice for the compression logic in the
1200 journal and in coredump handling.
1202 * Any reference to /etc/mtab has been dropped from
1203 systemd. The file has been obsolete since a while, but
1204 systemd refused to work on systems where it was incorrectly
1205 set up (it should be a symlink or non-existent). Please make
1206 sure to update to util-linux 2.27.1 or newer in conjunction
1207 with this systemd release, which also drops any reference to
1208 /etc/mtab. If you maintain a distribution make sure that no
1209 software you package still references it, as this is a
1210 likely source of bugs. There's also a glibc bug pending,
1211 asking for removal of any reference to this obsolete file:
1213 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
1215 Note that only util-linux versions built with
1216 --enable-libmount-force-mountinfo are supported.
1218 * Support for the ".snapshot" unit type has been removed. This
1219 feature turned out to be little useful and little used, and
1220 has now been removed from the core and from systemctl.
1222 * The dependency types RequiresOverridable= and
1223 RequisiteOverridable= have been removed from systemd. They
1224 have been used only very sparingly to our knowledge and
1225 other options that provide a similar effect (such as
1226 systemctl --mode=ignore-dependencies) are much more useful
1227 and commonly used. Moreover, they were only half-way
1228 implemented as the option to control behaviour regarding
1229 these dependencies was never added to systemctl. By removing
1230 these dependency types the execution engine becomes a bit
1231 simpler. Unit files that use these dependencies should be
1232 changed to use the non-Overridable dependency types
1233 instead. In fact, when parsing unit files with these
1234 options, that's what systemd will automatically convert them
1235 too, but it will also warn, asking users to fix the unit
1236 files accordingly. Removal of these dependency types should
1237 only affect a negligible number of unit files in the wild.
1239 * Behaviour of networkd's IPForward= option changed
1240 (again). It will no longer maintain a per-interface setting,
1241 but propagate one way from interfaces where this is enabled
1242 to the global kernel setting. The global setting will be
1243 enabled when requested by a network that is set up, but
1244 never be disabled again. This change was made to make sure
1245 IPv4 and IPv6 behaviour regarding packet forwarding is
1246 similar (as the Linux IPv6 stack does not support
1247 per-interface control of this setting) and to minimize
1250 * In unit files the behaviour of %u, %U, %h, %s has
1251 changed. These specifiers will now unconditionally resolve
1252 to the various user database fields of the user that the
1253 systemd instance is running as, instead of the user
1254 configured in the specific unit via User=. Note that this
1255 effectively doesn't change much, as resolving of these
1256 specifiers was already turned off in the --system instance
1257 of systemd, as we cannot do NSS lookups from PID 1. In the
1258 --user instance of systemd these specifiers where correctly
1259 resolved, but hardly made any sense, since the user instance
1260 lacks privileges to do user switches anyway, and User= is
1261 hence useless. Moreover, even in the --user instance of
1262 systemd behaviour was awkward as it would only take settings
1263 from User= assignment placed before the specifier into
1264 account. In order to unify and simplify the logic around
1265 this the specifiers will now always resolve to the
1266 credentials of the user invoking the manager (which in case
1267 of PID 1 is the root user).
1269 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
1270 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
1271 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
1272 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
1273 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
1274 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
1275 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
1276 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
1277 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
1278 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
1279 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
1280 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
1281 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
1282 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
1285 — Berlin, 2015-11-18
1289 * systemd now depends on util-linux v2.27. More specifically,
1290 the newly added mount monitor feature in libmount now
1291 replaces systemd's former own implementation.
1293 * libmount mandates /etc/mtab not to be regular file, and
1294 systemd now enforces this condition at early boot.
1295 /etc/mtab has been deprecated and warned about for a very
1296 long time, so systems running systemd should already have
1297 stopped having this file around as anything else than a
1298 symlink to /proc/self/mounts.
1300 * Support for the "pids" cgroup controller has been added. It
1301 allows accounting the number of tasks in a cgroup and
1302 enforcing limits on it. This adds two new setting
1303 TasksAccounting= and TasksMax= to each unit, as well as a
1304 global option DefaultTasksAccounting=.
1306 * Support for the "net_cls" cgroup controller has been added.
1307 It allows assigning a net class ID to each task in the
1308 cgroup, which can then be used in firewall rules and traffic
1309 shaping configurations. Note that the kernel netfilter net
1310 class code does not currently work reliably for ingress
1311 packets on unestablished sockets.
1313 This adds a new config directive called NetClass= to CGroup
1314 enabled units. Allowed values are positive numbers for fixed
1315 assignments and "auto" for picking a free value
1318 * 'systemctl is-system-running' now returns 'offline' if the
1319 system is not booted with systemd. This command can now be
1320 used as a substitute for 'systemd-notify --booted'.
1322 * Watchdog timeouts have been increased to 3 minutes for all
1323 in-tree service files. Apparently, disk IO issues are more
1324 frequent than we hoped, and user reported >1 minute waiting
1327 * 'machine-id-commit' functionality has been merged into
1328 'machine-id-setup --commit'. The separate binary has been
1331 * The WorkingDirectory= directive in unit files may now be set
1332 to the special value '~'. In this case, the working
1333 directory is set to the home directory of the user
1334 configured in User=.
1336 * "machinectl shell" will now open the shell in the home
1337 directory of the selected user by default.
1339 * The CrashChVT= configuration file setting is renamed to
1340 CrashChangeVT=, following our usual logic of not
1341 abbreviating unnecessarily. The old directive is still
1342 supported for compat reasons. Also, this directive now takes
1343 an integer value between 1 and 63, or a boolean value. The
1344 formerly supported '-1' value for disabling stays around for
1347 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
1348 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
1349 RootDirectory= properties can now be set for transient
1352 * The systemd-analyze tool gained a new "set-log-target" verb
1353 to change the logging target the system manager logs to
1354 dynamically during runtime. This is similar to how
1355 "systemd-analyze set-log-level" already changes the log
1358 * In nspawn /sys is now mounted as tmpfs, with only a selected
1359 set of subdirectories mounted in from the real sysfs. This
1360 enhances security slightly, and is useful for ensuring user
1361 namespaces work correctly.
1363 * Support for USB FunctionFS activation has been added. This
1364 allows implementation of USB gadget services that are
1365 activated as soon as they are requested, so that they don't
1366 have to run continuously, similar to classic socket
1369 * The "systemctl exit" command now optionally takes an
1370 additional parameter that sets the exit code to return from
1371 the systemd manager when exiting. This is only relevant when
1372 running the systemd user instance, or when running the
1373 system instance in a container.
1375 * sd-bus gained the new API calls sd_bus_path_encode_many()
1376 and sd_bus_path_decode_many() that allow easy encoding and
1377 decoding of multiple identifier strings inside a D-Bus
1378 object path. Another new call sd_bus_default_flush_close()
1379 has been added to flush and close per-thread default
1382 * systemd-cgtop gained support for a -M/--machine= switch to
1383 show the control groups within a certain container only.
1385 * "systemctl kill" gained support for an optional --fail
1386 switch. If specified the requested operation will fail of no
1387 processes have been killed, because the unit had no
1388 processes attached, or similar.
1390 * A new systemd.crash_reboot=1 kernel command line option has
1391 been added that triggers a reboot after crashing. This can
1392 also be set through CrashReboot= in systemd.conf.
1394 * The RuntimeDirectory= setting now understands unit
1395 specifiers like %i or %f.
1397 * A new (still internal) library API sd-ipv4acd has been added,
1398 that implements address conflict detection for IPv4. It's
1399 based on code from sd-ipv4ll, and will be useful for
1400 detecting DHCP address conflicts.
1402 * File descriptors passed during socket activation may now be
1403 named. A new API sd_listen_fds_with_names() is added to
1404 access the names. The default names may be overridden,
1405 either in the .socket file using the FileDescriptorName=
1406 parameter, or by passing FDNAME= when storing the file
1407 descriptors using sd_notify().
1409 * systemd-networkd gained support for:
1411 - Setting the IPv6 Router Advertisement settings via
1412 IPv6AcceptRouterAdvertisements= in .network files.
1414 - Configuring the HelloTimeSec=, MaxAgeSec= and
1415 ForwardDelaySec= bridge parameters in .netdev files.
1417 - Configuring PreferredSource= for static routes in
1420 * The "ask-password" framework used to query for LUKS harddisk
1421 passwords or SSL passwords during boot gained support for
1422 caching passwords in the kernel keyring, if it is
1423 available. This makes sure that the user only has to type in
1424 a passphrase once if there are multiple objects to unlock
1425 with the same one. Previously, such password caching was
1426 available only when Plymouth was used; this moves the
1427 caching logic into the systemd codebase itself. The
1428 "systemd-ask-password" utility gained a new --keyname=
1429 switch to control which kernel keyring key to use for
1430 caching a password in. This functionality is also useful for
1431 enabling display managers such as gdm to automatically
1432 unlock the user's GNOME keyring if its passphrase, the
1433 user's password and the harddisk password are the same, if
1434 gdm-autologin is used.
1436 * When downloading tar or raw images using "machinectl
1437 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
1438 file is now also downloaded, if it is available and stored
1439 next to the image file.
1441 * Units of type ".socket" gained a new boolean setting
1442 Writable= which is only useful in conjunction with
1443 ListenSpecial=. If true, enables opening the specified
1444 special file in O_RDWR mode rather than O_RDONLY mode.
1446 * systemd-rfkill has been reworked to become a singleton
1447 service that is activated through /dev/rfkill on each rfkill
1448 state change and saves the settings to disk. This way,
1449 systemd-rfkill is now compatible with devices that exist
1450 only intermittendly, and even restores state if the previous
1451 system shutdown was abrupt rather than clean.
1453 * The journal daemon gained support for vacuuming old journal
1454 files controlled by the number of files that shall remain,
1455 in addition to the already existing control by size and by
1456 date. This is useful as journal interleaving performance
1457 degrades with too many separate journal files, and allows
1458 putting an effective limit on them. The new setting defaults
1459 to 100, but this may be changed by setting SystemMaxFiles=
1460 and RuntimeMaxFiles= in journald.conf. Also, the
1461 "journalctl" tool gained the new --vacuum-files= switch to
1462 manually vacuum journal files to leave only the specified
1463 number of files in place.
1465 * udev will now create /dev/disk/by-path links for ATA devices
1466 on kernels where that is supported.
1468 * Galician, Serbian, Turkish and Korean translations were added.
1470 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
1471 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
1472 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
1473 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
1474 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
1475 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
1476 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
1477 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
1478 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
1479 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
1480 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
1481 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
1482 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
1483 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
1484 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
1485 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
1486 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
1487 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
1489 — Berlin, 2015-10-07
1493 * The DHCP implementation of systemd-networkd gained a set of
1496 - The DHCP server now supports emitting DNS and NTP
1497 information. It may be enabled and configured via
1498 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
1499 and NTP information is enabled, but no servers are
1500 configured, the corresponding uplink information (if there
1501 is any) is propagated.
1503 - Server and client now support transmission and reception
1504 of timezone information. It can be configured via the
1505 newly introduced network options UseTimezone=,
1506 EmitTimezone=, and Timezone=. Transmission of timezone
1507 information is enabled between host and containers by
1508 default now: the container will change its local timezone
1509 to what the host has set.
1511 - Lease timeouts can now be configured via
1512 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
1514 - The DHCP server improved on the stability of
1515 leases. Clients are more likely to get the same lease
1516 information back, even if the server loses state.
1518 - The DHCP server supports two new configuration options to
1519 control the lease address pool metrics, PoolOffset= and
1522 * The encapsulation limit of tunnels in systemd-networkd may
1523 now be configured via 'EncapsulationLimit='. It allows
1524 modifying the maximum additional levels of encapsulation
1525 that are permitted to be prepended to a packet.
1527 * systemd now supports the concept of user buses replacing
1528 session buses, if used with dbus-1.10 (and enabled via dbus
1529 --enable-user-session). It previously only supported this on
1530 kdbus-enabled systems, and this release expands this to
1531 'dbus-daemon' systems.
1533 * systemd-networkd now supports predictable interface names
1536 * systemd now optionally supports the new Linux kernel
1537 "unified" control group hierarchy. If enabled via the kernel
1538 command-line option 'systemd.unified_cgroup_hierarchy=1',
1539 systemd will try to mount the unified cgroup hierarchy
1540 directly on /sys/fs/cgroup. If not enabled, or not
1541 available, systemd will fall back to the legacy cgroup
1542 hierarchy setup, as before. Host system and containers can
1543 mix and match legacy and unified hierarchies as they
1544 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
1545 environment variable to individually select the hierarchy to
1546 use for executed containers. By default, nspawn will use the
1547 unified hierarchy for the containers if the host uses the
1548 unified hierarchy, and the legacy hierarchy otherwise.
1549 Please note that at this point the unified hierarchy is an
1550 experimental kernel feature and is likely to change in one
1551 of the next kernel releases. Therefore, it should not be
1552 enabled by default in downstream distributions yet. The
1553 minimum required kernel version for the unified hierarchy to
1554 work is 4.2. Note that when the unified hierarchy is used
1555 for the first time delegated access to controllers is
1556 safe. Because of this systemd-nspawn containers will get
1557 access to controllers now, as will systemd user
1558 sessions. This means containers and user sessions may now
1559 manage their own resources, partitioning up what the system
1562 * A new special scope unit "init.scope" has been introduced
1563 that encapsulates PID 1 of the system. It may be used to
1564 determine resource usage and enforce resource limits on PID
1565 1 itself. PID 1 hence moved out of the root of the control
1568 * The cgtop tool gained support for filtering out kernel
1569 threads when counting tasks in a control group. Also, the
1570 count of processes is now recursively summed up by
1571 default. Two options -k and --recursive= have been added to
1572 revert to old behaviour. The tool has also been updated to
1573 work correctly in containers now.
1575 * systemd-nspawn's --bind= and --bind-ro= options have been
1576 extended to allow creation of non-recursive bind mounts.
1578 * libsystemd gained two new calls sd_pid_get_cgroup() and
1579 sd_peer_get_cgroup() which return the control group path of
1580 a process or peer of a connected AF_UNIX socket. This
1581 function call is particularly useful when implementing
1582 delegated subtrees support in the control group hierarchy.
1584 * The "sd-event" event loop API of libsystemd now supports
1585 correct dequeuing of real-time signals, without losing
1588 * When systemd requests a PolicyKit decision when managing
1589 units it will now add additional fields to the request,
1590 including unit name and desired operation. This enables more
1591 powerful PolicyKit policies, that make decisions depending
1592 on these parameters.
1594 * nspawn learnt support for .nspawn settings files, that may
1595 accompany the image files or directories of containers, and
1596 may contain additional settings for the container. This is
1597 an alternative to configuring container parameters via the
1598 nspawn command line.
1600 Contributions from: Cristian Rodríguez, Daniel Mack, David
1601 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
1602 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
1603 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
1604 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
1605 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
1606 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
1607 Andersen, Tom Gundersen, Torstein Husebø
1609 — Berlin, 2015-09-08
1613 * machinectl gained a new verb 'shell' which opens a fresh
1614 shell on the target container or the host. It is similar to
1615 the existing 'login' command of machinectl, but spawns the
1616 shell directly without prompting for username or
1617 password. The pseudo machine '.host' now refers to the local
1618 host and is used by default. Hence, 'machinectl shell' can
1619 be used as replacement for 'su -' which spawns a session as
1620 a fresh systemd unit in a way that is fully isolated from
1621 the originating session.
1623 * systemd-networkd learned to cope with private-zone DHCP
1624 options and allows other programs to query the values.
1626 * SELinux access control when enabling/disabling units is no
1627 longer enforced with this release. The previous
1628 implementation was incorrect, and a new corrected
1629 implementation is not yet available. As unit file operations
1630 are still protected via PolicyKit and D-Bus policy this is
1631 not a security problem. Yet, distributions which care about
1632 optimal SELinux support should probably not stabilize on
1635 * sd-bus gained support for matches of type "arg0has=", that
1636 test for membership of strings in string arrays sent in bus
1639 * systemd-resolved now dumps the contents of its DNS and LLMNR
1640 caches to the logs on reception of the SIGUSR1 signal. This
1641 is useful to debug DNS behaviour.
1643 * The coredumpctl tool gained a new --directory= option to
1644 operate on journal files in a specific directory.
1646 * "systemctl reboot" and related commands gained a new
1647 "--message=" option which may be used to set a free-text
1648 wall message when shutting down or rebooting the
1649 system. This message is also logged, which is useful for
1650 figuring out the reason for a reboot or shutdown a
1653 * The "systemd-resolve-host" tool's -i switch now takes
1654 network interface numbers as alternative to interface names.
1656 * A new unit file setting for services has been introduced:
1657 UtmpMode= allows configuration of how precisely systemd
1658 handles utmp and wtmp entries for the service if this is
1659 enabled. This allows writing services that appear similar to
1660 user sessions in the output of the "w", "who", "last" and
1663 * systemd-resolved will now locally synthesize DNS resource
1664 records for the "localhost" and "gateway" domains as well as
1665 the local hostname. This should ensure that clients querying
1666 RRs via resolved will get similar results as those going via
1667 NSS, if nss-myhostname is enabled.
1669 Contributions from: Alastair Hughes, Alex Crawford, Daniel
1670 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
1671 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
1672 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
1673 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
1674 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
1675 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
1676 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
1677 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
1678 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
1679 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
1680 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
1682 — Berlin, 2015-08-27
1686 * The systemd-efi-boot-generator functionality was merged into
1687 systemd-gpt-auto-generator.
1689 * systemd-networkd now supports Group Policy for vxlan
1690 devices. It can be enabled via the new boolean configuration
1691 option called 'GroupPolicyExtension='.
1693 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
1694 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
1695 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
1697 — Berlin, 2015-07-31
1701 * The python-systemd code has been removed from the systemd repository.
1702 A new repository has been created which accommodates the code from
1703 now on, and we kindly ask distributions to create a separate package
1704 for this: https://github.com/systemd/python-systemd
1706 * The systemd daemon will now reload its main configuration
1707 (/etc/systemd/system.conf) on daemon-reload.
1709 * sd-dhcp now exposes vendor specific extensions via
1710 sd_dhcp_lease_get_vendor_specific().
1712 * systemd-networkd gained a number of new configuration options.
1714 - A new boolean configuration option for TAP devices called
1715 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
1716 device, thus allowing to send and receive GSO packets.
1718 - A new tunnel configuration option called 'CopyDSCP='.
1719 If enabled, the DSCP field of ip6 tunnels is copied into the
1720 decapsulated packet.
1722 - A set of boolean bridge configuration options were added.
1723 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
1724 and 'UnicastFlood=' are now parsed by networkd and applied to the
1725 respective bridge link device via the respective IFLA_BRPORT_*
1728 - A new string configuration option to override the hostname sent
1729 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
1730 is true, networkd will use the configured hostname instead of the
1731 system hostname when sending DHCP requests.
1733 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
1734 networkd will configure the IPv6 flow-label of the tunnel device
1735 according to RFC2460.
1737 - The 'macvtap' virtual network devices are now supported, similar to
1738 the already supported 'macvlan' devices.
1740 * systemd-resolved now implements RFC5452 to improve resilience against
1741 cache poisoning. Additionally, source port randomization is enabled
1742 by default to further protect against DNS spoofing attacks.
1744 * nss-mymachines now supports translating UIDs and GIDs of running
1745 containers with user-namespaces enabled. If a container 'foo'
1746 translates a host uid 'UID' to the container uid 'TUID', then
1747 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
1748 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
1749 mapped as 'vg-foo-TGID'.
1751 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
1752 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
1753 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
1754 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
1755 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
1756 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
1757 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
1758 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
1759 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
1760 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
1762 — Berlin, 2015-07-29
1766 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
1767 There are no known issues with current sysfs, and udev does not need
1768 or should be used to work around such bugs.
1770 * udev does no longer enable USB HID power management. Several reports
1771 indicate, that some devices cannot handle that setting.
1773 * The udev accelerometer helper was removed. The functionality
1774 is now fully included in iio-sensor-proxy. But this means,
1775 older iio-sensor-proxy versions will no longer provide
1776 accelerometer/orientation data with this systemd version.
1777 Please upgrade iio-sensor-proxy to version 1.0.
1779 * networkd gained a new configuration option IPv6PrivacyExtensions=
1780 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
1781 for Stateless Address") on selected networks.
1783 * For the sake of fewer build-time dependencies and less code in the
1784 main repository, the python bindings are about to be removed in the
1785 next release. A new repository has been created which accommodates
1786 the code from now on, and we kindly ask distributions to create a
1787 separate package for this. The removal will take place in v223.
1789 https://github.com/systemd/python-systemd
1791 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
1792 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
1793 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
1794 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
1795 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
1796 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
1797 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
1798 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
1799 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
1800 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
1802 — Berlin, 2015-07-07
1806 * The sd-bus.h and sd-event.h APIs have now been declared
1807 stable and have been added to the official interface of
1808 libsystemd.so. sd-bus implements an alternative D-Bus client
1809 library, that is relatively easy to use, very efficient and
1810 supports both classic D-Bus as well as kdbus as transport
1811 backend. sd-event is a generic event loop abstraction that
1812 is built around Linux epoll, but adds features such as event
1813 prioritization or efficient timer handling. Both APIs are good
1814 choices for C programs looking for a bus and/or event loop
1815 implementation that is minimal and does not have to be
1816 portable to other kernels.
1818 * kdbus support is no longer compile-time optional. It is now
1819 always built-in. However, it can still be disabled at
1820 runtime using the kdbus=0 kernel command line setting, and
1821 that setting may be changed to default to off, by specifying
1822 --disable-kdbus at build-time. Note though that the kernel
1823 command line setting has no effect if the kdbus.ko kernel
1824 module is not installed, in which case kdbus is (obviously)
1825 also disabled. We encourage all downstream distributions to
1826 begin testing kdbus by adding it to the kernel images in the
1827 development distributions, and leaving kdbus support in
1830 * The minimal required util-linux version has been bumped to
1833 * Support for chkconfig (--enable-chkconfig) was removed in
1834 favor of calling an abstraction tool
1835 /lib/systemd/systemd-sysv-install. This needs to be
1836 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
1837 in README for details.
1839 * If there's a systemd unit and a SysV init script for the
1840 same service name, and the user executes "systemctl enable"
1841 for it (or a related call), then this will now enable both
1842 (or execute the related operation on both), not just the
1845 * The libudev API documentation has been converted from gtkdoc
1848 * gudev has been removed from the systemd tree, it is now an
1851 * The systemd-cgtop tool learnt a new --raw switch to generate
1852 "raw" (machine parsable) output.
1854 * networkd's IPForwarding= .network file setting learnt the
1855 new setting "kernel", which ensures that networkd does not
1856 change the IP forwarding sysctl from the default kernel
1859 * The systemd-logind bus API now exposes a new boolean
1860 property "Docked" that reports whether logind considers the
1861 system "docked", i.e. connected to a docking station or not.
1863 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
1864 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
1865 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
1866 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
1867 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
1868 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
1869 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
1870 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
1871 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
1872 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
1873 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
1874 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
1875 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
1876 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
1877 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
1878 Fink, Zbigniew Jędrzejewski-Szmek
1880 — Berlin, 2015-06-19
1884 * The gudev library has been extracted into a separate repository
1885 available at: https://git.gnome.org/browse/libgudev/
1886 It is now managed as part of the Gnome project. Distributions
1887 are recommended to pass --disable-gudev to systemd and use
1888 gudev from the Gnome project instead. gudev is still included
1889 in systemd, for now. It will be removed soon, though. Please
1890 also see the announcement-thread on systemd-devel:
1891 http://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
1893 * systemd now exposes a CPUUsageNSec= property for each
1894 service unit on the bus, that contains the overall consumed
1895 CPU time of a service (the sum of what each process of the
1896 service consumed). This value is only available if
1897 CPUAccounting= is turned on for a service, and is then shown
1898 in the "systemctl status" output.
1900 * Support for configuring alternative mappings of the old SysV
1901 runlevels to systemd targets has been removed. They are now
1902 hardcoded in a way that runlevels 2, 3, 4 all map to
1903 multi-user.target and 5 to graphical.target (which
1904 previously was already the default behaviour).
1906 * The auto-mounter logic gained support for mount point
1907 expiry, using a new TimeoutIdleSec= setting in .automount
1908 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
1910 * The EFI System Partition (ESP) as mounted to /boot by
1911 systemd-efi-boot-generator will now be unmounted
1912 automatically after 2 minutes of not being used. This should
1913 minimize the risk of ESP corruptions.
1915 * New /etc/fstab options x-systemd.requires= and
1916 x-systemd.requires-mounts-for= are now supported to express
1917 additional dependencies for mounts. This is useful for
1918 journalling file systems that support external journal
1919 devices or overlay file systems that require underlying file
1920 systems to be mounted.
1922 * systemd does not support direct live-upgrades (via systemctl
1923 daemon-reexec) from versions older than v44 anymore. As no
1924 distribution we are aware of shipped such old versions in a
1925 stable release this should not be problematic.
1927 * When systemd forks off a new per-connection service instance
1928 it will now set the $REMOTE_ADDR environment variable to the
1929 remote IP address, and $REMOTE_PORT environment variable to
1930 the remote IP port. This behaviour is similar to the
1931 corresponding environment variables defined by CGI.
1933 * systemd-networkd gained support for uplink failure
1934 detection. The BindCarrier= option allows binding interface
1935 configuration dynamically to the link sense of other
1936 interfaces. This is useful to achieve behaviour like in
1939 * systemd-networkd gained support for configuring the DHCP
1940 client identifier to use when requesting leases.
1942 * systemd-networkd now has a per-network UseNTP= option to
1943 configure whether NTP server information acquired via DHCP
1944 is passed on to services like systemd-timesyncd.
1946 * systemd-networkd gained support for vti6 tunnels.
1948 * Note that systemd-networkd manages the sysctl variable
1949 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
1950 it is configured for since v219. The variable controls IP
1951 forwarding, and is a per-interface alternative to the global
1952 /proc/sys/net/ipv[46]/ip_forward. This setting is
1953 configurable in the IPForward= option, which defaults to
1954 "no". This means if networkd is used for an interface it is
1955 no longer sufficient to set the global sysctl option to turn
1956 on IP forwarding! Instead, the .network file option
1957 IPForward= needs to be turned on! Note that the
1958 implementation of this behaviour was broken in v219 and has
1961 * Many bonding and vxlan options are now configurable in
1964 * systemd-nspawn gained a new --property= setting to set unit
1965 properties for the container scope. This is useful for
1966 setting resource parameters (e.g. "CPUShares=500") on
1967 containers started from the command line.
1969 * systemd-nspawn gained a new --private-users= switch to make
1970 use of user namespacing available on recent Linux kernels.
1972 * systemd-nspawn may now be called as part of a shell pipeline
1973 in which case the pipes used for stdin and stdout are passed
1974 directly to the process invoked in the container, without
1975 indirection via a pseudo tty.
1977 * systemd-nspawn gained a new switch to control the UNIX
1978 signal to use when killing the init process of the container
1981 * systemd-nspawn gained a new --overlay= switch for mounting
1982 overlay file systems into the container using the new kernel
1985 * When a container image is imported via systemd-importd and
1986 the host file system is not btrfs, a loopback block device
1987 file is created in /var/lib/machines.raw with a btrfs file
1988 system inside. It is then mounted to /var/lib/machines to
1989 enable btrfs features for container management. The loopback
1990 file and btrfs file system is grown as needed when container
1991 images are imported via systemd-importd.
1993 * systemd-machined/systemd-importd gained support for btrfs
1994 quota, to enforce container disk space limits on disk. This
1995 is exposed in "machinectl set-limit".
1997 * systemd-importd now can import containers from local .tar,
1998 .raw and .qcow2 images, and export them to .tar and .raw. It
1999 can also import dkr v2 images now from the network (on top
2002 * systemd-importd gained support for verifying downloaded
2003 images with gpg2 (previously only gpg1 was supported).
2005 * systemd-machined, systemd-logind, systemd: most bus calls
2006 are now accessible to unprivileged processes via
2007 PolicyKit. Also, systemd-logind will now allow users to kill
2008 their own sessions without further privileges or
2011 * systemd-shutdownd has been removed. This service was
2012 previously responsible for implementing scheduled shutdowns
2013 as exposed in /usr/bin/shutdown's time parameter. This
2014 functionality has now been moved into systemd-logind and is
2015 accessible via a bus interface.
2017 * "systemctl reboot" gained a new switch --firmware-setup that
2018 can be used to reboot into the EFI firmware setup, if that
2019 is available. systemd-logind now exposes an API on the bus
2020 to trigger such reboots, in case graphical desktop UIs want
2021 to cover this functionality.
2023 * "systemctl enable", "systemctl disable" and "systemctl mask"
2024 now support a new "--now" switch. If specified the units
2025 that are enabled will also be started, and the ones
2026 disabled/masked also stopped.
2028 * The Gummiboot EFI boot loader tool has been merged into
2029 systemd, and renamed to "systemd-boot". The bootctl tool has been
2030 updated to support systemd-boot.
2032 * An EFI kernel stub has been added that may be used to create
2033 kernel EFI binaries that contain not only the actual kernel,
2034 but also an initrd, boot splash, command line and OS release
2035 information. This combined binary can then be signed as a
2036 single image, so that the firmware can verify it all in one
2037 step. systemd-boot has special support for EFI binaries created
2038 like this and can extract OS release information from them
2039 and show them in the boot menu. This functionality is useful
2040 to implement cryptographically verified boot schemes.
2042 * Optional support has been added to systemd-fsck to pass
2043 fsck's progress report to an AF_UNIX socket in the file
2046 * udev will no longer create device symlinks for all block
2047 devices by default. A blacklist for excluding special block
2048 devices from this logic has been turned into a whitelist
2049 that requires picking block devices explicitly that require
2052 * A new (currently still internal) API sd-device.h has been
2053 added to libsystemd. This modernized API is supposed to
2054 replace libudev eventually. In fact, already much of libudev
2055 is now just a wrapper around sd-device.h.
2057 * A new hwdb database for storing metadata about pointing
2058 stick devices has been added.
2060 * systemd-tmpfiles gained support for setting file attributes
2061 similar to the "chattr" tool with new 'h' and 'H' lines.
2063 * systemd-journald will no longer unconditionally set the
2064 btrfs NOCOW flag on new journal files. This is instead done
2065 with tmpfiles snippet using the new 'h' line type. This
2066 allows easy disabling of this logic, by masking the
2067 journal-nocow.conf tmpfiles file.
2069 * systemd-journald will now translate audit message types to
2070 human readable identifiers when writing them to the
2071 journal. This should improve readability of audit messages.
2073 * The LUKS logic gained support for the offset= and skip=
2074 options in /etc/crypttab, as previously implemented by
2077 * /usr/lib/os-release gained a new optional field VARIANT= for
2078 distributions that support multiple variants (such as a
2079 desktop edition, a server edition, ...)
2081 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
2082 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
2083 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
2084 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
2085 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
2086 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
2087 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
2088 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
2089 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
2090 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
2091 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
2092 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
2093 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
2094 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
2095 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
2096 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
2097 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
2098 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
2099 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
2100 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
2101 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
2102 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
2103 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
2104 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
2105 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
2106 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
2107 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
2109 — Berlin, 2015-05-22
2113 * Introduce a new API "sd-hwdb.h" for querying the hardware
2114 metadata database. With this minimal interface one can query
2115 and enumerate the udev hwdb, decoupled from the old libudev
2116 library. libudev's interface for this is now only a wrapper
2117 around sd-hwdb. A new tool systemd-hwdb has been added to
2118 interface with and update the database.
2120 * When any of systemd's tools copies files (for example due to
2121 tmpfiles' C lines) a btrfs reflink will attempted first,
2122 before bytewise copying is done.
2124 * systemd-nspawn gained a new --ephemeral switch. When
2125 specified a btrfs snapshot is taken of the container's root
2126 directory, and immediately removed when the container
2127 terminates again. Thus, a container can be started whose
2128 changes never alter the container's root directory, and are
2129 lost on container termination. This switch can also be used
2130 for starting a container off the root file system of the
2131 host without affecting the host OS. This switch is only
2132 available on btrfs file systems.
2134 * systemd-nspawn gained a new --template= switch. It takes the
2135 path to a container tree to use as template for the tree
2136 specified via --directory=, should that directory be
2137 missing. This allows instantiating containers dynamically,
2138 on first run. This switch is only available on btrfs file
2141 * When a .mount unit refers to a mount point on which multiple
2142 mounts are stacked, and the .mount unit is stopped all of
2143 the stacked mount points will now be unmounted until no
2144 mount point remains.
2146 * systemd now has an explicit notion of supported and
2147 unsupported unit types. Jobs enqueued for unsupported unit
2148 types will now fail with an "unsupported" error code. More
2149 specifically .swap, .automount and .device units are not
2150 supported in containers, .busname units are not supported on
2151 non-kdbus systems. .swap and .automount are also not
2152 supported if their respective kernel compile time options
2155 * machinectl gained support for two new "copy-from" and
2156 "copy-to" commands for copying files from a running
2157 container to the host or vice versa.
2159 * machinectl gained support for a new "bind" command to bind
2160 mount host directories into local containers. This is
2161 currently only supported for nspawn containers.
2163 * networkd gained support for configuring bridge forwarding
2164 database entries (fdb) from .network files.
2166 * A new tiny daemon "systemd-importd" has been added that can
2167 download container images in tar, raw, qcow2 or dkr formats,
2168 and make them available locally in /var/lib/machines, so
2169 that they can run as nspawn containers. The daemon can GPG
2170 verify the downloads (not supported for dkr, since it has no
2171 provisions for verifying downloads). It will transparently
2172 decompress bz2, xz, gzip compressed downloads if necessary,
2173 and restore sparse files on disk. The daemon uses privilege
2174 separation to ensure the actual download logic runs with
2175 fewer privileges than the daemon itself. machinectl has
2176 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
2177 make the functionality of importd available to the
2178 user. With this in place the Fedora and Ubuntu "Cloud"
2179 images can be downloaded and booted as containers unmodified
2180 (the Fedora images lack the appropriate GPG signature files
2181 currently, so they cannot be verified, but this will change
2182 soon, hopefully). Note that downloading images is currently
2183 only fully supported on btrfs.
2185 * machinectl is now able to list container images found in
2186 /var/lib/machines, along with some metadata about sizes of
2187 disk and similar. If the directory is located on btrfs and
2188 quota is enabled, this includes quota display. A new command
2189 "image-status" has been added that shows additional
2190 information about images.
2192 * machinectl is now able to clone container images
2193 efficiently, if the underlying file system (btrfs) supports
2194 it, with the new "machinectl clone" command. It also
2195 gained commands for renaming and removing images, as well as
2196 marking them read-only or read-write (supported also on
2197 legacy file systems).
2199 * networkd gained support for collecting LLDP network
2200 announcements, from hardware that supports this. This is
2201 shown in networkctl output.
2203 * systemd-run gained support for a new -t (--pty) switch for
2204 invoking a binary on a pty whose input and output is
2205 connected to the invoking terminal. This allows executing
2206 processes as system services while interactively
2207 communicating with them via the terminal. Most interestingly
2208 this is supported across container boundaries. Invoking
2209 "systemd-run -t /bin/bash" is an alternative to running a
2210 full login session, the difference being that the former
2211 will not register a session, nor go through the PAM session
2214 * tmpfiles gained support for a new "v" line type for creating
2215 btrfs subvolumes. If the underlying file system is a legacy
2216 file system, this automatically degrades to creating a
2217 normal directory. Among others /var/lib/machines is now
2218 created like this at boot, should it be missing.
2220 * The directory /var/lib/containers/ has been deprecated and
2221 been replaced by /var/lib/machines. The term "machines" has
2222 been used in the systemd context as generic term for both
2223 VMs and containers, and hence appears more appropriate for
2224 this, as the directory can also contain raw images bootable
2227 * systemd-nspawn when invoked with -M but without --directory=
2228 or --image= is now capable of searching for the container
2229 root directory, subvolume or disk image automatically, in
2230 /var/lib/machines. systemd-nspawn@.service has been updated
2231 to make use of this, thus allowing it to be used for raw
2234 * A new machines.target unit has been introduced that is
2235 supposed to group all containers/VMs invoked as services on
2236 the system. systemd-nspawn@.service has been updated to
2237 integrate with that.
2239 * machinectl gained a new "start" command, for invoking a
2240 container as a service. "machinectl start foo" is mostly
2241 equivalent to "systemctl start systemd-nspawn@foo.service",
2242 but handles escaping in a nicer way.
2244 * systemd-nspawn will now mount most of the cgroupfs tree
2245 read-only into each container, with the exception of the
2246 container's own subtree in the name=systemd hierarchy.
2248 * journald now sets the special FS_NOCOW file flag for its
2249 journal files. This should improve performance on btrfs, by
2250 avoiding heavy fragmentation when journald's write-pattern
2251 is used on COW file systems. It degrades btrfs' data
2252 integrity guarantees for the files to the same levels as for
2253 ext3/ext4 however. This should be OK though as journald does
2254 its own data integrity checks and all its objects are
2255 checksummed on disk. Also, journald should handle btrfs disk
2256 full events a lot more gracefully now, by processing SIGBUS
2257 errors, and not relying on fallocate() anymore.
2259 * When journald detects that journal files it is writing to
2260 have been deleted it will immediately start new journal
2263 * systemd now provides a way to store file descriptors
2264 per-service in PID 1. This is useful for daemons to ensure
2265 that fds they require are not lost during a daemon
2266 restart. The fds are passed to the daemon on the next
2267 invocation in the same way socket activation fds are
2268 passed. This is now used by journald to ensure that the
2269 various sockets connected to all the system's stdout/stderr
2270 are not lost when journald is restarted. File descriptors
2271 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
2272 an extension to sd_notify(). Note that a limit is enforced
2273 on the number of fds a service can store in PID 1, and it
2274 defaults to 0, so that no fds may be stored, unless this is
2275 explicitly turned on.
2277 * The default TERM variable to use for units connected to a
2278 terminal, when no other value is explicitly is set is now
2279 vt220 rather than vt102. This should be fairly safe still,
2280 but allows PgUp/PgDn work.
2282 * The /etc/crypttab option header= as known from Debian is now
2285 * "loginctl user-status" and "loginctl session-status" will
2286 now show the last 10 lines of log messages of the
2287 user/session following the status output. Similar,
2288 "machinectl status" will show the last 10 log lines
2289 associated with a virtual machine or container
2290 service. (Note that this is usually not the log messages
2291 done in the VM/container itself, but simply what the
2292 container manager logs. For nspawn this includes all console
2295 * "loginctl session-status" without further argument will now
2296 show the status of the session of the caller. Similar,
2297 "lock-session", "unlock-session", "activate",
2298 "enable-linger", "disable-linger" may now be called without
2299 session/user parameter in which case they apply to the
2300 caller's session/user.
2302 * An X11 session scriptlet is now shipped that uploads
2303 $DISPLAY and $XAUTHORITY into the environment of the systemd
2304 --user daemon if a session begins. This should improve
2305 compatibility with X11 enabled applications run as systemd
2308 * Generators are now subject to masking via /etc and /run, the
2309 same way as unit files.
2311 * networkd .network files gained support for configuring
2312 per-link IPv4/IPv6 packet forwarding as well as IPv4
2313 masquerading. This is by default turned on for veth links to
2314 containers, as registered by systemd-nspawn. This means that
2315 nspawn containers run with --network-veth will now get
2316 automatic routed access to the host's networks without any
2317 further configuration or setup, as long as networkd runs on
2320 * systemd-nspawn gained the --port= (-p) switch to expose TCP
2321 or UDP posts of a container on the host. With this in place
2322 it is possible to run containers with private veth links
2323 (--network-veth), and have their functionality exposed on
2324 the host as if their services were running directly on the
2327 * systemd-nspawn's --network-veth switch now gained a short
2328 version "-n", since with the changes above it is now truly
2329 useful out-of-the-box. The systemd-nspawn@.service has been
2330 updated to make use of it too by default.
2332 * systemd-nspawn will now maintain a per-image R/W lock, to
2333 ensure that the same image is not started more than once
2334 writable. (It's OK to run an image multiple times
2335 simultaneously in read-only mode.)
2337 * systemd-nspawn's --image= option is now capable of
2338 dissecting and booting MBR and GPT disk images that contain
2339 only a single active Linux partition. Previously it
2340 supported only GPT disk images with proper GPT type
2341 IDs. This allows running cloud images from major
2342 distributions directly with systemd-nspawn, without
2345 * In addition to collecting mouse dpi data in the udev
2346 hardware database, there's now support for collecting angle
2347 information for mouse scroll wheels. The database is
2348 supposed to guarantee similar scrolling behavior on mice
2349 that it knows about. There's also support for collecting
2350 information about Touchpad types.
2352 * udev's input_id built-in will now also collect touch screen
2353 dimension data and attach it to probed devices.
2355 * /etc/os-release gained support for a Distribution Privacy
2358 * networkd gained support for creating "ipvlan", "gretap",
2359 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
2361 * systemd-tmpfiles gained support for "a" lines for setting
2364 * systemd-nspawn will now mount /tmp in the container to
2365 tmpfs, automatically.
2367 * systemd now exposes the memory.usage_in_bytes cgroup
2368 attribute and shows it for each service in the "systemctl
2369 status" output, if available.
2371 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
2372 immediate reboot is triggered. This useful if shutdown is
2373 hung and is unable to complete, to expedite the
2374 operation. Note that this kind of reboot will still unmount
2375 all file systems, and hence should not result in fsck being
2378 * A .device unit for an optical block device will now be
2379 considered active only when a medium is in the drive. Also,
2380 mount units are now bound to their backing devices thus
2381 triggering automatic unmounting when devices become
2382 unavailable. With this in place systemd will now
2383 automatically unmount left-over mounts when a CD-ROM is
2384 ejected or an USB stick is yanked from the system.
2386 * networkd-wait-online now has support for waiting for
2387 specific interfaces only (with globbing), and for giving up
2388 after a configurable timeout.
2390 * networkd now exits when idle. It will be automatically
2391 restarted as soon as interfaces show up, are removed or
2392 change state. networkd will stay around as long as there is
2393 at least one DHCP state machine or similar around, that keep
2396 * networkd may now configure IPv6 link-local addressing in
2397 addition to IPv4 link-local addressing.
2399 * The IPv6 "token" for use in SLAAC may now be configured for
2400 each .network interface in networkd.
2402 * Routes configured with networkd may now be assigned a scope
2405 * networkd's [Match] sections now support globbing and lists
2406 of multiple space-separated matches per item.
2408 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
2409 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
2410 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
2411 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
2412 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
2413 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
2414 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
2415 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
2416 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
2417 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
2418 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
2419 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
2420 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
2421 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
2422 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
2423 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
2424 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
2425 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
2426 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
2427 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
2428 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
2429 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
2430 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
2431 Hoffmann, Zbigniew Jędrzejewski-Szmek
2433 — Berlin, 2015-02-16
2437 * When querying unit file enablement status (for example via
2438 "systemctl is-enabled"), a new state "indirect" is now known
2439 which indicates that a unit might not be enabled itself, but
2440 another unit listed in its Also= setting might be.
2442 * Similar to the various existing ConditionXYZ= settings for
2443 units, there are now matching AssertXYZ= settings. While
2444 failing conditions cause a unit to be skipped, but its job
2445 to succeed, failing assertions declared like this will cause
2446 a unit start operation and its job to fail.
2448 * hostnamed now knows a new chassis type "embedded".
2450 * systemctl gained a new "edit" command. When used on a unit
2451 file, this allows extending unit files with .d/ drop-in
2452 configuration snippets or editing the full file (after
2453 copying it from /usr/lib to /etc). This will invoke the
2454 user's editor (as configured with $EDITOR), and reload the
2455 modified configuration after editing.
2457 * "systemctl status" now shows the suggested enablement state
2458 for a unit, as declared in the (usually vendor-supplied)
2459 system preset files.
2461 * nss-myhostname will now resolve the single-label host name
2462 "gateway" to the locally configured default IP routing
2463 gateways, ordered by their metrics. This assigns a stable
2464 name to the used gateways, regardless which ones are
2465 currently configured. Note that the name will only be
2466 resolved after all other name sources (if nss-myhostname is
2467 configured properly) and should hence not negatively impact
2468 systems that use the single-label host name "gateway" in
2471 * systemd-inhibit now allows filtering by mode when listing
2474 * Scope and service units gained a new "Delegate" boolean
2475 property, which, when set, allows processes running inside the
2476 unit to further partition resources. This is primarily
2477 useful for systemd user instances as well as container
2480 * journald will now pick up audit messages directly from
2481 the kernel, and log them like any other log message. The
2482 audit fields are split up and fully indexed. This means that
2483 journalctl in many ways is now a (nicer!) alternative to
2484 ausearch, the traditional audit client. Note that this
2485 implements only a minimal audit client. If you want the
2486 special audit modes like reboot-on-log-overflow, please use
2487 the traditional auditd instead, which can be used in
2488 parallel to journald.
2490 * The ConditionSecurity= unit file option now understands the
2491 special string "audit" to check whether auditing is
2494 * journalctl gained two new commands --vacuum-size= and
2495 --vacuum-time= to delete old journal files until the
2496 remaining ones take up no more than the specified size on disk,
2497 or are not older than the specified time.
2499 * A new, native PPPoE library has been added to sd-network,
2500 systemd's library of light-weight networking protocols. This
2501 library will be used in a future version of networkd to
2502 enable PPPoE communication without an external pppd daemon.
2504 * The busctl tool now understands a new "capture" verb that
2505 works similar to "monitor", but writes a packet capture
2506 trace to STDOUT that can be redirected to a file which is
2507 compatible with libcap's capture file format. This can then
2508 be loaded in Wireshark and similar tools to inspect bus
2511 * The busctl tool now understands a new "tree" verb that shows
2512 the object trees of a specific service on the bus, or of all
2515 * The busctl tool now understands a new "introspect" verb that
2516 shows all interfaces and members of objects on the bus,
2517 including their signature and values. This is particularly
2518 useful to get more information about bus objects shown by
2519 the new "busctl tree" command.
2521 * The busctl tool now understands new verbs "call",
2522 "set-property" and "get-property" for invoking bus method
2523 calls, setting and getting bus object properties in a
2526 * busctl gained a new --augment-creds= argument that controls
2527 whether the tool shall augment credential information it
2528 gets from the bus with data from /proc, in a possibly
2531 * nspawn's --link-journal= switch gained two new values
2532 "try-guest" and "try-host" that work like "guest" and
2533 "host", but do not fail if the host has no persistent
2534 journalling enabled. -j is now equivalent to
2535 --link-journal=try-guest.
2537 * macvlan network devices created by nspawn will now have
2538 stable MAC addresses.
2540 * A new SmackProcessLabel= unit setting has been added, which
2541 controls the SMACK security label processes forked off by
2542 the respective unit shall use.
2544 * If compiled with --enable-xkbcommon, systemd-localed will
2545 verify x11 keymap settings by compiling the given keymap. It
2546 will spew out warnings if the compilation fails. This
2547 requires libxkbcommon to be installed.
2549 * When a coredump is collected, a larger number of metadata
2550 fields is now collected and included in the journal records
2551 created for it. More specifically, control group membership,
2552 environment variables, memory maps, working directory,
2553 chroot directory, /proc/$PID/status, and a list of open file
2554 descriptors is now stored in the log entry.
2556 * The udev hwdb now contains DPI information for mice. For
2559 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
2561 * All systemd programs that read standalone configuration
2562 files in /etc now also support a corresponding series of
2563 .conf.d configuration directories in /etc/, /run/,
2564 /usr/local/lib/, /usr/lib/, and (if configured with
2565 --enable-split-usr) /lib/. In particular, the following
2566 configuration files now have corresponding configuration
2567 directories: system.conf user.conf, logind.conf,
2568 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
2569 resolved.conf, timesyncd.conf, journal-remote.conf, and
2570 journal-upload.conf. Note that distributions should use the
2571 configuration directories in /usr/lib/; the directories in
2572 /etc/ are reserved for the system administrator.
2574 * systemd-rfkill will no longer take the rfkill device name
2575 into account when storing rfkill state on disk, as the name
2576 might be dynamically assigned and not stable. Instead, the
2577 ID_PATH udev variable combined with the rfkill type (wlan,
2578 bluetooth, ...) is used.
2580 * A new service systemd-machine-id-commit.service has been
2581 added. When used on systems where /etc is read-only during
2582 boot, and /etc/machine-id is not initialized (but an empty
2583 file), this service will copy the temporary machine ID
2584 created as replacement into /etc after the system is fully
2585 booted up. This is useful for systems that are freshly
2586 installed with a non-initialized machine ID, but should get
2587 a fixed machine ID for subsequent boots.
2589 * networkd's .netdev files now provide a large set of
2590 configuration parameters for VXLAN devices. Similarly, the
2591 bridge port cost parameter is now configurable in .network
2592 files. There's also new support for configuring IP source
2593 routing. networkd .link files gained support for a new
2594 OriginalName= match that is useful to match against the
2595 original interface name the kernel assigned. .network files
2596 may include MTU= and MACAddress= fields for altering the MTU
2597 and MAC address while being connected to a specific network
2600 * The LUKS logic gained supported for configuring
2601 UUID-specific key files. There's also new support for naming
2602 LUKS device from the kernel command line, using the new
2603 luks.name= argument.
2605 * Timer units may now be transiently created via the bus API
2606 (this was previously already available for scope and service
2607 units). In addition it is now possible to create multiple
2608 transient units at the same time with a single bus call. The
2609 "systemd-run" tool has been updated to make use of this for
2610 running commands on a specified time, in at(1)-style.
2612 * tmpfiles gained support for "t" lines, for assigning
2613 extended attributes to files. Among other uses this may be
2614 used to assign SMACK labels to files.
2616 Contributions from: Alin Rauta, Alison Chaiken, Andrej
2617 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
2618 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
2619 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
2620 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
2621 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
2622 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
2623 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
2624 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
2625 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
2626 Michael Biebl, Michael Chapman, Michael Marineau, Michal
2627 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
2628 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
2629 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
2630 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
2631 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
2632 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
2633 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
2635 — Berlin, 2014-12-10
2639 * journalctl gained the new options -t/--identifier= to match
2640 on the syslog identifier (aka "tag"), as well as --utc to
2641 show log timestamps in the UTC timezone. journalctl now also
2642 accepts -n/--lines=all to disable line capping in a pager.
2644 * journalctl gained a new switch, --flush, that synchronously
2645 flushes logs from /run/log/journal to /var/log/journal if
2646 persistent storage is enabled. systemd-journal-flush.service
2647 now waits until the operation is complete.
2649 * Services can notify the manager before they start a reload
2650 (by sending RELOADING=1) or shutdown (by sending
2651 STOPPING=1). This allows the manager to track and show the
2652 internal state of daemons and closes a race condition when
2653 the process is still running but has closed its D-Bus
2656 * Services with Type=oneshot do not have to have any ExecStart
2659 * User units are now loaded also from
2660 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
2661 /run/systemd/user directory that was already previously
2662 supported, but is under the control of the user.
2664 * Job timeouts (i.e. time-outs on the time a job that is
2665 queued stays in the run queue) can now optionally result in
2666 immediate reboot or power-off actions (JobTimeoutAction= and
2667 JobTimeoutRebootArgument=). This is useful on ".target"
2668 units, to limit the maximum time a target remains
2669 undispatched in the run queue, and to trigger an emergency
2670 operation in such a case. This is now used by default to
2671 turn off the system if boot-up (as defined by everything in
2672 basic.target) hangs and does not complete for at least
2673 15min. Also, if power-off or reboot hang for at least 30min
2674 an immediate power-off/reboot operation is triggered. This
2675 functionality is particularly useful to increase reliability
2676 on embedded devices, but also on laptops which might
2677 accidentally get powered on when carried in a backpack and
2678 whose boot stays stuck in a hard disk encryption passphrase
2681 * systemd-logind can be configured to also handle lid switch
2682 events even when the machine is docked or multiple displays
2683 are attached (HandleLidSwitchDocked= option).
2685 * A helper binary and a service have been added which can be
2686 used to resume from hibernation in the initramfs. A
2687 generator will parse the resume= option on the kernel
2688 command line to trigger resume.
2690 * A user console daemon systemd-consoled has been
2691 added. Currently, it is a preview, and will so far open a
2692 single terminal on each session of the user marked as
2693 Desktop=systemd-console.
2695 * Route metrics can be specified for DHCP routes added by
2698 * The SELinux context of socket-activated services can be set
2699 from the information provided by the networking stack
2700 (SELinuxContextFromNet= option).
2702 * Userspace firmware loading support has been removed and
2703 the minimum supported kernel version is thus bumped to 3.7.
2705 * Timeout for udev workers has been increased from 1 to 3
2706 minutes, but a warning will be printed after 1 minute to
2707 help diagnose kernel modules that take a long time to load.
2709 * Udev rules can now remove tags on devices with TAG-="foobar".
2711 * systemd's readahead implementation has been removed. In many
2712 circumstances it didn't give expected benefits even for
2713 rotational disk drives and was becoming less relevant in the
2714 age of SSDs. As none of the developers has been using
2715 rotating media anymore, and nobody stepped up to actively
2716 maintain this component of systemd it has now been removed.
2718 * Swap units can use Options= to specify discard options.
2719 Discard options specified for swaps in /etc/fstab are now
2722 * Docker containers are now detected as a separate type of
2725 * The Password Agent protocol gained support for queries where
2726 the user input is shown, useful e.g. for user names.
2727 systemd-ask-password gained a new --echo option to turn that
2730 * The default sysctl.d/ snippets will now set:
2732 net.core.default_qdisc = fq_codel
2734 This selects Fair Queuing Controlled Delay as the default
2735 queuing discipline for network interfaces. fq_codel helps
2736 fight the network bufferbloat problem. It is believed to be
2737 a good default with no tuning required for most workloads.
2738 Downstream distributions may override this choice. On 10Gbit
2739 servers that do not do forwarding, "fq" may perform better.
2740 Systems without a good clocksource should use "pfifo_fast".
2742 * If kdbus is enabled during build a new option BusPolicy= is
2743 available for service units, that allows locking all service
2744 processes into a stricter bus policy, in order to limit
2745 access to various bus services, or even hide most of them
2746 from the service's view entirely.
2748 * networkctl will now show the .network and .link file
2749 networkd has applied to a specific interface.
2751 * sd-login gained a new API call sd_session_get_desktop() to
2752 query which desktop environment has been selected for a
2755 * UNIX utmp support is now compile-time optional to support
2756 legacy-free systems.
2758 * systemctl gained two new commands "add-wants" and
2759 "add-requires" for pulling in units from specific targets
2762 * If the word "rescue" is specified on the kernel command line
2763 the system will now boot into rescue mode (aka
2764 rescue.target), which was previously available only by
2765 specifying "1" or "systemd.unit=rescue.target" on the kernel
2766 command line. This new kernel command line option nicely
2767 mirrors the already existing "emergency" kernel command line
2770 * New kernel command line options mount.usr=, mount.usrflags=,
2771 mount.usrfstype= have been added that match root=, rootflags=,
2772 rootfstype= but allow mounting a specific file system to
2775 * The $NOTIFY_SOCKET is now also passed to control processes of
2776 services, not only the main process.
2778 * This version reenables support for fsck's -l switch. This
2779 means at least version v2.25 of util-linux is required for
2780 operation, otherwise dead-locks on device nodes may
2781 occur. Again: you need to update util-linux to at least
2782 v2.25 when updating systemd to v217.
2784 * The "multi-seat-x" tool has been removed from systemd, as
2785 its functionality has been integrated into X servers 1.16,
2786 and the tool is hence redundant. It is recommended to update
2787 display managers invoking this tool to simply invoke X
2788 directly from now on, again.
2790 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
2791 message flag has been added for all of systemd's PolicyKit
2792 authenticated method calls has been added. In particular
2793 this now allows optional interactive authorization via
2794 PolicyKit for many of PID1's privileged operations such as
2795 unit file enabling and disabling.
2797 * "udevadm hwdb --update" learnt a new switch "--usr" for
2798 placing the rebuilt hardware database in /usr instead of
2799 /etc. When used only hardware database entries stored in
2800 /usr will be used, and any user database entries in /etc are
2801 ignored. This functionality is useful for vendors to ship a
2802 pre-built database on systems where local configuration is
2803 unnecessary or unlikely.
2805 * Calendar time specifications in .timer units now also
2806 understand the strings "semi-annually", "quarterly" and
2807 "minutely" as shortcuts (in addition to the preexisting
2808 "anually", "hourly", ...).
2810 * systemd-tmpfiles will now correctly create files in /dev
2811 at boot which are marked for creation only at boot. It is
2812 recommended to always create static device nodes with 'c!'
2813 and 'b!', so that they are created only at boot and not
2814 overwritten at runtime.
2816 * When the watchdog logic is used for a service (WatchdogSec=)
2817 and the watchdog timeout is hit the service will now be
2818 terminated with SIGABRT (instead of just SIGTERM), in order
2819 to make sure a proper coredump and backtrace is
2820 generated. This ensures that hanging services will result in
2821 similar coredump/backtrace behaviour as services that hit a
2824 Contributions from: Andreas Henriksson, Andrei Borzenkov,
2825 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
2826 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
2827 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
2828 Herrmann, David Sommerseth, David Strauss, Emil Renner
2829 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
2830 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
2831 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
2832 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
2833 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
2834 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
2835 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
2836 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
2837 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
2838 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
2839 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
2840 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
2841 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
2842 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
2843 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
2844 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
2847 — Berlin, 2014-10-28
2851 * timedated no longer reads NTP implementation unit names from
2852 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
2853 implementations should add a
2855 Conflicts=systemd-timesyncd.service
2857 to their unit files to take over and replace systemd's NTP
2858 default functionality.
2860 * systemd-sysusers gained a new line type "r" for configuring
2861 which UID/GID ranges to allocate system users/groups
2862 from. Lines of type "u" may now add an additional column
2863 that specifies the home directory for the system user to be
2864 created. Also, systemd-sysusers may now optionally read user
2865 information from STDIN instead of a file. This is useful for
2866 invoking it from RPM preinst scriptlets that need to create
2867 users before the first RPM file is installed since these
2868 files might need to be owned by them. A new
2869 %sysusers_create_inline RPM macro has been introduced to do
2870 just that. systemd-sysusers now updates the shadow files as
2871 well as the user/group databases, which should enhance
2872 compatibility with certain tools like grpck.
2874 * A number of bus APIs of PID 1 now optionally consult
2875 PolicyKit to permit access for otherwise unprivileged
2876 clients under certain conditions. Note that this currently
2877 doesn't support interactive authentication yet, but this is
2878 expected to be added eventually, too.
2880 * /etc/machine-info now has new fields for configuring the
2881 deployment environment of the machine, as well as the
2882 location of the machine. hostnamectl has been updated with
2883 new command to update these fields.
2885 * systemd-timesyncd has been updated to automatically acquire
2886 NTP server information from systemd-networkd, which might
2887 have been discovered via DHCP.
2889 * systemd-resolved now includes a caching DNS stub resolver
2890 and a complete LLMNR name resolution implementation. A new
2891 NSS module "nss-resolve" has been added which can be used
2892 instead of glibc's own "nss-dns" to resolve hostnames via
2893 systemd-resolved. Hostnames, addresses and arbitrary RRs may
2894 be resolved via systemd-resolved D-Bus APIs. In contrast to
2895 the glibc internal resolver systemd-resolved is aware of
2896 multi-homed system, and keeps DNS server and caches separate
2897 and per-interface. Queries are sent simultaneously on all
2898 interfaces that have DNS servers configured, in order to
2899 properly handle VPNs and local LANs which might resolve
2900 separate sets of domain names. systemd-resolved may acquire
2901 DNS server information from systemd-networkd automatically,
2902 which in turn might have discovered them via DHCP. A tool
2903 "systemd-resolve-host" has been added that may be used to
2904 query the DNS logic in resolved. systemd-resolved implements
2905 IDNA and automatically uses IDNA or UTF-8 encoding depending
2906 on whether classic DNS or LLMNR is used as transport. In the
2907 next releases we intend to add a DNSSEC and mDNS/DNS-SD
2908 implementation to systemd-resolved.
2910 * A new NSS module nss-mymachines has been added, that
2911 automatically resolves the names of all local registered
2912 containers to their respective IP addresses.
2914 * A new client tool "networkctl" for systemd-networkd has been
2915 added. It currently is entirely passive and will query
2916 networking configuration from udev, rtnetlink and networkd,
2917 and present it to the user in a very friendly
2918 way. Eventually, we hope to extend it to become a full
2919 control utility for networkd.
2921 * .socket units gained a new DeferAcceptSec= setting that
2922 controls the kernels' TCP_DEFER_ACCEPT sockopt for
2923 TCP. Similarly, support for controlling TCP keep-alive
2924 settings has been added (KeepAliveTimeSec=,
2925 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
2926 turning off Nagle's algorithm on TCP has been added
2929 * logind learned a new session type "web", for use in projects
2930 like Cockpit which register web clients as PAM sessions.
2932 * timer units with at least one OnCalendar= setting will now
2933 be started only after timer-sync.target has been
2934 reached. This way they will not elapse before the system
2935 clock has been corrected by a local NTP client or
2936 similar. This is particular useful on RTC-less embedded
2937 machines, that come up with an invalid system clock.
2939 * systemd-nspawn's --network-veth= switch should now result in
2940 stable MAC addresses for both the outer and the inner side
2943 * systemd-nspawn gained a new --volatile= switch for running
2944 container instances with /etc or /var unpopulated.
2946 * The kdbus client code has been updated to use the new Linux
2947 3.17 memfd subsystem instead of the old kdbus-specific one.
2949 * systemd-networkd's DHCP client and server now support
2950 FORCERENEW. There are also new configuration options to
2951 configure the vendor client identifier and broadcast mode
2954 * systemd will no longer inform the kernel about the current
2955 timezone, as this is necessarily incorrect and racy as the
2956 kernel has no understanding of DST and similar
2957 concepts. This hence means FAT timestamps will be always
2958 considered UTC, similar to what Android is already
2959 doing. Also, when the RTC is configured to the local time
2960 (rather than UTC) systemd will never synchronize back to it,
2961 as this might confuse Windows at a later boot.
2963 * systemd-analyze gained a new command "verify" for offline
2964 validation of unit files.
2966 * systemd-networkd gained support for a couple of additional
2967 settings for bonding networking setups. Also, the metric for
2968 statically configured routes may now be configured. For
2969 network interfaces where this is appropriate the peer IP
2970 address may now be configured.
2972 * systemd-networkd's DHCP client will no longer request
2973 broadcasting by default, as this tripped up some networks.
2974 For hardware where broadcast is required the feature should
2975 be switched back on using RequestBroadcast=yes.
2977 * systemd-networkd will now set up IPv4LL addresses (when
2978 enabled) even if DHCP is configured successfully.
2980 * udev will now default to respect network device names given
2981 by the kernel when the kernel indicates that these are
2982 predictable. This behavior can be tweaked by changing
2983 NamePolicy= in the relevant .link file.
2985 * A new library systemd-terminal has been added that
2986 implements full TTY stream parsing and rendering. This
2987 library is supposed to be used later on for implementing a
2988 full userspace VT subsystem, replacing the current kernel
2991 * A new tool systemd-journal-upload has been added to push
2992 journal data to a remote system running
2993 systemd-journal-remote.
2995 * journald will no longer forward all local data to another
2996 running syslog daemon. This change has been made because
2997 rsyslog (which appears to be the most commonly used syslog
2998 implementation these days) no longer makes use of this, and
2999 instead pulls the data out of the journal on its own. Since
3000 forwarding the messages to a non-existent syslog server is
3001 more expensive than we assumed we have now turned this
3002 off. If you run a syslog server that is not a recent rsyslog
3003 version, you have to turn this option on again
3004 (ForwardToSyslog= in journald.conf).
3006 * journald now optionally supports the LZ4 compressor for
3007 larger journal fields. This compressor should perform much
3008 better than XZ which was the previous default.
3010 * machinectl now shows the IP addresses of local containers,
3011 if it knows them, plus the interface name of the container.
3013 * A new tool "systemd-escape" has been added that makes it
3014 easy to escape strings to build unit names and similar.
3016 * sd_notify() messages may now include a new ERRNO= field
3017 which is parsed and collected by systemd and shown among the
3018 "systemctl status" output for a service.
3020 * A new component "systemd-firstboot" has been added that
3021 queries the most basic systemd information (timezone,
3022 hostname, root password) interactively on first
3023 boot. Alternatively it may also be used to provision these
3024 things offline on OS images installed into directories.
3026 * The default sysctl.d/ snippets will now set
3028 net.ipv4.conf.default.promote_secondaries=1
3030 This has the benefit of no flushing secondary IP addresses
3031 when primary addresses are removed.
3033 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
3034 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
3035 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
3036 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
3037 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
3038 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
3039 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
3040 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
3041 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
3042 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
3043 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
3044 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
3045 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
3046 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
3047 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
3049 — Berlin, 2014-08-19
3053 * A new tool systemd-sysusers has been added. This tool
3054 creates system users and groups in /etc/passwd and
3055 /etc/group, based on static declarative system user/group
3056 definitions in /usr/lib/sysusers.d/. This is useful to
3057 enable factory resets and volatile systems that boot up with
3058 an empty /etc directory, and thus need system users and
3059 groups created during early boot. systemd now also ships
3060 with two default sysusers.d/ files for the most basic
3061 users and groups systemd and the core operating system
3064 * A new tmpfiles snippet has been added that rebuilds the
3065 essential files in /etc on boot, should they be missing.
3067 * A directive for ensuring automatic clean-up of
3068 /var/cache/man/ has been removed from the default
3069 configuration. This line should now be shipped by the man
3070 implementation. The necessary change has been made to the
3071 man-db implementation. Note that you need to update your man
3072 implementation to one that ships this line, otherwise no
3073 automatic clean-up of /var/cache/man will take place.
3075 * A new condition ConditionNeedsUpdate= has been added that
3076 may conditionalize services to only run when /etc or /var
3077 are "older" than the vendor operating system resources in
3078 /usr. This is useful for reconstructing or updating /etc
3079 after an offline update of /usr or a factory reset, on the
3080 next reboot. Services that want to run once after such an
3081 update or reset should use this condition and order
3082 themselves before the new systemd-update-done.service, which
3083 will mark the two directories as fully updated. A number of
3084 service files have been added making use of this, to rebuild
3085 the udev hardware database, the journald message catalog and
3086 dynamic loader cache (ldconfig). The systemd-sysusers tool
3087 described above also makes use of this now. With this in
3088 place it is now possible to start up a minimal operating
3089 system with /etc empty cleanly. For more information on the
3090 concepts involved see this recent blog story:
3092 http://0pointer.de/blog/projects/stateless.html
3094 * A new system group "input" has been introduced, and all
3095 input device nodes get this group assigned. This is useful
3096 for system-level software to get access to input devices. It
3097 complements what is already done for "audio" and "video".
3099 * systemd-networkd learnt minimal DHCPv4 server support in
3100 addition to the existing DHCPv4 client support. It also
3101 learnt DHCPv6 client and IPv6 Router Solicitation client
3102 support. The DHCPv4 client gained support for static routes
3103 passed in from the server. Note that the [DHCPv4] section
3104 known in older systemd-networkd versions has been renamed to
3105 [DHCP] and is now also used by the DHCPv6 client. Existing
3106 .network files using settings of this section should be
3107 updated, though compatibility is maintained. Optionally, the
3108 client hostname may now be sent to the DHCP server.
3110 * networkd gained support for vxlan virtual networks as well
3111 as tun/tap and dummy devices.
3113 * networkd gained support for automatic allocation of address
3114 ranges for interfaces from a system-wide pool of
3115 addresses. This is useful for dynamically managing a large
3116 number of interfaces with a single network configuration
3117 file. In particular this is useful to easily assign
3118 appropriate IP addresses to the veth links of a large number
3119 of nspawn instances.
3121 * RPM macros for processing sysusers, sysctl and binfmt
3122 drop-in snippets at package installation time have been
3125 * The /etc/os-release file should now be placed in
3126 /usr/lib/os-release. The old location is automatically
3127 created as symlink. /usr/lib is the more appropriate
3128 location of this file, since it shall actually describe the
3129 vendor operating system shipped in /usr, and not the
3130 configuration stored in /etc.
3132 * .mount units gained a new boolean SloppyOptions= setting
3133 that maps to mount(8)'s -s option which enables permissive
3134 parsing of unknown mount options.
3136 * tmpfiles learnt a new "L+" directive which creates a symlink
3137 but (unlike "L") deletes a pre-existing file first, should
3138 it already exist and not already be the correct
3139 symlink. Similarly, "b+", "c+" and "p+" directives have been
3140 added as well, which create block and character devices, as
3141 well as fifos in the filesystem, possibly removing any
3142 pre-existing files of different types.
3144 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
3145 'argument' field (which so far specified the source to
3146 symlink/copy the files from) is now optional. If omitted the
3147 same file os copied from /usr/share/factory/ suffixed by the
3148 full destination path. This is useful for populating /etc
3149 with essential files, by copying them from vendor defaults
3150 shipped in /usr/share/factory/etc.
3152 * A new command "systemctl preset-all" has been added that
3153 applies the service preset settings to all installed unit
3154 files. A new switch --preset-mode= has been added that
3155 controls whether only enable or only disable operations
3158 * A new command "systemctl is-system-running" has been added
3159 that allows checking the overall state of the system, for
3160 example whether it is fully up and running.
3162 * When the system boots up with an empty /etc, the equivalent
3163 to "systemctl preset-all" is executed during early boot, to
3164 make sure all default services are enabled after a factory
3167 * systemd now contains a minimal preset file that enables the
3168 most basic services systemd ships by default.
3170 * Unit files' [Install] section gained a new DefaultInstance=
3171 field for defining the default instance to create if a
3172 template unit is enabled with no instance specified.
3174 * A new passive target cryptsetup-pre.target has been added
3175 that may be used by services that need to make they run and
3176 finish before the first LUKS cryptographic device is set up.
3178 * The /dev/loop-control and /dev/btrfs-control device nodes
3179 are now owned by the "disk" group by default, opening up
3180 access to this group.
3182 * systemd-coredump will now automatically generate a
3183 stack trace of all core dumps taking place on the system,
3184 based on elfutils' libdw library. This stack trace is logged
3187 * systemd-coredump may now optionally store coredumps directly
3188 on disk (in /var/lib/systemd/coredump, possibly compressed),
3189 instead of storing them unconditionally in the journal. This
3190 mode is the new default. A new configuration file
3191 /etc/systemd/coredump.conf has been added to configure this
3192 and other parameters of systemd-coredump.
3194 * coredumpctl gained a new "info" verb to show details about a
3195 specific coredump. A new switch "-1" has also been added
3196 that makes sure to only show information about the most
3197 recent entry instead of all entries. Also, as the tool is
3198 generally useful now the "systemd-" prefix of the binary
3199 name has been removed. Distributions that want to maintain
3200 compatibility with the old name should add a symlink from
3201 the old name to the new name.
3203 * journald's SplitMode= now defaults to "uid". This makes sure
3204 that unprivileged users can access their own coredumps with
3205 coredumpctl without restrictions.
3207 * New kernel command line options "systemd.wants=" (for
3208 pulling an additional unit during boot), "systemd.mask="
3209 (for masking a specific unit for the boot), and
3210 "systemd.debug-shell" (for enabling the debug shell on tty9)
3211 have been added. This is implemented in the new generator
3212 "systemd-debug-generator".
3214 * systemd-nspawn will now by default filter a couple of
3215 syscalls for containers, among them those required for
3216 kernel module loading, direct x86 IO port access, swap
3217 management, and kexec. Most importantly though
3218 open_by_handle_at() is now prohibited for containers,
3219 closing a hole similar to a recently discussed vulnerability
3220 in docker regarding access to files on file hierarchies the
3221 container should normally not have access to. Note that, for
3222 nspawn, we generally make no security claims anyway (and
3223 this is explicitly documented in the man page), so this is
3224 just a fix for one of the most obvious problems.
3226 * A new man page file-hierarchy(7) has been added that
3227 contains a minimized, modernized version of the file system
3228 layout systemd expects, similar in style to the FHS
3229 specification or hier(5). A new tool systemd-path(1) has
3230 been added to query many of these paths for the local
3233 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
3234 longer done. Since the directory now has a per-user size
3235 limit, and is cleaned on logout this appears unnecessary,
3236 in particular since this now brings the lifecycle of this
3237 directory closer in line with how IPC objects are handled.
3239 * systemd.pc now exports a number of additional directories,
3240 including $libdir (which is useful to identify the library
3241 path for the primary architecture of the system), and a
3242 couple of drop-in directories.
3244 * udev's predictable network interface names now use the dev_port
3245 sysfs attribute, introduced in linux 3.15 instead of dev_id to
3246 distinguish between ports of the same PCI function. dev_id should
3247 only be used for ports using the same HW address, hence the need
3250 * machined has been updated to export the OS version of a
3251 container (read from /etc/os-release and
3252 /usr/lib/os-release) on the bus. This is now shown in
3253 "machinectl status" for a machine.
3255 * A new service setting RestartForceExitStatus= has been
3256 added. If configured to a set of exit signals or process
3257 return values, the service will be restarted when the main
3258 daemon process exits with any of them, regardless of the
3261 * systemctl's -H switch for connecting to remote systemd
3262 machines has been extended so that it may be used to
3263 directly connect to a specific container on the
3264 host. "systemctl -H root@foobar:waldi" will now connect as
3265 user "root" to host "foobar", and then proceed directly to
3266 the container named "waldi". Note that currently you have to
3267 authenticate as user "root" for this to work, as entering
3268 containers is a privileged operation.
3270 Contributions from: Andreas Henriksson, Benjamin Steinwender,
3271 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
3272 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
3273 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
3274 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
3275 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
3276 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
3277 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
3278 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
3279 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
3280 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
3281 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
3283 — Berlin, 2014-07-03
3287 * As an experimental feature, udev now tries to lock the
3288 disk device node (flock(LOCK_SH|LOCK_NB)) while it
3289 executes events for the disk or any of its partitions.
3290 Applications like partitioning programs can lock the
3291 disk device node (flock(LOCK_EX)) and claim temporary
3292 device ownership that way; udev will entirely skip all event
3293 handling for this disk and its partitions. If the disk
3294 was opened for writing, the close will trigger a partition
3295 table rescan in udev's "watch" facility, and if needed
3296 synthesize "change" events for the disk and all its partitions.
3297 This is now unconditionally enabled, and if it turns out to
3298 cause major problems, we might turn it on only for specific
3299 devices, or might need to disable it entirely. Device Mapper
3300 devices are excluded from this logic.
3302 * We temporarily dropped the "-l" switch for fsck invocations,
3303 since they collide with the flock() logic above. util-linux
3304 upstream has been changed already to avoid this conflict,
3305 and we will readd "-l" as soon as util-linux with this
3306 change has been released.
3308 * The dependency on libattr has been removed. Since a long
3309 time, the extended attribute calls have moved to glibc, and
3310 libattr is thus unnecessary.
3312 * Virtualization detection works without privileges now. This
3313 means the systemd-detect-virt binary no longer requires
3314 CAP_SYS_PTRACE file capabilities, and our daemons can run
3315 with fewer privileges.
3317 * systemd-networkd now runs under its own "systemd-network"
3318 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
3319 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
3320 loses the ability to write to files owned by root this way.
3322 * Similarly, systemd-resolved now runs under its own
3323 "systemd-resolve" user with no capabilities remaining.
3325 * Similarly, systemd-bus-proxyd now runs under its own
3326 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
3328 * systemd-networkd gained support for setting up "veth"
3329 virtual Ethernet devices for container connectivity, as well
3330 as GRE and VTI tunnels.
3332 * systemd-networkd will no longer automatically attempt to
3333 manually load kernel modules necessary for certain tunnel
3334 transports. Instead, it is assumed the kernel loads them
3335 automatically when required. This only works correctly on
3336 very new kernels. On older kernels, please consider adding
3337 the kernel modules to /etc/modules-load.d/ as a work-around.
3339 * The resolv.conf file systemd-resolved generates has been
3340 moved to /run/systemd/resolve/. If you have a symlink from
3341 /etc/resolv.conf, it might be necessary to correct it.
3343 * Two new service settings, ProtectHome= and ProtectSystem=,
3344 have been added. When enabled, they will make the user data
3345 (such as /home) inaccessible or read-only and the system
3346 (such as /usr) read-only, for specific services. This allows
3347 very light-weight per-service sandboxing to avoid
3348 modifications of user data or system files from
3349 services. These two new switches have been enabled for all
3350 of systemd's long-running services, where appropriate.
3352 * Socket units gained new SocketUser= and SocketGroup=
3353 settings to set the owner user and group of AF_UNIX sockets
3354 and FIFOs in the file system.
3356 * Socket units gained a new RemoveOnStop= setting. If enabled,
3357 all FIFOS and sockets in the file system will be removed
3358 when the specific socket unit is stopped.
3360 * Socket units gained a new Symlinks= setting. It takes a list
3361 of symlinks to create to file system sockets or FIFOs
3362 created by the specific Unix sockets. This is useful to
3363 manage symlinks to socket nodes with the same life-cycle as
3366 * The /dev/log socket and /dev/initctl FIFO have been moved to
3367 /run, and have been replaced by symlinks. This allows
3368 connecting to these facilities even if PrivateDevices=yes is
3369 used for a service (which makes /dev/log itself unavailable,
3370 but /run is left). This also has the benefit of ensuring
3371 that /dev only contains device nodes, directories and
3372 symlinks, and nothing else.
3374 * sd-daemon gained two new calls sd_pid_notify() and
3375 sd_pid_notifyf(). They are similar to sd_notify() and
3376 sd_notifyf(), but allow overriding of the source PID of
3377 notification messages if permissions permit this. This is
3378 useful to send notify messages on behalf of a different
3379 process (for example, the parent process). The
3380 systemd-notify tool has been updated to make use of this
3381 when sending messages (so that notification messages now
3382 originate from the shell script invoking systemd-notify and
3383 not the systemd-notify process itself. This should minimize
3384 a race where systemd fails to associate notification
3385 messages to services when the originating process already
3388 * A new "on-abnormal" setting for Restart= has been added. If
3389 set, it will result in automatic restarts on all "abnormal"
3390 reasons for a process to exit, which includes unclean
3391 signals, core dumps, timeouts and watchdog timeouts, but
3392 does not include clean and unclean exit codes or clean
3393 signals. Restart=on-abnormal is an alternative for
3394 Restart=on-failure for services that shall be able to
3395 terminate and avoid restarts on certain errors, by
3396 indicating so with an unclean exit code. Restart=on-failure
3397 or Restart=on-abnormal is now the recommended setting for
3398 all long-running services.
3400 * If the InaccessibleDirectories= service setting points to a
3401 mount point (or if there are any submounts contained within
3402 it), it is now attempted to completely unmount it, to make
3403 the file systems truly unavailable for the respective
3406 * The ReadOnlyDirectories= service setting and
3407 systemd-nspawn's --read-only parameter are now recursively
3408 applied to all submounts, too.
3410 * Mount units may now be created transiently via the bus APIs.
3412 * The support for SysV and LSB init scripts has been removed
3413 from the systemd daemon itself. Instead, it is now
3414 implemented as a generator that creates native systemd units
3415 from these scripts when needed. This enables us to remove a
3416 substantial amount of legacy code from PID 1, following the
3417 fact that many distributions only ship a very small number
3418 of LSB/SysV init scripts nowadays.
3420 * Privileged Xen (dom0) domains are not considered
3421 virtualization anymore by the virtualization detection
3422 logic. After all, they generally have unrestricted access to
3423 the hardware and usually are used to manage the unprivileged
3426 * systemd-tmpfiles gained a new "C" line type, for copying
3427 files or entire directories.
3429 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
3430 lines. So far, they have been non-globbing versions of the
3431 latter, and have thus been redundant. In future, it is
3432 recommended to only use "z". "m" has hence been removed
3433 from the documentation, even though it stays supported.
3435 * A tmpfiles snippet to recreate the most basic structure in
3436 /var has been added. This is enough to create the /var/run →
3437 /run symlink and create a couple of structural
3438 directories. This allows systems to boot up with an empty or
3439 volatile /var. Of course, while with this change, the core OS
3440 now is capable with dealing with a volatile /var, not all
3441 user services are ready for it. However, we hope that sooner
3442 or later, many service daemons will be changed upstream so
3443 that they are able to automatically create their necessary
3444 directories in /var at boot, should they be missing. This is
3445 the first step to allow state-less systems that only require
3446 the vendor image for /usr to boot.
3448 * systemd-nspawn has gained a new --tmpfs= switch to mount an
3449 empty tmpfs instance to a specific directory. This is
3450 particularly useful for making use of the automatic
3451 reconstruction of /var (see above), by passing --tmpfs=/var.
3453 * Access modes specified in tmpfiles snippets may now be
3454 prefixed with "~", which indicates that they shall be masked
3455 by whether the existing file or directory is currently
3456 writable, readable or executable at all. Also, if specified,
3457 the sgid/suid/sticky bits will be masked for all
3460 * A new passive target unit "network-pre.target" has been
3461 added which is useful for services that shall run before any
3462 network is configured, for example firewall scripts.
3464 * The "floppy" group that previously owned the /dev/fd*
3465 devices is no longer used. The "disk" group is now used
3466 instead. Distributions should probably deprecate usage of
3469 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
3470 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
3471 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
3472 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
3473 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
3474 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
3477 — Berlin, 2014-06-11
3481 * A new "systemd-timesyncd" daemon has been added for
3482 synchronizing the system clock across the network. It
3483 implements an SNTP client. In contrast to NTP
3484 implementations such as chrony or the NTP reference server,
3485 this only implements a client side, and does not bother with
3486 the full NTP complexity, focusing only on querying time from
3487 one remote server and synchronizing the local clock to
3488 it. Unless you intend to serve NTP to networked clients or
3489 want to connect to local hardware clocks, this simple NTP
3490 client should be more than appropriate for most
3491 installations. The daemon runs with minimal privileges, and
3492 has been hooked up with networkd to only operate when
3493 network connectivity is available. The daemon saves the
3494 current clock to disk every time a new NTP sync has been
3495 acquired, and uses this to possibly correct the system clock
3496 early at bootup, in order to accommodate for systems that
3497 lack an RTC such as the Raspberry Pi and embedded devices,
3498 and to make sure that time monotonically progresses on these
3499 systems, even if it is not always correct. To make use of
3500 this daemon, a new system user and group "systemd-timesync"
3501 needs to be created on installation of systemd.
3503 * The queue "seqnum" interface of libudev has been disabled, as
3504 it was generally incompatible with device namespacing as
3505 sequence numbers of devices go "missing" if the devices are
3506 part of a different namespace.
3508 * "systemctl list-timers" and "systemctl list-sockets" gained
3509 a --recursive switch for showing units of these types also
3510 for all local containers, similar in style to the already
3511 supported --recursive switch for "systemctl list-units".
3513 * A new RebootArgument= setting has been added for service
3514 units, which may be used to specify a kernel reboot argument
3515 to use when triggering reboots with StartLimitAction=.
3517 * A new FailureAction= setting has been added for service
3518 units which may be used to specify an operation to trigger
3519 when a service fails. This works similarly to
3520 StartLimitAction=, but unlike it, controls what is done
3521 immediately rather than only after several attempts to
3522 restart the service in question.
3524 * hostnamed got updated to also expose the kernel name,
3525 release, and version on the bus. This is useful for
3526 executing commands like hostnamectl with the -H switch.
3527 systemd-analyze makes use of this to properly display
3528 details when running non-locally.
3530 * The bootchart tool can now show cgroup information in the
3531 graphs it generates.
3533 * The CFS CPU quota cgroup attribute is now exposed for
3534 services. The new CPUQuota= switch has been added for this
3535 which takes a percentage value. Setting this will have the
3536 result that a service may never get more CPU time than the
3537 specified percentage, even if the machine is otherwise idle.
3539 * systemd-networkd learned IPIP and SIT tunnel support.
3541 * LSB init scripts exposing a dependency on $network will now
3542 get a dependency on network-online.target rather than simply
3543 network.target. This should bring LSB handling closer to
3544 what it was on SysV systems.
3546 * A new fsck.repair= kernel option has been added to control
3547 how fsck shall deal with unclean file systems at boot.
3549 * The (.ini) configuration file parser will now silently
3550 ignore sections whose name begins with "X-". This may be
3551 used to maintain application-specific extension sections in unit
3554 * machined gained a new API to query the IP addresses of
3555 registered containers. "machinectl status" has been updated
3556 to show these addresses in its output.
3558 * A new call sd_uid_get_display() has been added to the
3559 sd-login APIs for querying the "primary" session of a
3560 user. The "primary" session of the user is elected from the
3561 user's sessions and generally a graphical session is
3562 preferred over a text one.
3564 * A minimal systemd-resolved daemon has been added. It
3565 currently simply acts as a companion to systemd-networkd and
3566 manages resolv.conf based on per-interface DNS
3567 configuration, possibly supplied via DHCP. In the long run
3568 we hope to extend this into a local DNSSEC enabled DNS and
3571 * The systemd-networkd-wait-online tool is now enabled by
3572 default. It will delay network-online.target until a network
3573 connection has been configured. The tool primarily integrates
3574 with networkd, but will also make a best effort to make sense
3575 of network configuration performed in some other way.
3577 * Two new service options StartupCPUShares= and
3578 StartupBlockIOWeight= have been added that work similarly to
3579 CPUShares= and BlockIOWeight= however only apply during
3580 system startup. This is useful to prioritize certain services
3581 differently during bootup than during normal runtime.
3583 * hostnamed has been changed to prefer the statically
3584 configured hostname in /etc/hostname (unless set to
3585 'localhost' or empty) over any dynamic one supplied by
3586 dhcp. With this change, the rules for picking the hostname
3587 match more closely the rules of other configuration settings
3588 where the local administrator's configuration in /etc always
3589 overrides any other settings.
3591 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
3592 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
3593 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
3594 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
3595 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
3596 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
3597 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
3598 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
3599 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
3600 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
3601 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
3602 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
3603 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
3604 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
3605 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
3606 Lindskog, WaLyong Cho, Will Woods, Zbigniew
3609 — Beijing, 2014-05-28
3613 * When restoring the screen brightness at boot, stay away from
3614 the darkest setting or from the lowest 5% of the available
3615 range, depending on which is the larger value of both. This
3616 should effectively protect the user from rebooting into a
3617 black screen, should the brightness have been set to minimum
3620 * sd-login gained a new sd_machine_get_class() call to
3621 determine the class ("vm" or "container") of a machine
3622 registered with machined.
3624 * sd-login gained new calls
3625 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
3626 to query the identity of the peer of a local AF_UNIX
3627 connection. They operate similarly to their sd_pid_get_xyz()
3630 * PID 1 will now maintain a system-wide system state engine
3631 with the states "starting", "running", "degraded",
3632 "maintenance", "stopping". These states are bound to system
3633 startup, normal runtime, runtime with at least one failed
3634 service, rescue/emergency mode and system shutdown. This
3635 state is shown in the "systemctl status" output when no unit
3636 name is passed. It is useful to determine system state, in
3637 particularly when doing so for many systems or containers at
3640 * A new command "list-machines" has been added to "systemctl"
3641 that lists all local OS containers and shows their system
3642 state (see above), if systemd runs inside of them.
3644 * systemctl gained a new "-r" switch to recursively enumerate
3645 units on all local containers, when used with the
3646 "list-unit" command (which is the default one that is
3647 executed when no parameters are specified).
3649 * The GPT automatic partition discovery logic will now honour
3650 two GPT partition flags: one may be set on a partition to
3651 cause it to be mounted read-only, and the other may be set
3652 on a partition to ignore it during automatic discovery.
3654 * Two new GPT type UUIDs have been added for automatic root
3655 partition discovery, for 32-bit and 64-bit ARM. This is not
3656 particularly useful for discovering the root directory on
3657 these architectures during bare-metal boots (since UEFI is
3658 not common there), but still very useful to allow booting of
3659 ARM disk images in nspawn with the -i option.
3661 * MAC addresses of interfaces created with nspawn's
3662 --network-interface= switch will now be generated from the
3663 machine name, and thus be stable between multiple invocations
3666 * logind will now automatically remove all IPC objects owned
3667 by a user if she or he fully logs out. This makes sure that
3668 users who are logged out cannot continue to consume IPC
3669 resources. This covers SysV memory, semaphores and message
3670 queues as well as POSIX shared memory and message
3671 queues. Traditionally, SysV and POSIX IPC had no life-cycle
3672 limits. With this functionality, that is corrected. This may
3673 be turned off by using the RemoveIPC= switch of logind.conf.
3675 * The systemd-machine-id-setup and tmpfiles tools gained a
3676 --root= switch to operate on a specific root directory,
3679 * journald can now forward logged messages to the TTYs of all
3680 logged in users ("wall"). This is the default for all
3681 emergency messages now.
3683 * A new tool systemd-journal-remote has been added to stream
3684 journal log messages across the network.
3686 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
3687 controller trees are mounted into it. Note that the
3688 directories mounted beneath it are not read-only. This is a
3689 security measure and is particularly useful because glibc
3690 actually includes a search logic to pick any tmpfs it can
3691 find to implement shm_open() if /dev/shm is not available
3692 (which it might very well be in namespaced setups).
3694 * machinectl gained a new "poweroff" command to cleanly power
3695 down a local OS container.
3697 * The PrivateDevices= unit file setting will now also drop the
3698 CAP_MKNOD capability from the capability bound set, and
3699 imply DevicePolicy=closed.
3701 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
3702 comprehensively on all long-running systemd services where
3703 this is appropriate.
3705 * systemd-udevd will now run in a disassociated mount
3706 namespace. To mount directories from udev rules, make sure to
3707 pull in mount units via SYSTEMD_WANTS properties.
3709 * The kdbus support gained support for uploading policy into
3710 the kernel. sd-bus gained support for creating "monitoring"
3711 connections that can eavesdrop into all bus communication
3712 for debugging purposes.
3714 * Timestamps may now be specified in seconds since the UNIX
3715 epoch Jan 1st, 1970 by specifying "@" followed by the value
3718 * Native tcpwrap support in systemd has been removed. tcpwrap
3719 is old code, not really maintained anymore and has serious
3720 shortcomings, and better options such as firewalls
3721 exist. For setups that require tcpwrap usage, please
3722 consider invoking your socket-activated service via tcpd,
3723 like on traditional inetd.
3725 * A new system.conf configuration option
3726 DefaultTimerAccuracySec= has been added that controls the
3727 default AccuracySec= setting of .timer units.
3729 * Timer units gained a new WakeSystem= switch. If enabled,
3730 timers configured this way will cause the system to resume
3731 from system suspend (if the system supports that, which most
3734 * Timer units gained a new Persistent= switch. If enabled,
3735 timers configured this way will save to disk when they have
3736 been last triggered. This information is then used on next
3737 reboot to possible execute overdue timer events, that
3738 could not take place because the system was powered off.
3739 This enables simple anacron-like behaviour for timer units.
3741 * systemctl's "list-timers" will now also list the time a
3742 timer unit was last triggered in addition to the next time
3743 it will be triggered.
3745 * systemd-networkd will now assign predictable IPv4LL
3746 addresses to its local interfaces.
3748 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
3749 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
3750 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
3751 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
3752 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
3753 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
3754 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
3755 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
3758 — Berlin, 2014-03-25
3762 * A new unit file setting RestrictAddressFamilies= has been
3763 added to restrict which socket address families unit
3764 processes gain access to. This takes address family names
3765 like "AF_INET" or "AF_UNIX", and is useful to minimize the
3766 attack surface of services via exotic protocol stacks. This
3767 is built on seccomp system call filters.
3769 * Two new unit file settings RuntimeDirectory= and
3770 RuntimeDirectoryMode= have been added that may be used to
3771 manage a per-daemon runtime directories below /run. This is
3772 an alternative for setting up directory permissions with
3773 tmpfiles snippets, and has the advantage that the runtime
3774 directory's lifetime is bound to the daemon runtime and that
3775 the daemon starts up with an empty directory each time. This
3776 is particularly useful when writing services that drop
3777 privileges using the User= or Group= setting.
3779 * The DeviceAllow= unit setting now supports globbing for
3780 matching against device group names.
3782 * The systemd configuration file system.conf gained new
3783 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
3784 DefaultMemoryAccounting= to globally turn on/off accounting
3785 for specific resources (cgroups) for all units. These
3786 settings may still be overridden individually in each unit
3789 * systemd-gpt-auto-generator is now able to discover /srv and
3790 root partitions in addition to /home and swap partitions. It
3791 also supports LUKS-encrypted partitions now. With this in
3792 place, automatic discovery of partitions to mount following
3793 the Discoverable Partitions Specification
3794 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
3795 is now a lot more complete. This allows booting without
3796 /etc/fstab and without root= on the kernel command line on
3797 systems prepared appropriately.
3799 * systemd-nspawn gained a new --image= switch which allows
3800 booting up disk images and Linux installations on any block
3801 device that follow the Discoverable Partitions Specification
3802 (see above). This means that installations made with
3803 appropriately updated installers may now be started and
3804 deployed using container managers, completely
3805 unmodified. (We hope that libvirt-lxc will add support for
3806 this feature soon, too.)
3808 * systemd-nspawn gained a new --network-macvlan= setting to
3809 set up a private macvlan interface for the
3810 container. Similarly, systemd-networkd gained a new
3811 Kind=macvlan setting in .netdev files.
3813 * systemd-networkd now supports configuring local addresses
3816 * A new tool systemd-network-wait-online has been added to
3817 synchronously wait for network connectivity using
3820 * The sd-bus.h bus API gained a new sd_bus_track object for
3821 tracking the life-cycle of bus peers. Note that sd-bus.h is
3822 still not a public API though (unless you specify
3823 --enable-kdbus on the configure command line, which however
3824 voids your warranty and you get no API stability guarantee).
3826 * The $XDG_RUNTIME_DIR runtime directories for each user are
3827 now individual tmpfs instances, which has the benefit of
3828 introducing separate pools for each user, with individual
3829 size limits, and thus making sure that unprivileged clients
3830 can no longer negatively impact the system or other users by
3831 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
3832 RuntimeDirectorySize= has been introduced that allows
3833 controlling the default size limit for all users. It
3834 defaults to 10% of the available physical memory. This is no
3835 replacement for quotas on tmpfs though (which the kernel
3836 still does not support), as /dev/shm and /tmp are still
3837 shared resources used by both the system and unprivileged
3840 * logind will now automatically turn off automatic suspending
3841 on laptop lid close when more than one display is
3842 connected. This was previously expected to be implemented
3843 individually in desktop environments (such as GNOME),
3844 however has been added to logind now, in order to fix a
3845 boot-time race where a desktop environment might not have
3846 been started yet and thus not been able to take an inhibitor
3847 lock at the time where logind already suspends the system
3848 due to a closed lid.
3850 * logind will now wait at least 30s after each system
3851 suspend/resume cycle, and 3min after system boot before
3852 suspending the system due to a closed laptop lid. This
3853 should give USB docking stations and similar enough time to
3854 be probed and configured after system resume and boot in
3855 order to then act as suspend blocker.
3857 * systemd-run gained a new --property= setting which allows
3858 initialization of resource control properties (and others)
3859 for the created scope or service unit. Example: "systemd-run
3860 --property=BlockIOWeight=10 updatedb" may be used to run
3861 updatedb at a low block IO scheduling weight.
3863 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
3864 now also work in --scope mode.
3866 * When systemd is compiled with kdbus support, basic support
3867 for enforced policies is now in place. (Note that enabling
3868 kdbus still voids your warranty and no API compatibility
3871 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
3872 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
3873 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
3874 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
3875 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
3876 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
3877 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
3878 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
3879 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
3880 Zbigniew Jędrzejewski-Szmek
3882 — Berlin, 2014-03-12
3886 * systemd will now relabel /dev after loading the SMACK policy
3887 according to SMACK rules.
3889 * A new unit file option AppArmorProfile= has been added to
3890 set the AppArmor profile for the processes of a unit.
3892 * A new condition check ConditionArchitecture= has been added
3893 to conditionalize units based on the system architecture, as
3894 reported by uname()'s "machine" field.
3896 * systemd-networkd now supports matching on the system
3897 virtualization, architecture, kernel command line, host name
3900 * logind is now a lot more aggressive when suspending the
3901 machine due to a closed laptop lid. Instead of acting only
3902 on the lid close action, it will continuously watch the lid
3903 status and act on it. This is useful for laptops where the
3904 power button is on the outside of the chassis so that it can
3905 be reached without opening the lid (such as the Lenovo
3906 Yoga). On those machines, logind will now immediately
3907 re-suspend the machine if the power button has been
3908 accidentally pressed while the laptop was suspended and in a
3909 backpack or similar.
3911 * logind will now watch SW_DOCK switches and inhibit reaction
3912 to the lid switch if it is pressed. This means that logind
3913 will not suspend the machine anymore if the lid is closed
3914 and the system is docked, if the laptop supports SW_DOCK
3915 notifications via the input layer. Note that ACPI docking
3916 stations do not generate this currently. Also note that this
3917 logic is usually not fully sufficient and Desktop
3918 Environments should take a lid switch inhibitor lock when an
3919 external display is connected, as systemd will not watch
3922 * nspawn will now make use of the devices cgroup controller by
3923 default, and only permit creation of and access to the usual
3924 API device nodes like /dev/null or /dev/random, as well as
3925 access to (but not creation of) the pty devices.
3927 * We will now ship a default .network file for
3928 systemd-networkd that automatically configures DHCP for
3929 network interfaces created by nspawn's --network-veth or
3930 --network-bridge= switches.
3932 * systemd will now understand the usual M, K, G, T suffixes
3933 according to SI conventions (i.e. to the base 1000) when
3934 referring to throughput and hardware metrics. It will stay
3935 with IEC conventions (i.e. to the base 1024) for software
3936 metrics, according to what is customary according to
3937 Wikipedia. We explicitly document which base applies for
3938 each configuration option.
3940 * The DeviceAllow= setting in unit files now supports a syntax
3941 to whitelist an entire group of devices node majors at once,
3942 based on the /proc/devices listing. For example, with the
3943 string "char-pts", it is now possible to whitelist all
3944 current and future pseudo-TTYs at once.
3946 * sd-event learned a new "post" event source. Event sources of
3947 this type are triggered by the dispatching of any event
3948 source of a type that is not "post". This is useful for
3949 implementing clean-up and check event sources that are
3950 triggered by other work being done in the program.
3952 * systemd-networkd is no longer statically enabled, but uses
3953 the usual [Install] sections so that it can be
3954 enabled/disabled using systemctl. It still is enabled by
3957 * When creating a veth interface pair with systemd-nspawn, the
3958 host side will now be prefixed with "vb-" if
3959 --network-bridge= is used, and with "ve-" if --network-veth
3960 is used. This way, it is easy to distinguish these cases on
3961 the host, for example to apply different configuration to
3962 them with systemd-networkd.
3964 * The compatibility libraries for libsystemd-journal.so,
3965 libsystem-id128.so, libsystemd-login.so and
3966 libsystemd-daemon.so do not make use of IFUNC
3967 anymore. Instead, we now build libsystemd.so multiple times
3968 under these alternative names. This means that the footprint
3969 is drastically increased, but given that these are
3970 transitional compatibility libraries, this should not matter
3971 much. This change has been made necessary to support the ARM
3972 platform for these compatibility libraries, as the ARM
3973 toolchain is not really at the same level as the toolchain
3974 for other architectures like x86 and does not support
3975 IFUNC. Please make sure to use --enable-compat-libs only
3976 during a transitional period!
3978 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
3979 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
3980 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
3981 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
3982 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
3983 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
3984 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
3985 Zbigniew Jędrzejewski-Szmek
3987 — Berlin, 2014-02-24
3991 * A new component "systemd-networkd" has been added that can
3992 be used to configure local network interfaces statically or
3993 via DHCP. It is capable of bringing up bridges, VLANs, and
3994 bonding. Currently, no hook-ups for interactive network
3995 configuration are provided. Use this for your initrd,
3996 container, embedded, or server setup if you need a simple,
3997 yet powerful, network configuration solution. This
3998 configuration subsystem is quite nifty, as it allows wildcard
3999 hotplug matching in interfaces. For example, with a single
4000 configuration snippet, you can configure that all Ethernet
4001 interfaces showing up are automatically added to a bridge,
4002 or similar. It supports link-sensing and more.
4004 * A new tool "systemd-socket-proxyd" has been added which can
4005 act as a bidirectional proxy for TCP sockets. This is
4006 useful for adding socket activation support to services that
4007 do not actually support socket activation, including virtual
4008 machines and the like.
4010 * Add a new tool to save/restore rfkill state on
4013 * Save/restore state of keyboard backlights in addition to
4014 display backlights on shutdown/boot.
4016 * udev learned a new SECLABEL{} construct to label device
4017 nodes with a specific security label when they appear. For
4018 now, only SECLABEL{selinux} is supported, but the syntax is
4019 prepared for additional security frameworks.
4021 * udev gained a new scheme to configure link-level attributes
4022 from files in /etc/systemd/network/*.link. These files can
4023 match against MAC address, device path, driver name and type,
4024 and will apply attributes like the naming policy, link speed,
4025 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
4026 address assignment policy (randomized, ...).
4028 * The configuration of network interface naming rules for
4029 "permanent interface names" has changed: a new NamePolicy=
4030 setting in the [Link] section of .link files determines the
4031 priority of possible naming schemes (onboard, slot, MAC,
4032 path). The default value of this setting is determined by
4033 /usr/lib/net/links/99-default.link. Old
4034 80-net-name-slot.rules udev configuration file has been
4035 removed, so local configuration overriding this file should
4036 be adapted to override 99-default.link instead.
4038 * When the User= switch is used in a unit file, also
4039 initialize $SHELL= based on the user database entry.
4041 * systemd no longer depends on libdbus. All communication is
4042 now done with sd-bus, systemd's low-level bus library
4045 * kdbus support has been added to PID 1 itself. When kdbus is
4046 enabled, this causes PID 1 to set up the system bus and
4047 enable support for a new ".busname" unit type that
4048 encapsulates bus name activation on kdbus. It works a little
4049 bit like ".socket" units, except for bus names. A new
4050 generator has been added that converts classic dbus1 service
4051 activation files automatically into native systemd .busname
4054 * sd-bus: add a light-weight vtable implementation that allows
4055 defining objects on the bus with a simple static const
4056 vtable array of its methods, signals and properties.
4058 * systemd will not generate or install static dbus
4059 introspection data anymore to /usr/share/dbus-1/interfaces,
4060 as the precise format of these files is unclear, and
4061 nothing makes use of it.
4063 * A proxy daemon is now provided to proxy clients connecting
4064 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
4065 compatibility with classic D-Bus.
4067 * A bus driver implementation has been added that supports the
4068 classic D-Bus bus driver calls on kdbus, also for
4069 compatibility purposes.
4071 * A new API "sd-event.h" has been added that implements a
4072 minimal event loop API built around epoll. It provides a
4073 couple of features that direct epoll usage is lacking:
4074 prioritization of events, scales to large numbers of timer
4075 events, per-event timer slack (accuracy), system-wide
4076 coalescing of timer events, exit handlers, watchdog
4077 supervision support using systemd's sd_notify() API, child
4080 * A new API "sd-rntl.h" has been added that provides an API
4081 around the route netlink interface of the kernel, similar in
4082 style to "sd-bus.h".
4084 * A new API "sd-dhcp-client.h" has been added that provides a
4085 small DHCPv4 client-side implementation. This is used by
4088 * There is a new kernel command line option
4089 "systemd.restore_state=0|1". When set to "0", none of the
4090 systemd tools will restore saved runtime state to hardware
4091 devices. More specifically, the rfkill and backlight states
4094 * The FsckPassNo= compatibility option in mount/service units
4095 has been removed. The fstab generator will now add the
4096 necessary dependencies automatically, and does not require
4097 PID1's support for that anymore.
4099 * journalctl gained a new switch, --list-boots, that lists
4100 recent boots with their times and boot IDs.
4102 * The various tools like systemctl, loginctl, timedatectl,
4103 busctl, systemd-run, ... have gained a new switch "-M" to
4104 connect to a specific, local OS container (as direct
4105 connection, without requiring SSH). This works on any
4106 container that is registered with machined, such as those
4107 created by libvirt-lxc or nspawn.
4109 * systemd-run and systemd-analyze also gained support for "-H"
4110 to connect to remote hosts via SSH. This is particularly
4111 useful for systemd-run because it enables queuing of jobs
4112 onto remote systems.
4114 * machinectl gained a new command "login" to open a getty
4115 login in any local container. This works with any container
4116 that is registered with machined (such as those created by
4117 libvirt-lxc or nspawn), and which runs systemd inside.
4119 * machinectl gained a new "reboot" command that may be used to
4120 trigger a reboot on a specific container that is registered
4121 with machined. This works on any container that runs an init
4122 system of some kind.
4124 * systemctl gained a new "list-timers" command to print a nice
4125 listing of installed timer units with the times they elapse
4128 * Alternative reboot() parameters may now be specified on the
4129 "systemctl reboot" command line and are passed to the
4130 reboot() system call.
4132 * systemctl gained a new --job-mode= switch to configure the
4133 mode to queue a job with. This is a more generic version of
4134 --fail, --irreversible, and --ignore-dependencies, which are
4135 still available but not advertised anymore.
4137 * /etc/systemd/system.conf gained new settings to configure
4138 various default timeouts of units, as well as the default
4139 start limit interval and burst. These may still be overridden
4142 * PID1 will now export on the bus profile data of the security
4143 policy upload process (such as the SELinux policy upload to
4146 * journald: when forwarding logs to the console, include
4147 timestamps (following the setting in
4148 /sys/module/printk/parameters/time).
4150 * OnCalendar= in timer units now understands the special
4151 strings "yearly" and "annually". (Both are equivalent)
4153 * The accuracy of timer units is now configurable with the new
4154 AccuracySec= setting. It defaults to 1min.
4156 * A new dependency type JoinsNamespaceOf= has been added that
4157 allows running two services within the same /tmp and network
4158 namespace, if PrivateNetwork= or PrivateTmp= are used.
4160 * A new command "cat" has been added to systemctl. It outputs
4161 the original unit file of a unit, and concatenates the
4162 contents of additional "drop-in" unit file snippets, so that
4163 the full configuration is shown.
4165 * systemctl now supports globbing on the various "list-xyz"
4166 commands, like "list-units" or "list-sockets", as well as on
4167 those commands which take multiple unit names.
4169 * journalctl's --unit= switch gained support for globbing.
4171 * All systemd daemons now make use of the watchdog logic so
4172 that systemd automatically notices when they hang.
4174 * If the $container_ttys environment variable is set,
4175 getty-generator will automatically spawn a getty for each
4176 listed tty. This is useful for container managers to request
4177 login gettys to be spawned on as many ttys as needed.
4179 * %h, %s, %U specifier support is not available anymore when
4180 used in unit files for PID 1. This is because NSS calls are
4181 not safe from PID 1. They stay available for --user
4182 instances of systemd, and as special case for the root user.
4184 * loginctl gained a new "--no-legend" switch to turn off output
4187 * The "sd-login.h" API gained three new calls:
4188 sd_session_is_remote(), sd_session_get_remote_user(),
4189 sd_session_get_remote_host() to query information about
4192 * The udev hardware database now also carries vendor/product
4193 information of SDIO devices.
4195 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
4196 determine whether watchdog notifications are requested by
4199 * Socket-activated per-connection services now include a
4200 short description of the connection parameters in the
4203 * tmpfiles gained a new "--boot" option. When this is not used,
4204 only lines where the command character is not suffixed with
4205 "!" are executed. When this option is specified, those
4206 options are executed too. This partitions tmpfiles
4207 directives into those that can be safely executed at any
4208 time, and those which should be run only at boot (for
4209 example, a line that creates /run/nologin).
4211 * A new API "sd-resolve.h" has been added which provides a simple
4212 asynchronous wrapper around glibc NSS host name resolution
4213 calls, such as getaddrinfo(). In contrast to glibc's
4214 getaddrinfo_a(), it does not use signals. In contrast to most
4215 other asynchronous name resolution libraries, this one does
4216 not reimplement DNS, but reuses NSS, so that alternate
4217 host name resolution systems continue to work, such as mDNS,
4218 LDAP, etc. This API is based on libasyncns, but it has been
4219 cleaned up for inclusion in systemd.
4221 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
4222 "sd-daemon.h" are no longer found in individual libraries
4223 libsystemd-journal.so, libsystemd-login.so,
4224 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
4225 merged them into a single library, libsystemd.so, which
4226 provides all symbols. The reason for this is cyclic
4227 dependencies, as these libraries tend to use each other's
4228 symbols. So far, we have managed to workaround that by linking
4229 a copy of a good part of our code into each of these
4230 libraries again and again, which, however, makes certain
4231 things hard to do, like sharing static variables. Also, it
4232 substantially increases footprint. With this change, there
4233 is only one library for the basic APIs systemd
4234 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
4235 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
4236 library as well, however are subject to the --enable-kdbus
4237 switch (see below). Note that "sd-dhcp-client.h" is not part
4238 of this library (this is because it only consumes, never
4239 provides, services of/to other APIs). To make the transition
4240 easy from the separate libraries to the unified one, we
4241 provide the --enable-compat-libs compile-time switch which
4242 will generate stub libraries that are compatible with the
4243 old ones but redirect all calls to the new one.
4245 * All of the kdbus logic and the new APIs "sd-bus.h",
4246 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
4247 and "sd-utf8.h" are compile-time optional via the
4248 "--enable-kdbus" switch, and they are not compiled in by
4249 default. To make use of kdbus, you have to explicitly enable
4250 the switch. Note however, that neither the kernel nor the
4251 userspace API for all of this is considered stable yet. We
4252 want to maintain the freedom to still change the APIs for
4253 now. By specifying this build-time switch, you acknowledge
4254 that you are aware of the instability of the current
4257 * Also, note that while kdbus is pretty much complete,
4258 it lacks one thing: proper policy support. This means you
4259 can build a fully working system with all features; however,
4260 it will be highly insecure. Policy support will be added in
4261 one of the next releases, at the same time that we will
4262 declare the APIs stable.
4264 * When the kernel command line argument "kdbus" is specified,
4265 systemd will automatically load the kdbus.ko kernel module. At
4266 this stage of development, it is only useful for testing kdbus
4267 and should not be used in production. Note: if "--enable-kdbus"
4268 is specified, and the kdbus.ko kernel module is available, and
4269 "kdbus" is added to the kernel command line, the entire system
4270 runs with kdbus instead of dbus-daemon, with the above mentioned
4271 problem of missing the system policy enforcement. Also a future
4272 version of kdbus.ko or a newer systemd will not be compatible with
4273 each other, and will unlikely be able to boot the machine if only
4274 one of them is updated.
4276 * systemctl gained a new "import-environment" command which
4277 uploads the caller's environment (or parts thereof) into the
4278 service manager so that it is inherited by services started
4279 by the manager. This is useful to upload variables like
4280 $DISPLAY into the user service manager.
4282 * A new PrivateDevices= switch has been added to service units
4283 which allows running a service with a namespaced /dev
4284 directory that does not contain any device nodes for
4285 physical devices. More specifically, it only includes devices
4286 such as /dev/null, /dev/urandom, and /dev/zero which are API
4289 * logind has been extended to support behaviour like VT
4290 switching on seats that do not support a VT. This makes
4291 multi-session available on seats that are not the first seat
4292 (seat0), and on systems where kernel support for VTs has
4293 been disabled at compile-time.
4295 * If a process holds a delay lock for system sleep or shutdown
4296 and fails to release it in time, we will now log its
4297 identity. This makes it easier to identify processes that
4298 cause slow suspends or power-offs.
4300 * When parsing /etc/crypttab, support for a new key-slot=
4301 option as supported by Debian is added. It allows indicating
4302 which LUKS slot to use on disk, speeding up key loading.
4304 * The sd_journald_sendv() API call has been checked and
4305 officially declared to be async-signal-safe so that it may
4306 be invoked from signal handlers for logging purposes.
4308 * Boot-time status output is now enabled automatically after a
4309 short timeout if boot does not progress, in order to give
4310 the user an indication what she or he is waiting for.
4312 * The boot-time output has been improved to show how much time
4313 remains until jobs expire.
4315 * The KillMode= switch in service units gained a new possible
4316 value "mixed". If set, and the unit is shut down, then the
4317 initial SIGTERM signal is sent only to the main daemon
4318 process, while the following SIGKILL signal is sent to
4319 all remaining processes of the service.
4321 * When a scope unit is registered, a new property "Controller"
4322 may be set. If set to a valid bus name, systemd will send a
4323 RequestStop() signal to this name when it would like to shut
4324 down the scope. This may be used to hook manager logic into
4325 the shutdown logic of scope units. Also, scope units may now
4326 be put in a special "abandoned" state, in which case the
4327 manager process which created them takes no further
4328 responsibilities for it.
4330 * When reading unit files, systemd will now verify
4331 the access mode of these files, and warn about certain
4332 suspicious combinations. This has been added to make it
4333 easier to track down packaging bugs where unit files are
4334 marked executable or world-writable.
4336 * systemd-nspawn gained a new "--setenv=" switch to set
4337 container-wide environment variables. The similar option in
4338 systemd-activate was renamed from "--environment=" to
4339 "--setenv=" for consistency.
4341 * systemd-nspawn has been updated to create a new kdbus domain
4342 for each container that is invoked, thus allowing each
4343 container to have its own set of system and user buses,
4344 independent of the host.
4346 * systemd-nspawn gained a new --drop-capability= switch to run
4347 the container with less capabilities than the default. Both
4348 --drop-capability= and --capability= now take the special
4349 string "all" for dropping or keeping all capabilities.
4351 * systemd-nspawn gained new switches for executing containers
4352 with specific SELinux labels set.
4354 * systemd-nspawn gained a new --quiet switch to not generate
4355 any additional output but the container's own console
4358 * systemd-nspawn gained a new --share-system switch to run a
4359 container without PID namespacing enabled.
4361 * systemd-nspawn gained a new --register= switch to control
4362 whether the container is registered with systemd-machined or
4363 not. This is useful for containers that do not run full
4364 OS images, but only specific apps.
4366 * systemd-nspawn gained a new --keep-unit which may be used
4367 when invoked as the only program from a service unit, and
4368 results in registration of the unit service itself in
4369 systemd-machined, instead of a newly opened scope unit.
4371 * systemd-nspawn gained a new --network-interface= switch for
4372 moving arbitrary interfaces to the container. The new
4373 --network-veth switch creates a virtual Ethernet connection
4374 between host and container. The new --network-bridge=
4375 switch then allows assigning the host side of this virtual
4376 Ethernet connection to a bridge device.
4378 * systemd-nspawn gained a new --personality= switch for
4379 setting the kernel personality for the container. This is
4380 useful when running a 32-bit container on a 64-bit host. A
4381 similar option Personality= is now also available for service
4384 * logind will now also track a "Desktop" identifier for each
4385 session which encodes the desktop environment of it. This is
4386 useful for desktop environments that want to identify
4387 multiple running sessions of itself easily.
4389 * A new SELinuxContext= setting for service units has been
4390 added that allows setting a specific SELinux execution
4391 context for a service.
4393 * Most systemd client tools will now honour $SYSTEMD_LESS for
4394 settings of the "less" pager. By default, these tools will
4395 override $LESS to allow certain operations to work, such as
4396 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
4397 influence this logic.
4399 * systemd's "seccomp" hook-up has been changed to make use of
4400 the libseccomp library instead of using its own
4401 implementation. This has benefits for portability among
4404 * For usage together with SystemCallFilter=, a new
4405 SystemCallErrorNumber= setting has been introduced that
4406 allows configuration of a system error number to be returned
4407 on filtered system calls, instead of immediately killing the
4408 process. Also, SystemCallArchitectures= has been added to
4409 limit access to system calls of a particular architecture
4410 (in order to turn off support for unused secondary
4411 architectures). There is also a global
4412 SystemCallArchitectures= setting in system.conf now to turn
4413 off support for non-native system calls system-wide.
4415 * systemd requires a kernel with a working name_to_handle_at(),
4416 please see the kernel config requirements in the README file.
4418 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
4419 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
4420 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
4421 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
4422 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
4423 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
4424 Elia Pinto, Florian Weimer, George McCollister, Goffredo
4425 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
4426 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
4427 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
4428 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
4429 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
4430 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
4431 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
4432 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
4433 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
4434 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
4435 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
4436 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
4437 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
4438 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
4439 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
4440 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
4441 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
4443 — Berlin, 2014-02-20
4447 * logind has gained support for facilitating privileged input
4448 and drm device access for unprivileged clients. This work is
4449 useful to allow Wayland display servers (and similar
4450 programs, such as kmscon) to run under the user's ID and
4451 access input and drm devices which are normally
4452 protected. When this is used (and the kernel is new enough)
4453 logind will "mute" IO on the file descriptors passed to
4454 Wayland as long as it is in the background and "unmute" it
4455 if it returns into the foreground. This allows secure
4456 session switching without allowing background sessions to
4457 eavesdrop on input and display data. This also introduces
4458 session switching support if VT support is turned off in the
4459 kernel, and on seats that are not seat0.
4461 * A new kernel command line option luks.options= is understood
4462 now which allows specifying LUKS options for usage for LUKS
4463 encrypted partitions specified with luks.uuid=.
4465 * tmpfiles.d(5) snippets may now use specifier expansion in
4466 path names. More specifically %m, %b, %H, %v, are now
4467 replaced by the local machine id, boot id, hostname, and
4468 kernel version number.
4470 * A new tmpfiles.d(5) command "m" has been introduced which
4471 may be used to change the owner/group/access mode of a file
4472 or directory if it exists, but do nothing if it does not.
4474 * This release removes high-level support for the
4475 MemorySoftLimit= cgroup setting. The underlying kernel
4476 cgroup attribute memory.soft_limit= is currently badly
4477 designed and likely to be removed from the kernel API in its
4478 current form, hence we should not expose it for now.
4480 * The memory.use_hierarchy cgroup attribute is now enabled for
4481 all cgroups systemd creates in the memory cgroup
4482 hierarchy. This option is likely to be come the built-in
4483 default in the kernel anyway, and the non-hierarchical mode
4484 never made much sense in the intrinsically hierarchical
4487 * A new field _SYSTEMD_SLICE= is logged along with all journal
4488 messages containing the slice a message was generated
4489 from. This is useful to allow easy per-customer filtering of
4490 logs among other things.
4492 * systemd-journald will no longer adjust the group of journal
4493 files it creates to the "systemd-journal" group. Instead we
4494 rely on the journal directory to be owned by the
4495 "systemd-journal" group, and its setgid bit set, so that the
4496 kernel file system layer will automatically enforce that
4497 journal files inherit this group assignment. The reason for
4498 this change is that we cannot allow NSS look-ups from
4499 journald which would be necessary to resolve
4500 "systemd-journal" to a numeric GID, because this might
4501 create deadlocks if NSS involves synchronous queries to
4502 other daemons (such as nscd, or sssd) which in turn are
4503 logging clients of journald and might block on it, which
4504 would then dead lock. A tmpfiles.d(5) snippet included in
4505 systemd will make sure the setgid bit and group are
4506 properly set on the journal directory if it exists on every
4507 boot. However, we recommend adjusting it manually after
4508 upgrades too (or from RPM scriptlets), so that the change is
4509 not delayed until next reboot.
4511 * Backlight and random seed files in /var/lib/ have moved into
4512 the /var/lib/systemd/ directory, in order to centralize all
4513 systemd generated files in one directory.
4515 * Boot time performance measurements (as displayed by
4516 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
4517 performance information if that's available to determine how
4518 much time BIOS and boot loader initialization required. With
4519 a sufficiently new BIOS you hence no longer need to boot
4520 with Gummiboot to get access to such information.
4522 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
4523 Cristian Rodríguez, Dave Reisner, David Herrmann, David
4524 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
4525 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
4526 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
4527 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
4528 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
4530 — Berlin, 2013-10-02
4534 * The Restart= option for services now understands a new
4535 on-watchdog setting, which will restart the service
4536 automatically if the service stops sending out watchdog keep
4537 alive messages (as configured with WatchdogSec=).
4539 * The getty generator (which is responsible for bringing up a
4540 getty on configured serial consoles) will no longer only
4541 start a getty on the primary kernel console but on all
4542 others, too. This makes the order in which console= is
4543 specified on the kernel command line less important.
4545 * libsystemd-logind gained a new sd_session_get_vt() call to
4546 retrieve the VT number of a session.
4548 * If the option "tries=0" is set for an entry of /etc/crypttab
4549 its passphrase is queried indefinitely instead of any
4550 maximum number of tries.
4552 * If a service with a configure PID file terminates its PID
4553 file will now be removed automatically if it still exists
4554 afterwards. This should put an end to stale PID files.
4556 * systemd-run will now also take relative binary path names
4557 for execution and no longer insists on absolute paths.
4559 * InaccessibleDirectories= and ReadOnlyDirectories= now take
4560 paths that are optionally prefixed with "-" to indicate that
4561 it should not be considered a failure if they do not exist.
4563 * journalctl -o (and similar commands) now understands a new
4564 output mode "short-precise", it is similar to "short" but
4565 shows timestamps with usec accuracy.
4567 * The option "discard" (as known from Debian) is now
4568 synonymous to "allow-discards" in /etc/crypttab. In fact,
4569 "discard" is preferred now (since it is easier to remember
4572 * Some licensing clean-ups were made, so that more code is now
4573 LGPL-2.1 licensed than before.
4575 * A minimal tool to save/restore the display backlight
4576 brightness across reboots has been added. It will store the
4577 backlight setting as late as possible at shutdown, and
4578 restore it as early as possible during reboot.
4580 * A logic to automatically discover and enable home and swap
4581 partitions on GPT disks has been added. With this in place
4582 /etc/fstab becomes optional for many setups as systemd can
4583 discover certain partitions located on the root disk
4584 automatically. Home partitions are recognized under their
4585 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
4586 partitions are recognized under their GPT type ID
4587 0657fd6da4ab43c484e50933c84b4f4f.
4589 * systemd will no longer pass any environment from the kernel
4590 or initrd to system services. If you want to set an
4591 environment for all services, do so via the kernel command
4592 line systemd.setenv= assignment.
4594 * The systemd-sysctl tool no longer natively reads the file
4595 /etc/sysctl.conf. If desired, the file should be symlinked
4596 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
4597 legacy support by a symlink rather than built-in code, it
4598 also makes the otherwise hidden order of application of the
4599 different files visible. (Note that this partly reverts to a
4600 pre-198 application order of sysctl knobs!)
4602 * The "systemctl set-log-level" and "systemctl dump" commands
4603 have been moved to systemd-analyze.
4605 * systemd-run learned the new --remain-after-exit switch,
4606 which causes the scope unit not to be cleaned up
4607 automatically after the process terminated.
4609 * tmpfiles learned a new --exclude-prefix= switch to exclude
4610 certain paths from operation.
4612 * journald will now automatically flush all messages to disk
4613 as soon as a message at the log level CRIT, ALERT or EMERG
4616 Contributions from: Andrew Cook, Brandon Philips, Christian
4617 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
4618 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
4619 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
4620 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
4621 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
4622 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
4623 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
4624 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
4625 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
4626 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
4627 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
4628 William Giokas, Zbigniew Jędrzejewski-Szmek
4630 — Berlin, 2013-09-13
4634 * The documentation has been updated to cover the various new
4635 concepts introduced with 205.
4637 * Unit files now understand the new %v specifier which
4638 resolves to the kernel version string as returned by "uname
4641 * systemctl now supports filtering the unit list output by
4642 load state, active state and sub state, using the new
4645 * "systemctl status" will now show the results of the
4646 condition checks (like ConditionPathExists= and similar) of
4647 the last start attempts of the unit. They are also logged to
4650 * "journalctl -b" may now be used to look for boot output of a
4651 specific boot. Try "journalctl -b -1" for the previous boot,
4652 but the syntax is substantially more powerful.
4654 * "journalctl --show-cursor" has been added which prints the
4655 cursor string the last shown log line. This may then be used
4656 with the new "journalctl --after-cursor=" switch to continue
4657 browsing logs from that point on.
4659 * "journalctl --force" may now be used to force regeneration
4662 * Creation of "dead" device nodes has been moved from udev
4663 into kmod and tmpfiles. Previously, udev would read the kmod
4664 databases to pre-generate dead device nodes based on meta
4665 information contained in kernel modules, so that these would
4666 be auto-loaded on access rather then at boot. As this
4667 does not really have much to do with the exposing actual
4668 kernel devices to userspace this has always been slightly
4669 alien in the udev codebase. Following the new scheme kmod
4670 will now generate a runtime snippet for tmpfiles from the
4671 module meta information and it now is tmpfiles' job to the
4672 create the nodes. This also allows overriding access and
4673 other parameters for the nodes using the usual tmpfiles
4674 facilities. As side effect this allows us to remove the
4675 CAP_SYS_MKNOD capability bit from udevd entirely.
4677 * logind's device ACLs may now be applied to these "dead"
4678 devices nodes too, thus finally allowing managed access to
4679 devices such as /dev/snd/sequencer without loading the
4680 backing module right-away.
4682 * A new RPM macro has been added that may be used to apply
4683 tmpfiles configuration during package installation.
4685 * systemd-detect-virt and ConditionVirtualization= now can
4686 detect User-Mode-Linux machines (UML).
4688 * journald will now implicitly log the effective capabilities
4689 set of processes in the message metadata.
4691 * systemd-cryptsetup has gained support for TrueCrypt volumes.
4693 * The initrd interface has been simplified (more specifically,
4694 support for passing performance data via environment
4695 variables and fsck results via files in /run has been
4696 removed). These features were non-essential, and are
4697 nowadays available in a much nicer way by having systemd in
4698 the initrd serialize its state and have the hosts systemd
4699 deserialize it again.
4701 * The udev "keymap" data files and tools to apply keyboard
4702 specific mappings of scan to key codes, and force-release
4703 scan code lists have been entirely replaced by a udev
4704 "keyboard" builtin and a hwdb data file.
4706 * systemd will now honour the kernel's "quiet" command line
4707 argument also during late shutdown, resulting in a
4708 completely silent shutdown when used.
4710 * There's now an option to control the SO_REUSEPORT socket
4711 option in .socket units.
4713 * Instance units will now automatically get a per-template
4714 subslice of system.slice unless something else is explicitly
4715 configured. For example, instances of sshd@.service will now
4716 implicitly be placed in system-sshd.slice rather than
4717 system.slice as before.
4719 * Test coverage support may now be enabled at build time.
4721 Contributions from: Dave Reisner, Frederic Crozat, Harald
4722 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
4723 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
4724 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
4725 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
4726 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
4727 Giokas, Zbigniew Jędrzejewski-Szmek
4729 — Berlin, 2013-07-23
4733 * Two new unit types have been introduced:
4735 Scope units are very similar to service units, however, are
4736 created out of pre-existing processes — instead of PID 1
4737 forking off the processes. By using scope units it is
4738 possible for system services and applications to group their
4739 own child processes (worker processes) in a powerful way
4740 which then maybe used to organize them, or kill them
4741 together, or apply resource limits on them.
4743 Slice units may be used to partition system resources in an
4744 hierarchical fashion and then assign other units to them. By
4745 default there are now three slices: system.slice (for all
4746 system services), user.slice (for all user sessions),
4747 machine.slice (for VMs and containers).
4749 Slices and scopes have been introduced primarily in
4750 context of the work to move cgroup handling to a
4751 single-writer scheme, where only PID 1
4752 creates/removes/manages cgroups.
4754 * There's a new concept of "transient" units. In contrast to
4755 normal units these units are created via an API at runtime,
4756 not from configuration from disk. More specifically this
4757 means it is now possible to run arbitrary programs as
4758 independent services, with all execution parameters passed
4759 in via bus APIs rather than read from disk. Transient units
4760 make systemd substantially more dynamic then it ever was,
4761 and useful as a general batch manager.
4763 * logind has been updated to make use of scope and slice units
4764 for managing user sessions. As a user logs in he will get
4765 his own private slice unit, to which all sessions are added
4766 as scope units. We also added support for automatically
4767 adding an instance of user@.service for the user into the
4768 slice. Effectively logind will no longer create cgroup
4769 hierarchies on its own now, it will defer entirely to PID 1
4770 for this by means of scope, service and slice units. Since
4771 user sessions this way become entities managed by PID 1
4772 the output of "systemctl" is now a lot more comprehensive.
4774 * A new mini-daemon "systemd-machined" has been added which
4775 may be used by virtualization managers to register local
4776 VMs/containers. nspawn has been updated accordingly, and
4777 libvirt will be updated shortly. machined will collect a bit
4778 of meta information about the VMs/containers, and assign
4779 them their own scope unit (see above). The collected
4780 meta-data is then made available via the "machinectl" tool,
4781 and exposed in "ps" and similar tools. machined/machinectl
4782 is compile-time optional.
4784 * As discussed earlier, the low-level cgroup configuration
4785 options ControlGroup=, ControlGroupModify=,
4786 ControlGroupPersistent=, ControlGroupAttribute= have been
4787 removed. Please use high-level attribute settings instead as
4788 well as slice units.
4790 * A new bus call SetUnitProperties() has been added to alter
4791 various runtime parameters of a unit. This is primarily
4792 useful to alter cgroup parameters dynamically in a nice way,
4793 but will be extended later on to make more properties
4794 modifiable at runtime. systemctl gained a new set-properties
4795 command that wraps this call.
4797 * A new tool "systemd-run" has been added which can be used to
4798 run arbitrary command lines as transient services or scopes,
4799 while configuring a number of settings via the command
4800 line. This tool is currently very basic, however already
4801 very useful. We plan to extend this tool to even allow
4802 queuing of execution jobs with time triggers from the
4803 command line, similar in fashion to "at".
4805 * nspawn will now inform the user explicitly that kernels with
4806 audit enabled break containers, and suggest the user to turn
4809 * Support for detecting the IMA and AppArmor security
4810 frameworks with ConditionSecurity= has been added.
4812 * journalctl gained a new "-k" switch for showing only kernel
4813 messages, mimicking dmesg output; in addition to "--user"
4814 and "--system" switches for showing only user's own logs
4817 * systemd-delta can now show information about drop-in
4818 snippets extending unit files.
4820 * libsystemd-bus has been substantially updated but is still
4821 not available as public API.
4823 * systemd will now look for the "debug" argument on the kernel
4824 command line and enable debug logging, similar to what
4825 "systemd.log_level=debug" already did before.
4827 * "systemctl set-default", "systemctl get-default" has been
4828 added to configure the default.target symlink, which
4829 controls what to boot into by default.
4831 * "systemctl set-log-level" has been added as a convenient
4832 way to raise and lower systemd logging threshold.
4834 * "systemd-analyze plot" will now show the time the various
4835 generators needed for execution, as well as information
4836 about the unit file loading.
4838 * libsystemd-journal gained a new sd_journal_open_files() call
4839 for opening specific journal files. journactl also gained a
4840 new switch to expose this new functionality. Previously we
4841 only supported opening all files from a directory, or all
4842 files from the system, as opening individual files only is
4843 racy due to journal file rotation.
4845 * systemd gained the new DefaultEnvironment= setting in
4846 /etc/systemd/system.conf to set environment variables for
4849 * If a privileged process logs a journal message with the
4850 OBJECT_PID= field set, then journald will automatically
4851 augment this with additional OBJECT_UID=, OBJECT_GID=,
4852 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
4853 system services want to log events about specific client
4854 processes. journactl/systemctl has been updated to make use
4855 of this information if all log messages regarding a specific
4858 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
4859 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
4860 Reisner, David Coppa, David King, David Strauss, Eelco
4861 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
4862 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
4863 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
4864 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
4865 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
4866 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
4867 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
4868 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
4869 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
4870 Łukasz Stelmach, 장동준
4874 * The Python bindings gained some minimal support for the APIs
4875 exposed by libsystemd-logind.
4877 * ConditionSecurity= gained support for detecting SMACK. Since
4878 this condition already supports SELinux and AppArmor we only
4879 miss IMA for this. Patches welcome!
4881 Contributions from: Karol Lewandowski, Lennart Poettering,
4882 Zbigniew Jędrzejewski-Szmek
4886 * systemd-nspawn will now create /etc/resolv.conf if
4887 necessary, before bind-mounting the host's file onto it.
4889 * systemd-nspawn will now store meta information about a
4890 container on the container's cgroup as extended attribute
4891 fields, including the root directory.
4893 * The cgroup hierarchy has been reworked in many ways. All
4894 objects any of the components systemd creates in the cgroup
4895 tree are now suffixed. More specifically, user sessions are
4896 now placed in cgroups suffixed with ".session", users in
4897 cgroups suffixed with ".user", and nspawn containers in
4898 cgroups suffixed with ".nspawn". Furthermore, all cgroup
4899 names are now escaped in a simple scheme to avoid collision
4900 of userspace object names with kernel filenames. This work
4901 is preparation for making these objects relocatable in the
4902 cgroup tree, in order to allow easy resource partitioning of
4903 these objects without causing naming conflicts.
4905 * systemctl list-dependencies gained the new switches
4906 --plain, --reverse, --after and --before.
4908 * systemd-inhibit now shows the process name of processes that
4909 have taken an inhibitor lock.
4911 * nss-myhostname will now also resolve "localhost"
4912 implicitly. This makes /etc/hosts an optional file and
4913 nicely handles that on IPv6 ::1 maps to both "localhost" and
4916 * libsystemd-logind.so gained a new call
4917 sd_get_machine_names() to enumerate running containers and
4918 VMs (currently only supported by very new libvirt and
4919 nspawn). sd_login_monitor can now be used to watch
4920 VMs/containers coming and going.
4922 * .include is not allowed recursively anymore, and only in
4923 unit files. Usually it is better to use drop-in snippets in
4924 .d/*.conf anyway, as introduced with systemd 198.
4926 * systemd-analyze gained a new "critical-chain" command that
4927 determines the slowest chain of units run during system
4928 boot-up. It is very useful for tracking down where
4929 optimizing boot time is the most beneficial.
4931 * systemd will no longer allow manipulating service paths in
4932 the name=systemd:/system cgroup tree using ControlGroup= in
4933 units. (But is still fine with it in all other dirs.)
4935 * There's a new systemd-nspawn@.service service file that may
4936 be used to easily run nspawn containers as system
4937 services. With the container's root directory in
4938 /var/lib/container/foobar it is now sufficient to run
4939 "systemctl start systemd-nspawn@foobar.service" to boot it.
4941 * systemd-cgls gained a new parameter "--machine" to list only
4942 the processes within a certain container.
4944 * ConditionSecurity= now can check for "apparmor". We still
4945 are lacking checks for SMACK and IMA for this condition
4946 check though. Patches welcome!
4948 * A new configuration file /etc/systemd/sleep.conf has been
4949 added that may be used to configure which kernel operation
4950 systemd is supposed to execute when "suspend", "hibernate"
4951 or "hybrid-sleep" is requested. This makes the new kernel
4952 "freeze" state accessible to the user.
4954 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
4955 the passed argument if applicable.
4957 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
4958 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
4959 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
4960 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
4961 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
4962 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
4963 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
4968 * The output of 'systemctl list-jobs' got some polishing. The
4969 '--type=' argument may now be passed more than once. A new
4970 command 'systemctl list-sockets' has been added which shows
4971 a list of kernel sockets systemd is listening on with the
4972 socket units they belong to, plus the units these socket
4975 * The experimental libsystemd-bus library got substantial
4976 updates to work in conjunction with the (also experimental)
4977 kdbus kernel project. It works well enough to exchange
4978 messages with some sophistication. Note that kdbus is not
4979 ready yet, and the library is mostly an elaborate test case
4980 for now, and not installable.
4982 * systemd gained a new unit 'systemd-static-nodes.service'
4983 that generates static device nodes earlier during boot, and
4984 can run in conjunction with udev.
4986 * libsystemd-login gained a new call sd_pid_get_user_unit()
4987 to retrieve the user systemd unit a process is running
4988 in. This is useful for systems where systemd is used as
4991 * systemd-nspawn now places all containers in the new /machine
4992 top-level cgroup directory in the name=systemd
4993 hierarchy. libvirt will soon do the same, so that we get a
4994 uniform separation of /system, /user and /machine for system
4995 services, user processes and containers/virtual
4996 machines. This new cgroup hierarchy is also useful to stick
4997 stable names to specific container instances, which can be
4998 recognized later this way (this name may be controlled
4999 via systemd-nspawn's new -M switch). libsystemd-login also
5000 gained a new call sd_pid_get_machine_name() to retrieve the
5001 name of the container/VM a specific process belongs to.
5003 * bootchart can now store its data in the journal.
5005 * libsystemd-journal gained a new call
5006 sd_journal_add_conjunction() for AND expressions to the
5007 matching logic. This can be used to express more complex
5008 logical expressions.
5010 * journactl can now take multiple --unit= and --user-unit=
5013 * The cryptsetup logic now understands the "luks.key=" kernel
5014 command line switch for specifying a file to read the
5015 decryption key from. Also, if a configured key file is not
5016 found the tool will now automatically fall back to prompting
5019 * Python systemd.journal module was updated to wrap recently
5020 added functions from libsystemd-journal. The interface was
5021 changed to bring the low level interface in s.j._Reader
5022 closer to the C API, and the high level interface in
5023 s.j.Reader was updated to wrap and convert all data about
5026 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
5027 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
5028 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
5029 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
5030 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
5031 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
5035 * journalctl --update-catalog now understands a new --root=
5036 option to operate on catalogs found in a different root
5039 * During shutdown after systemd has terminated all running
5040 services a final killing loop kills all remaining left-over
5041 processes. We will now print the name of these processes
5042 when we send SIGKILL to them, since this usually indicates a
5045 * If /etc/crypttab refers to password files stored on
5046 configured mount points automatic dependencies will now be
5047 generated to ensure the specific mount is established first
5048 before the key file is attempted to be read.
5050 * 'systemctl status' will now show information about the
5051 network sockets a socket unit is listening on.
5053 * 'systemctl status' will also shown information about any
5054 drop-in configuration file for units. (Drop-In configuration
5055 files in this context are files such as
5056 /etc/systemd/systemd/foobar.service.d/*.conf)
5058 * systemd-cgtop now optionally shows summed up CPU times of
5059 cgroups. Press '%' while running cgtop to switch between
5060 percentage and absolute mode. This is useful to determine
5061 which cgroups use up the most CPU time over the entire
5062 runtime of the system. systemd-cgtop has also been updated
5063 to be 'pipeable' for processing with further shell tools.
5065 * 'hostnamectl set-hostname' will now allow setting of FQDN
5068 * The formatting and parsing of time span values has been
5069 changed. The parser now understands fractional expressions
5070 such as "5.5h". The formatter will now output fractional
5071 expressions for all time spans under 1min, i.e. "5.123456s"
5072 rather than "5s 123ms 456us". For time spans under 1s
5073 millisecond values are shown, for those under 1ms
5074 microsecond values are shown. This should greatly improve
5075 all time-related output of systemd.
5077 * libsystemd-login and libsystemd-journal gained new
5078 functions for querying the poll() events mask and poll()
5079 timeout value for integration into arbitrary event
5082 * localectl gained the ability to list available X11 keymaps
5083 (models, layouts, variants, options).
5085 * 'systemd-analyze dot' gained the ability to filter for
5086 specific units via shell-style globs, to create smaller,
5087 more useful graphs. I.e. it is now possible to create simple
5088 graphs of all the dependencies between only target units, or
5089 of all units that Avahi has dependencies with.
5091 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
5092 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
5093 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
5094 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
5095 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
5096 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
5097 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
5101 * The boot-time readahead implementation for rotating media
5102 will now read the read-ahead data in multiple passes which
5103 consist of all read requests made in equidistant time
5104 intervals. This means instead of strictly reading read-ahead
5105 data in its physical order on disk we now try to find a
5106 middle ground between physical and access time order.
5108 * /etc/os-release files gained a new BUILD_ID= field for usage
5109 on operating systems that provide continuous builds of OS
5112 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
5113 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
5114 William Douglas, Zbigniew Jędrzejewski-Szmek
5118 * systemd-python gained an API exposing libsystemd-daemon.
5120 * The SMACK setup logic gained support for uploading CIPSO
5123 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
5124 ReadOnlyDirectories= and InaccessibleDirectories= has
5125 changed. The private /tmp and /var/tmp directories are now
5126 shared by all processes of a service (which means
5127 ExecStartPre= may now leave data in /tmp that ExecStart= of
5128 the same service can still access). When a service is
5129 stopped its temporary directories are immediately deleted
5130 (normal clean-up with tmpfiles is still done in addition to
5133 * By default, systemd will now set a couple of sysctl
5134 variables in the kernel: the safe sysrq options are turned
5135 on, IP route verification is turned on, and source routing
5136 disabled. The recently added hardlink and softlink
5137 protection of the kernel is turned on. These settings should
5138 be reasonably safe, and good defaults for all new systems.
5140 * The predictable network naming logic may now be turned off
5141 with a new kernel command line switch: net.ifnames=0.
5143 * A new libsystemd-bus module has been added that implements a
5144 pretty complete D-Bus client library. For details see:
5146 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
5148 * journald will now explicitly flush the journal files to disk
5149 at the latest 5min after each write. The file will then also
5150 be marked offline until the next write. This should increase
5151 reliability in case of a crash. The synchronization delay
5152 can be configured via SyncIntervalSec= in journald.conf.
5154 * There's a new remote-fs-setup.target unit that can be used
5155 to pull in specific services when at least one remote file
5156 system is to be mounted.
5158 * There are new targets timers.target and paths.target as
5159 canonical targets to pull user timer and path units in
5160 from. This complements sockets.target with a similar
5161 purpose for socket units.
5163 * libudev gained a new call udev_device_set_attribute_value()
5164 to set sysfs attributes of a device.
5166 * The udev daemon now sets the default number of worker
5167 processes executed in parallel based on the number of available
5168 CPUs instead of the amount of available RAM. This is supposed
5169 to provide a more reliable default and limit a too aggressive
5170 parallelism for setups with 1000s of devices connected.
5172 Contributions from: Auke Kok, Colin Walters, Cristian
5173 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
5174 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
5175 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
5176 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
5177 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
5178 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
5179 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
5180 Zbigniew Jędrzejewski-Szmek
5184 * Configuration of unit files may now be extended via drop-in
5185 files without having to edit/override the unit files
5186 themselves. More specifically, if the administrator wants to
5187 change one value for a service file foobar.service he can
5188 now do so by dropping in a configuration snippet into
5189 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
5190 will load all these snippets and apply them on top of the
5191 main unit configuration file, possibly extending or
5192 overriding its settings. Using these drop-in snippets is
5193 generally nicer than the two earlier options for changing
5194 unit files locally: copying the files from
5195 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
5196 them there; or creating a new file in /etc/systemd/system/
5197 that incorporates the original one via ".include". Drop-in
5198 snippets into these .d/ directories can be placed in any
5199 directory systemd looks for units in, and the usual
5200 overriding semantics between /usr/lib, /etc and /run apply
5203 * Most unit file settings which take lists of items can now be
5204 reset by assigning the empty string to them. For example,
5205 normally, settings such as Environment=FOO=BAR append a new
5206 environment variable assignment to the environment block,
5207 each time they are used. By assigning Environment= the empty
5208 string the environment block can be reset to empty. This is
5209 particularly useful with the .d/*.conf drop-in snippets
5210 mentioned above, since this adds the ability to reset list
5211 settings from vendor unit files via these drop-ins.
5213 * systemctl gained a new "list-dependencies" command for
5214 listing the dependencies of a unit recursively.
5216 * Inhibitors are now honored and listed by "systemctl
5217 suspend", "systemctl poweroff" (and similar) too, not only
5218 GNOME. These commands will also list active sessions by
5221 * Resource limits (as exposed by the various control group
5222 controllers) can now be controlled dynamically at runtime
5223 for all units. More specifically, you can now use a command
5224 like "systemctl set-cgroup-attr foobar.service cpu.shares
5225 2000" to alter the CPU shares a specific service gets. These
5226 settings are stored persistently on disk, and thus allow the
5227 administrator to easily adjust the resource usage of
5228 services with a few simple commands. This dynamic resource
5229 management logic is also available to other programs via the
5230 bus. Almost any kernel cgroup attribute and controller is
5233 * systemd-vconsole-setup will now copy all font settings to
5234 all allocated VTs, where it previously applied them only to
5237 * libsystemd-login gained the new sd_session_get_tty() API
5240 * This release drops support for a few legacy or
5241 distribution-specific LSB facility names when parsing init
5242 scripts: $x-display-manager, $mail-transfer-agent,
5243 $mail-transport-agent, $mail-transfer-agent, $smtp,
5244 $null. Also, the mail-transfer-agent.target unit backing
5245 this has been removed. Distributions which want to retain
5246 compatibility with this should carry the burden for
5247 supporting this themselves and patch support for these back
5248 in, if they really need to. Also, the facilities $syslog and
5249 $local_fs are now ignored, since systemd does not support
5250 early-boot LSB init scripts anymore, and these facilities
5251 are implied anyway for normal services. syslog.target has
5254 * There are new bus calls on PID1's Manager object for
5255 cancelling jobs, and removing snapshot units. Previously,
5256 both calls were only available on the Job and Snapshot
5259 * systemd-journal-gatewayd gained SSL support.
5261 * The various "environment" files, such as /etc/locale.conf
5262 now support continuation lines with a backslash ("\") as
5263 last character in the line, similarly in style (but different)
5264 to how this is supported in shells.
5266 * For normal user processes the _SYSTEMD_USER_UNIT= field is
5267 now implicitly appended to every log entry logged. systemctl
5268 has been updated to filter by this field when operating on a
5269 user systemd instance.
5271 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
5272 CAP_AUDIT_CONTROL capabilities to the capabilities set for
5273 the container. This makes it easier to boot unmodified
5274 Fedora systems in a container, which however still requires
5275 audit=0 to be passed on the kernel command line. Auditing in
5276 kernel and userspace is unfortunately still too broken in
5277 context of containers, hence we recommend compiling it out
5278 of the kernel or using audit=0. Hopefully this will be fixed
5279 one day for good in the kernel.
5281 * nspawn gained the new --bind= and --bind-ro= parameters to
5282 bind mount specific directories from the host into the
5285 * nspawn will now mount its own devpts file system instance
5286 into the container, in order not to leak pty devices from
5287 the host into the container.
5289 * systemd will now read the firmware boot time performance
5290 information from the EFI variables, if the used boot loader
5291 supports this, and takes it into account for boot performance
5292 analysis via "systemd-analyze". This is currently supported
5293 only in conjunction with Gummiboot, but could be supported
5294 by other boot loaders too. For details see:
5296 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
5298 * A new generator has been added that automatically mounts the
5299 EFI System Partition (ESP) to /boot, if that directory
5300 exists, is empty, and no other file system has been
5301 configured to be mounted there.
5303 * logind will now send out PrepareForSleep(false) out
5304 unconditionally, after coming back from suspend. This may be
5305 used by applications as asynchronous notification for
5306 system resume events.
5308 * "systemctl unlock-sessions" has been added, that allows
5309 unlocking the screens of all user sessions at once, similar
5310 to how "systemctl lock-sessions" already locked all users
5311 sessions. This is backed by a new D-Bus call UnlockSessions().
5313 * "loginctl seat-status" will now show the master device of a
5314 seat. (i.e. the device of a seat that needs to be around for
5315 the seat to be considered available, usually the graphics
5318 * tmpfiles gained a new "X" line type, that allows
5319 configuration of files and directories (with wildcards) that
5320 shall be excluded from automatic cleanup ("aging").
5322 * udev default rules set the device node permissions now only
5323 at "add" events, and do not change them any longer with a
5324 later "change" event.
5326 * The log messages for lid events and power/sleep keypresses
5327 now carry a message ID.
5329 * We now have a substantially larger unit test suite, but this
5330 continues to be work in progress.
5332 * udevadm hwdb gained a new --root= parameter to change the
5333 root directory to operate relative to.
5335 * logind will now issue a background sync() request to the kernel
5336 early at shutdown, so that dirty buffers are flushed to disk early
5337 instead of at the last moment, in order to optimize shutdown
5340 * A new bootctl tool has been added that is an interface for
5341 certain boot loader operations. This is currently a preview
5342 and is likely to be extended into a small mechanism daemon
5343 like timedated, localed, hostnamed, and can be used by
5344 graphical UIs to enumerate available boot options, and
5345 request boot into firmware operations.
5347 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
5348 the rest of the package. It also has been updated to work
5349 correctly in initrds.
5351 * Policykit previously has been runtime optional, and is now
5352 also compile time optional via a configure switch.
5354 * systemd-analyze has been reimplemented in C. Also "systemctl
5355 dot" has moved into systemd-analyze.
5357 * "systemctl status" with no further parameters will now print
5358 the status of all active or failed units.
5360 * Operations such as "systemctl start" can now be executed
5361 with a new mode "--irreversible" which may be used to queue
5362 operations that cannot accidentally be reversed by a later
5363 job queuing. This is by default used to make shutdown
5364 requests more robust.
5366 * The Python API of systemd now gained a new module for
5367 reading journal files.
5369 * A new tool kernel-install has been added that can install
5370 kernel images according to the Boot Loader Specification:
5372 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
5374 * Boot time console output has been improved to provide
5375 animated boot time output for hanging jobs.
5377 * A new tool systemd-activate has been added which can be used
5378 to test socket activation with, directly from the command
5379 line. This should make it much easier to test and debug
5380 socket activation in daemons.
5382 * journalctl gained a new "--reverse" (or -r) option to show
5383 journal output in reverse order (i.e. newest line first).
5385 * journalctl gained a new "--pager-end" (or -e) option to jump
5386 to immediately jump to the end of the journal in the
5387 pager. This is only supported in conjunction with "less".
5389 * journalctl gained a new "--user-unit=" option, that works
5390 similarly to "--unit=" but filters for user units rather than
5393 * A number of unit files to ease adoption of systemd in
5394 initrds has been added. This moves some minimal logic from
5395 the various initrd implementations into systemd proper.
5397 * The journal files are now owned by a new group
5398 "systemd-journal", which exists specifically to allow access
5399 to the journal, and nothing else. Previously, we used the
5400 "adm" group for that, which however possibly covers more
5401 than just journal/log file access. This new group is now
5402 already used by systemd-journal-gatewayd to ensure this
5403 daemon gets access to the journal files and as little else
5404 as possible. Note that "make install" will also set FS ACLs
5405 up for /var/log/journal to give "adm" and "wheel" read
5406 access to it, in addition to "systemd-journal" which owns
5407 the journal files. We recommend that packaging scripts also
5408 add read access to "adm" + "wheel" to /var/log/journal, and
5409 all existing/future journal files. To normal users and
5410 administrators little changes, however packagers need to
5411 ensure to create the "systemd-journal" system group at
5412 package installation time.
5414 * The systemd-journal-gatewayd now runs as unprivileged user
5415 systemd-journal-gateway:systemd-journal-gateway. Packaging
5416 scripts need to create these system user/group at
5419 * timedated now exposes a new boolean property CanNTP that
5420 indicates whether a local NTP service is available or not.
5422 * systemd-detect-virt will now also detect xen PVs
5424 * The pstore file system is now mounted by default, if it is
5427 * In addition to the SELinux and IMA policies we will now also
5428 load SMACK policies at early boot.
5430 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
5431 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
5432 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
5433 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
5434 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
5435 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
5436 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
5437 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
5438 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
5439 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
5440 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
5441 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
5442 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
5443 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
5447 * Timer units now support calendar time events in addition to
5448 monotonic time events. That means you can now trigger a unit
5449 based on a calendar time specification such as "Thu,Fri
5450 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
5451 or fifth day of any month of the year 2013, given that it is
5452 a thursday or friday. This brings timer event support
5453 considerably closer to cron's capabilities. For details on
5454 the supported calendar time specification language see
5457 * udev now supports a number of different naming policies for
5458 network interfaces for predictable names, and a combination
5459 of these policies is now the default. Please see this wiki
5460 document for details:
5462 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
5464 * Auke Kok's bootchart implementation has been added to the
5465 systemd tree. It is an optional component that can graph the
5466 boot in quite some detail. It is one of the best bootchart
5467 implementations around and minimal in its code and
5470 * nss-myhostname has been integrated into the systemd source
5471 tree. nss-myhostname guarantees that the local hostname
5472 always stays resolvable via NSS. It has been a weak
5473 requirement of systemd-hostnamed since a long time, and
5474 since its code is actually trivial we decided to just
5475 include it in systemd's source tree. It can be turned off
5476 with a configure switch.
5478 * The read-ahead logic is now capable of properly detecting
5479 whether a btrfs file system is on SSD or rotating media, in
5480 order to optimize the read-ahead scheme. Previously, it was
5481 only capable of detecting this on traditional file systems
5484 * In udev, additional device properties are now read from the
5485 IAB in addition to the OUI database. Also, Bluetooth company
5486 identities are attached to the devices as well.
5488 * In service files %U may be used as specifier that is
5489 replaced by the configured user name of the service.
5491 * nspawn may now be invoked without a controlling TTY. This
5492 makes it suitable for invocation as its own service. This
5493 may be used to set up a simple containerized server system
5494 using only core OS tools.
5496 * systemd and nspawn can now accept socket file descriptors
5497 when they are started for socket activation. This enables
5498 implementation of socket activated nspawn
5499 containers. i.e. think about autospawning an entire OS image
5500 when the first SSH or HTTP connection is received. We expect
5501 that similar functionality will also be added to libvirt-lxc
5504 * journalctl will now suppress ANSI color codes when
5505 presenting log data.
5507 * systemctl will no longer show control group information for
5508 a unit if the control group is empty anyway.
5510 * logind can now automatically suspend/hibernate/shutdown the
5513 * /etc/machine-info and hostnamed now also expose the chassis
5514 type of the system. This can be used to determine whether
5515 the local system is a laptop, desktop, handset or
5516 tablet. This information may either be configured by the
5517 user/vendor or is automatically determined from ACPI and DMI
5518 information if possible.
5520 * A number of PolicyKit actions are now bound together with
5521 "imply" rules. This should simplify creating UIs because
5522 many actions will now authenticate similar ones as well.
5524 * Unit files learnt a new condition ConditionACPower= which
5525 may be used to conditionalize a unit depending on whether an
5526 AC power source is connected or not, of whether the system
5527 is running on battery power.
5529 * systemctl gained a new "is-failed" verb that may be used in
5530 shell scripts and suchlike to check whether a specific unit
5531 is in the "failed" state.
5533 * The EnvironmentFile= setting in unit files now supports file
5534 globbing, and can hence be used to easily read a number of
5535 environment files at once.
5537 * systemd will no longer detect and recognize specific
5538 distributions. All distribution-specific #ifdeffery has been
5539 removed, systemd is now fully generic and
5540 distribution-agnostic. Effectively, not too much is lost as
5541 a lot of the code is still accessible via explicit configure
5542 switches. However, support for some distribution specific
5543 legacy configuration file formats has been dropped. We
5544 recommend distributions to simply adopt the configuration
5545 files everybody else uses now and convert the old
5546 configuration from packaging scripts. Most distributions
5547 already did that. If that's not possible or desirable,
5548 distributions are welcome to forward port the specific
5549 pieces of code locally from the git history.
5551 * When logging a message about a unit systemd will now always
5552 log the unit name in the message meta data.
5554 * localectl will now also discover system locale data that is
5555 not stored in locale archives, but directly unpacked.
5557 * logind will no longer unconditionally use framebuffer
5558 devices as seat masters, i.e. as devices that are required
5559 to be existing before a seat is considered preset. Instead,
5560 it will now look for all devices that are tagged as
5561 "seat-master" in udev. By default, framebuffer devices will
5562 be marked as such, but depending on local systems, other
5563 devices might be marked as well. This may be used to
5564 integrate graphics cards using closed source drivers (such
5565 as NVidia ones) more nicely into logind. Note however, that
5566 we recommend using the open source NVidia drivers instead,
5567 and no udev rules for the closed-source drivers will be
5568 shipped from us upstream.
5570 Contributions from: Adam Williamson, Alessandro Crismani, Auke
5571 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
5572 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
5573 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
5574 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
5575 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
5576 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
5577 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
5578 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
5579 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
5580 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
5581 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
5586 * udev gained support for loading additional device properties
5587 from an indexed database that is keyed by vendor/product IDs
5588 and similar device identifiers. For the beginning this
5589 "hwdb" is populated with data from the well-known PCI and
5590 USB database, but also includes PNP, ACPI and OID data. In
5591 the longer run this indexed database shall grow into
5592 becoming the one central database for non-essential
5593 userspace device metadata. Previously, data from the PCI/USB
5594 database was only attached to select devices, since the
5595 lookup was a relatively expensive operation due to O(n) time
5596 complexity (with n being the number of entries in the
5597 database). Since this is now O(1), we decided to add in this
5598 data for all devices where this is available, by
5599 default. Note that the indexed database needs to be rebuilt
5600 when new data files are installed. To achieve this you need
5601 to update your packaging scripts to invoke "udevadm hwdb
5602 --update" after installation of hwdb data files. For
5603 RPM-based distributions we introduced the new
5604 %udev_hwdb_update macro for this purpose.
5606 * The Journal gained support for the "Message Catalog", an
5607 indexed database to link up additional information with
5608 journal entries. For further details please check:
5610 http://www.freedesktop.org/wiki/Software/systemd/catalog
5612 The indexed message catalog database also needs to be
5613 rebuilt after installation of message catalog files. Use
5614 "journalctl --update-catalog" for this. For RPM-based
5615 distributions we introduced the %journal_catalog_update
5616 macro for this purpose.
5618 * The Python Journal bindings gained support for the standard
5619 Python logging framework.
5621 * The Journal API gained new functions for checking whether
5622 the underlying file system of a journal file is capable of
5623 properly reporting file change notifications, or whether
5624 applications that want to reflect journal changes "live"
5625 need to recheck journal files continuously in appropriate
5628 * It is now possible to set the "age" field for tmpfiles
5629 entries to 0, indicating that files matching this entry
5630 shall always be removed when the directories are cleaned up.
5632 * coredumpctl gained a new "gdb" verb which invokes gdb
5633 right-away on the selected coredump.
5635 * There's now support for "hybrid sleep" on kernels that
5636 support this, in addition to "suspend" and "hibernate". Use
5637 "systemctl hybrid-sleep" to make use of this.
5639 * logind's HandleSuspendKey= setting (and related settings)
5640 now gained support for a new "lock" setting to simply
5641 request the screen lock on all local sessions, instead of
5642 actually executing a suspend or hibernation.
5644 * systemd will now mount the EFI variables file system by
5647 * Socket units now gained support for configuration of the
5648 SMACK security label.
5650 * timedatectl will now output the time of the last and next
5651 daylight saving change.
5653 * We dropped support for various legacy and distro-specific
5654 concepts, such as insserv, early-boot SysV services
5655 (i.e. those for non-standard runlevels such as 'b' or 'S')
5656 or ArchLinux /etc/rc.conf support. We recommend the
5657 distributions who still need support this to either continue
5658 to maintain the necessary patches downstream, or find a
5659 different solution. (Talk to us if you have questions!)
5661 * Various systemd components will now bypass PolicyKit checks
5662 for root and otherwise handle properly if PolicyKit is not
5663 found to be around. This should fix most issues for
5664 PolicyKit-less systems. Quite frankly this should have been
5665 this way since day one. It is absolutely our intention to
5666 make systemd work fine on PolicyKit-less systems, and we
5667 consider it a bug if something does not work as it should if
5668 PolicyKit is not around.
5670 * For embedded systems it is now possible to build udev and
5671 systemd without blkid and/or kmod support.
5673 * "systemctl switch-root" is now capable of switching root
5674 more than once. I.e. in addition to transitions from the
5675 initrd to the host OS it is now possible to transition to
5676 further OS images from the host. This is useful to implement
5677 offline updating tools.
5679 * Various other additions have been made to the RPM macros
5680 shipped with systemd. Use %udev_rules_update() after
5681 installing new udev rules files. %_udevhwdbdir,
5682 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
5683 %_sysctldir are now available which resolve to the right
5684 directories for packages to place various data files in.
5686 * journalctl gained the new --full switch (in addition to
5687 --all, to disable ellipsation for long messages.
5689 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
5690 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
5691 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
5692 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
5693 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
5694 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
5695 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
5696 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
5697 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
5701 * journalctl gained new --since= and --until= switches to
5702 filter by time. It also now supports nice filtering for
5703 units via --unit=/-u.
5705 * Type=oneshot services may use ExecReload= and do the
5708 * The journal daemon now supports time-based rotation and
5709 vacuuming, in addition to the usual disk-space based
5712 * The journal will now index the available field values for
5713 each field name. This enables clients to show pretty drop
5714 downs of available match values when filtering. The bash
5715 completion of journalctl has been updated
5716 accordingly. journalctl gained a new switch -F to list all
5717 values a certain field takes in the journal database.
5719 * More service events are now written as structured messages
5720 to the journal, and made recognizable via message IDs.
5722 * The timedated, localed and hostnamed mini-services which
5723 previously only provided support for changing time, locale
5724 and hostname settings from graphical DEs such as GNOME now
5725 also have a minimal (but very useful) text-based client
5726 utility each. This is probably the nicest way to changing
5727 these settings from the command line now, especially since
5728 it lists available options and is fully integrated with bash
5731 * There's now a new tool "systemd-coredumpctl" to list and
5732 extract coredumps from the journal.
5734 * We now install a README each in /var/log/ and
5735 /etc/rc.d/init.d explaining where the system logs and init
5736 scripts went. This hopefully should help folks who go to
5737 that dirs and look into the otherwise now empty void and
5738 scratch their heads.
5740 * When user-services are invoked (by systemd --user) the
5741 $MANAGERPID env var is set to the PID of systemd.
5743 * SIGRTMIN+24 when sent to a --user instance will now result
5744 in immediate termination of systemd.
5746 * gatewayd received numerous feature additions such as a
5747 "follow" mode, for live syncing and filtering.
5749 * browse.html now allows filtering and showing detailed
5750 information on specific entries. Keyboard navigation and
5751 mouse screen support has been added.
5753 * gatewayd/journalctl now supports HTML5/JSON
5754 Server-Sent-Events as output.
5756 * The SysV init script compatibility logic will now
5757 heuristically determine whether a script supports the
5758 "reload" verb, and only then make this available as
5761 * "systemctl status --follow" has been removed, use "journalctl
5764 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
5765 have been removed since they are hardly useful to be
5768 * And I'd like to take the opportunity to specifically mention
5769 Zbigniew for his great contributions. Zbigniew, you rock!
5771 Contributions from: Andrew Eikum, Christian Hesse, Colin
5772 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
5773 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
5774 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
5775 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
5776 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
5777 Jędrzejewski-Szmek, Сковорода Никита Андреевич
5781 * If /etc/vconsole.conf is non-existent or empty we will no
5782 longer load any console font or key map at boot by
5783 default. Instead the kernel defaults will be left
5784 intact. This is definitely the right thing to do, as no
5785 configuration should mean no configuration, and hard-coding
5786 font names that are different on all archs is probably a bad
5787 idea. Also, the kernel default key map and font should be
5788 good enough for most cases anyway, and mostly identical to
5789 the userspace fonts/key maps we previously overloaded them
5790 with. If distributions want to continue to default to a
5791 non-kernel font or key map they should ship a default
5792 /etc/vconsole.conf with the appropriate contents.
5794 Contributions from: Colin Walters, Daniel J Walsh, Dave
5795 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
5796 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
5800 * journalctl gained a new --cursor= switch to show entries
5801 starting from the specified location in the journal.
5803 * We now enforce a size limit on journal entry fields exported
5804 with "-o json" in journalctl. Fields larger than 4K will be
5805 assigned null. This can be turned off with --all.
5807 * An (optional) journal gateway daemon is now available as
5808 "systemd-journal-gatewayd.service". This service provides
5809 access to the journal via HTTP and JSON. This functionality
5810 will be used to implement live log synchronization in both
5811 pull and push modes, but has various other users too, such
5812 as easy log access for debugging of embedded devices. Right
5813 now it is already useful to retrieve the journal via HTTP:
5815 # systemctl start systemd-journal-gatewayd.service
5816 # wget http://localhost:19531/entries
5818 This will download the journal contents in a
5819 /var/log/messages compatible format. The same as JSON:
5821 # curl -H"Accept: application/json" http://localhost:19531/entries
5823 This service is also accessible via a web browser where a
5824 single static HTML5 app is served that uses the JSON logic
5825 to enable the user to do some basic browsing of the
5826 journal. This will be extended later on. Here's an example
5827 screenshot of this app in its current state:
5829 http://0pointer.de/public/journal-gatewayd
5831 Contributions from: Kay Sievers, Lennart Poettering, Robert
5832 Milasan, Tom Gundersen
5836 * The bash completion logic is now available for journalctl
5839 * We do not mount the "cpuset" controller anymore together with
5840 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
5841 started if no parameters are assigned to it. "cpuset" hence
5842 broke code that assumed it could create "cpu" groups and
5845 * journalctl -f will now subscribe to terminal size changes,
5846 and line break accordingly.
5848 Contributions from: Dave Reisner, Kay Sievers, Lennart
5849 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
5853 * nspawn will now create a symlink /etc/localtime in the
5854 container environment, copying the host's timezone
5855 setting. Previously this has been done via a bind mount, but
5856 since symlinks cannot be bind mounted this has now been
5857 changed to create/update the appropriate symlink.
5859 * journalctl -n's line number argument is now optional, and
5860 will default to 10 if omitted.
5862 * journald will now log the maximum size the journal files may
5863 take up on disk. This is particularly useful if the default
5864 built-in logic of determining this parameter from the file
5865 system size is used. Use "systemctl status
5866 systemd-journald.service" to see this information.
5868 * The multi-seat X wrapper tool has been stripped down. As X
5869 is now capable of enumerating graphics devices via udev in a
5870 seat-aware way the wrapper is not strictly necessary
5871 anymore. A stripped down temporary stop-gap is still shipped
5872 until the upstream display managers have been updated to
5873 fully support the new X logic. Expect this wrapper to be
5874 removed entirely in one of the next releases.
5876 * HandleSleepKey= in logind.conf has been split up into
5877 HandleSuspendKey= and HandleHibernateKey=. The old setting
5878 is not available anymore. X11 and the kernel are
5879 distinguishing between these keys and we should too. This
5880 also means the inhibition lock for these keys has been split
5883 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
5884 Poettering, Lukas Nykryn, Václav Pavlín
5888 * Whenever a unit changes state we will now log this to the
5889 journal and show along the unit's own log output in
5892 * ConditionPathIsMountPoint= can now properly detect bind
5893 mount points too. (Previously, a bind mount of one file
5894 system to another place in the same file system could not be
5895 detected as mount, since they shared struct stat's st_dev
5898 * We will now mount the cgroup controllers cpu, cpuacct,
5899 cpuset and the controllers net_cls, net_prio together by
5902 * nspawn containers will now have a virtualized boot
5903 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
5904 over with a randomized ID at container initialization). This
5905 has the effect of making "journalctl -b" do the right thing
5908 * The JSON output journal serialization has been updated not
5909 to generate "endless" list objects anymore, but rather one
5910 JSON object per line. This is more in line how most JSON
5911 parsers expect JSON objects. The new output mode
5912 "json-pretty" has been added to provide similar output, but
5913 neatly aligned for readability by humans.
5915 * We dropped all explicit sync() invocations in the shutdown
5916 code. The kernel does this implicitly anyway in the kernel
5917 reboot() syscall. halt(8)'s -n option is now a compatibility
5920 * We now support virtualized reboot() in containers, as
5921 supported by newer kernels. We will fall back to exit() if
5922 CAP_SYS_REBOOT is not available to the container. Also,
5923 nspawn makes use of this now and will actually reboot the
5924 container if the containerized OS asks for that.
5926 * journalctl will only show local log output by default
5927 now. Use --merge (-m) to show remote log output, too.
5929 * libsystemd-journal gained the new sd_journal_get_usage()
5930 call to determine the current disk usage of all journal
5931 files. This is exposed in the new "journalctl --disk-usage"
5934 * journald gained a new configuration setting SplitMode= in
5935 journald.conf which may be used to control how user journals
5936 are split off. See journald.conf(5) for details.
5938 * A new condition type ConditionFileNotEmpty= has been added.
5940 * tmpfiles' "w" lines now support file globbing, to write
5941 multiple files at once.
5943 * We added Python bindings for the journal submission
5944 APIs. More Python APIs for a number of selected APIs will
5945 likely follow. Note that we intend to add native bindings
5946 only for the Python language, as we consider it common
5947 enough to deserve bindings shipped within systemd. There are
5948 various projects outside of systemd that provide bindings
5949 for languages such as PHP or Lua.
5951 * Many conditions will now resolve specifiers such as %i. In
5952 addition, PathChanged= and related directives of .path units
5953 now support specifiers as well.
5955 * There's now a new RPM macro definition for the system preset
5958 * journald will now warn if it ca not forward a message to the
5959 syslog daemon because its socket is full.
5961 * timedated will no longer write or process /etc/timezone,
5962 except on Debian. As we do not support late mounted /usr
5963 anymore /etc/localtime always being a symlink is now safe,
5964 and hence the information in /etc/timezone is not necessary
5967 * logind will now always reserve one VT for a text getty (VT6
5968 by default). Previously if more than 6 X sessions where
5969 started they took up all the VTs with auto-spawned gettys,
5970 so that no text gettys were available anymore.
5972 * udev will now automatically inform the btrfs kernel logic
5973 about btrfs RAID components showing up. This should make
5974 simple hotplug based btrfs RAID assembly work.
5976 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
5977 (but not for its children which will stay at the kernel
5978 default). This should allow setups with a lot more listening
5981 * systemd will now always pass the configured timezone to the
5982 kernel at boot. timedated will do the same when the timezone
5985 * logind's inhibition logic has been updated. By default,
5986 logind will now handle the lid switch, the power and sleep
5987 keys all the time, even in graphical sessions. If DEs want
5988 to handle these events on their own they should take the new
5989 handle-power-key, handle-sleep-key and handle-lid-switch
5990 inhibitors during their runtime. A simple way to achieve
5991 that is to invoke the DE wrapped in an invocation of:
5993 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
5995 * Access to unit operations is now checked via SELinux taking
5996 the unit file label and client process label into account.
5998 * systemd will now notify the administrator in the journal
5999 when he over-mounts a non-empty directory.
6001 * There are new specifiers that are resolved in unit files,
6002 for the host name (%H), the machine ID (%m) and the boot ID
6005 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
6006 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
6007 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
6008 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
6009 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
6010 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
6011 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
6015 * Support for reading structured kernel messages from
6016 /dev/kmsg has now been added and is enabled by default.
6018 * Support for reading kernel messages from /proc/kmsg has now
6019 been removed. If you want kernel messages in the journal
6020 make sure to run a recent kernel (>= 3.5) that supports
6021 reading structured messages from /dev/kmsg (see
6022 above). /proc/kmsg is now exclusive property of classic
6023 syslog daemons again.
6025 * The libudev API gained the new
6026 udev_device_new_from_device_id() call.
6028 * The logic for file system namespace (ReadOnlyDirectory=,
6029 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
6030 require pivot_root() anymore. This means fewer temporary
6031 directories are created below /tmp for this feature.
6033 * nspawn containers will now see and receive all submounts
6034 made on the host OS below the root file system of the
6037 * Forward Secure Sealing is now supported for Journal files,
6038 which provide cryptographical sealing of journal files so
6039 that attackers cannot alter log history anymore without this
6040 being detectable. Lennart will soon post a blog story about
6041 this explaining it in more detail.
6043 * There are two new service settings RestartPreventExitStatus=
6044 and SuccessExitStatus= which allow configuration of exit
6045 status (exit code or signal) which will be excepted from the
6046 restart logic, resp. consider successful.
6048 * journalctl gained the new --verify switch that can be used
6049 to check the integrity of the structure of journal files and
6050 (if Forward Secure Sealing is enabled) the contents of
6053 * nspawn containers will now be run with /dev/stdin, /dev/fd/
6054 and similar symlinks pre-created. This makes running shells
6055 as container init process a lot more fun.
6057 * The fstab support can now handle PARTUUID= and PARTLABEL=
6060 * A new ConditionHost= condition has been added to match
6061 against the hostname (with globs) and machine ID. This is
6062 useful for clusters where a single OS image is used to
6063 provision a large number of hosts which shall run slightly
6064 different sets of services.
6066 * Services which hit the restart limit will now be placed in a
6069 Contributions from: Bertram Poettering, Dave Reisner, Huang
6070 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
6071 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
6075 * When running in --user mode systemd will now become a
6076 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
6077 tree a lot more organized.
6079 * A new PartOf= unit dependency type has been introduced that
6080 may be used to group services in a natural way.
6082 * "systemctl enable" may now be used to enable instances of
6085 * journalctl now prints error log levels in red, and
6086 warning/notice log levels in bright white. It also supports
6087 filtering by log level now.
6089 * cgtop gained a new -n switch (similar to top), to configure
6090 the maximum number of iterations to run for. It also gained
6091 -b, to run in batch mode (accepting no input).
6093 * The suffix ".service" may now be omitted on most systemctl
6094 command lines involving service unit names.
6096 * There's a new bus call in logind to lock all sessions, as
6097 well as a loginctl verb for it "lock-sessions".
6099 * libsystemd-logind.so gained a new call sd_journal_perror()
6100 that works similar to libc perror() but logs to the journal
6101 and encodes structured information about the error number.
6103 * /etc/crypttab entries now understand the new keyfile-size=
6106 * shutdown(8) now can send a (configurable) wall message when
6107 a shutdown is cancelled.
6109 * The mount propagation mode for the root file system will now
6110 default to "shared", which is useful to make containers work
6111 nicely out-of-the-box so that they receive new mounts from
6112 the host. This can be undone locally by running "mount
6113 --make-rprivate /" if needed.
6115 * The prefdm.service file has been removed. Distributions
6116 should maintain this unit downstream if they intend to keep
6117 it around. However, we recommend writing normal unit files
6118 for display managers instead.
6120 * Since systemd is a crucial part of the OS we will now
6121 default to a number of compiler switches that improve
6122 security (hardening) such as read-only relocations, stack
6123 protection, and suchlike.
6125 * The TimeoutSec= setting for services is now split into
6126 TimeoutStartSec= and TimeoutStopSec= to allow configuration
6127 of individual time outs for the start and the stop phase of
6130 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
6131 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
6132 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
6133 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
6134 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
6135 Gundersen, Zbigniew Jędrzejewski-Szmek
6139 * The journal and id128 C APIs are now fully documented as man
6142 * Extra safety checks have been added when transitioning from
6143 the initial RAM disk to the main system to avoid accidental
6146 * /etc/crypttab entries now understand the new keyfile-offset=
6149 * systemctl -t can now be used to filter by unit load state.
6151 * The journal C API gained the new sd_journal_wait() call to
6152 make writing synchronous journal clients easier.
6154 * journalctl gained the new -D switch to show journals from a
6157 * journalctl now displays a special marker between log
6158 messages of two different boots.
6160 * The journal is now explicitly flushed to /var via a service
6161 systemd-journal-flush.service, rather than implicitly simply
6162 by seeing /var/log/journal to be writable.
6164 * journalctl (and the journal C APIs) can now match for much
6165 more complex expressions, with alternatives and
6168 * When transitioning from the initial RAM disk to the main
6169 system we will now kill all processes in a killing spree to
6170 ensure no processes stay around by accident.
6172 * Three new specifiers may be used in unit files: %u, %h, %s
6173 resolve to the user name, user home directory resp. user
6174 shell. This is useful for running systemd user instances.
6176 * We now automatically rotate journal files if their data
6177 object hash table gets a fill level > 75%. We also size the
6178 hash table based on the configured maximum file size. This
6179 together should lower hash collisions drastically and thus
6180 speed things up a bit.
6182 * journalctl gained the new "--header" switch to introspect
6183 header data of journal files.
6185 * A new setting SystemCallFilters= has been added to services
6186 which may be used to apply blacklists or whitelists to
6187 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
6189 * nspawn gained a new --link-journal= switch (and quicker: -j)
6190 to link the container journal with the host. This makes it
6191 very easy to centralize log viewing on the host for all
6192 guests while still keeping the journal files separated.
6194 * Many bugfixes and optimizations
6196 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
6197 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
6198 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
6203 * Several tools now understand kernel command line arguments,
6204 which are only read when run in an initial RAM disk. They
6205 usually follow closely their normal counterparts, but are
6208 * There's a new tool to analyze the readahead files that are
6209 automatically generated at boot. Use:
6211 /usr/lib/systemd/systemd-readahead analyze /.readahead
6213 * We now provide an early debug shell on tty9 if this enabled. Use:
6215 systemctl enable debug-shell.service
6217 * All plymouth related units have been moved into the Plymouth
6218 package. Please make sure to upgrade your Plymouth version
6221 * systemd-tmpfiles now supports getting passed the basename of
6222 a configuration file only, in which case it will look for it
6223 in all appropriate directories automatically.
6225 * udevadm info now takes a /dev or /sys path as argument, and
6226 does the right thing. Example:
6228 udevadm info /dev/sda
6229 udevadm info /sys/class/block/sda
6231 * systemctl now prints a warning if a unit is stopped but a
6232 unit that might trigger it continues to run. Example: a
6233 service is stopped but the socket that activates it is left
6236 * "systemctl status" will now mention if the log output was
6237 shortened due to rotation since a service has been started.
6239 * The journal API now exposes functions to determine the
6240 "cutoff" times due to rotation.
6242 * journald now understands SIGUSR1 and SIGUSR2 for triggering
6243 immediately flushing of runtime logs to /var if possible,
6244 resp. for triggering immediate rotation of the journal
6247 * It is now considered an error if a service is attempted to
6248 be stopped that is not loaded.
6250 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
6252 * systemd-analyze now supports Python 3
6254 * tmpfiles now supports cleaning up directories via aging
6255 where the first level dirs are always kept around but
6256 directories beneath it automatically aged. This is enabled
6257 by prefixing the age field with '~'.
6259 * Seat objects now expose CanGraphical, CanTTY properties
6260 which is required to deal with very fast bootups where the
6261 display manager might be running before the graphics drivers
6262 completed initialization.
6264 * Seat objects now expose a State property.
6266 * We now include RPM macros for service enabling/disabling
6267 based on the preset logic. We recommend RPM based
6268 distributions to make use of these macros if possible. This
6269 makes it simpler to reuse RPM spec files across
6272 * We now make sure that the collected systemd unit name is
6273 always valid when services log to the journal via
6276 * There's a new man page kernel-command-line(7) detailing all
6277 command line options we understand.
6279 * The fstab generator may now be disabled at boot by passing
6280 fstab=0 on the kernel command line.
6282 * A new kernel command line option modules-load= is now understood
6283 to load a specific kernel module statically, early at boot.
6285 * Unit names specified on the systemctl command line are now
6286 automatically escaped as needed. Also, if file system or
6287 device paths are specified they are automatically turned
6288 into the appropriate mount or device unit names. Example:
6290 systemctl status /home
6291 systemctl status /dev/sda
6293 * The SysVConsole= configuration option has been removed from
6294 system.conf parsing.
6296 * The SysV search path is no longer exported on the D-Bus
6299 * The Names= option has been removed from unit file parsing.
6301 * There's a new man page bootup(7) detailing the boot process.
6303 * Every unit and every generator we ship with systemd now
6304 comes with full documentation. The self-explanatory boot is
6307 * A couple of services gained "systemd-" prefixes in their
6308 name if they wrap systemd code, rather than only external
6309 code. Among them fsck@.service which is now
6310 systemd-fsck@.service.
6312 * The HaveWatchdog property has been removed from the D-Bus
6315 * systemd.confirm_spawn= on the kernel command line should now
6318 * There's a new man page crypttab(5) which details all options
6319 we actually understand.
6321 * systemd-nspawn gained a new --capability= switch to pass
6322 additional capabilities to the container.
6324 * timedated will now read known NTP implementation unit names
6325 from /usr/lib/systemd/ntp-units.d/*.list,
6326 systemd-timedated-ntp.target has been removed.
6328 * journalctl gained a new switch "-b" that lists log data of
6329 the current boot only.
6331 * The notify socket is in the abstract namespace again, in
6332 order to support daemons which chroot() at start-up.
6334 * There is a new Storage= configuration option for journald
6335 which allows configuration of where log data should go. This
6336 also provides a way to disable journal logging entirely, so
6337 that data collected is only forwarded to the console, the
6338 kernel log buffer or another syslog implementation.
6340 * Many bugfixes and optimizations
6342 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
6343 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
6344 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
6345 Shawn Landden, Tom Gundersen
6349 * "systemctl help <unit>" now shows the man page if one is
6352 * Several new man pages have been added.
6354 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
6355 MaxLevelConsole= can now be specified in
6356 journald.conf. These options allow reducing the amount of
6357 data stored on disk or forwarded by the log level.
6359 * TimerSlackNSec= can now be specified in system.conf for
6360 PID1. This allows system-wide power savings.
6362 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
6363 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
6368 * logind is now capable of (optionally) handling power and
6369 sleep keys as well as the lid switch.
6371 * journalctl now understands the syntax "journalctl
6372 /usr/bin/avahi-daemon" to get all log output of a specific
6375 * CapabilityBoundingSet= in system.conf now also influences
6376 the capability bound set of usermode helpers of the kernel.
6378 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
6379 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
6380 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
6381 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
6385 * Note that we skipped 139 releases here in order to set the
6386 new version to something that is greater than both udev's
6387 and systemd's most recent version number.
6389 * udev: all udev sources are merged into the systemd source tree now.
6390 All future udev development will happen in the systemd tree. It
6391 is still fully supported to use the udev daemon and tools without
6392 systemd running, like in initramfs or other init systems. Building
6393 udev though, will require the *build* of the systemd tree, but
6394 udev can be properly *run* without systemd.
6396 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
6397 should be used to create dead device nodes as workarounds for broken
6400 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
6401 no longer supported. udev_monitor_new_from_netlink() needs to be
6402 used to subscribe to events.
6404 * udev: when udevd is started by systemd, processes which are left
6405 behind by forking them off of udev rules, are unconditionally cleaned
6406 up and killed now after the event handling has finished. Services or
6407 daemons must be started as systemd services. Services can be
6408 pulled-in by udev to get started, but they can no longer be directly
6409 forked by udev rules.
6411 * udev: the daemon binary is called systemd-udevd now and installed
6412 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
6413 to adapt to that, create symlink, or rename the binary after building
6416 * libudev no longer provides these symbols:
6417 udev_monitor_from_socket()
6418 udev_queue_get_failed_list_entry()
6419 udev_get_{dev,sys,run}_path()
6420 The versions number was bumped and symbol versioning introduced.
6422 * systemd-loginctl and systemd-journalctl have been renamed
6423 to loginctl and journalctl to match systemctl.
6425 * The config files: /etc/systemd/systemd-logind.conf and
6426 /etc/systemd/systemd-journald.conf have been renamed to
6427 logind.conf and journald.conf. Package updates should rename
6428 the files to the new names on upgrade.
6430 * For almost all files the license is now LGPL2.1+, changed
6431 from the previous GPL2.0+. Exceptions are some minor stuff
6432 of udev (which will be changed to LGPL2.1 eventually, too),
6433 and the MIT licensed sd-daemon.[ch] library that is suitable
6434 to be used as drop-in files.
6436 * systemd and logind now handle system sleep states, in
6437 particular suspending and hibernating.
6439 * logind now implements a sleep/shutdown/idle inhibiting logic
6440 suitable for a variety of uses. Soonishly Lennart will blog
6441 about this in more detail.
6443 * var-run.mount and var-lock.mount are no longer provided
6444 (which previously bind mounted these directories to their new
6445 places). Distributions which have not converted these
6446 directories to symlinks should consider stealing these files
6447 from git history and add them downstream.
6449 * We introduced the Documentation= field for units and added
6450 this to all our shipped units. This is useful to make it
6451 easier to explore the boot and the purpose of the various
6454 * All smaller setup units (such as
6455 systemd-vconsole-setup.service) now detect properly if they
6456 are run in a container and are skipped when
6457 appropriate. This guarantees an entirely noise-free boot in
6458 Linux container environments such as systemd-nspawn.
6460 * A framework for implementing offline system updates is now
6461 integrated, for details see:
6462 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
6464 * A new service type Type=idle is available now which helps us
6465 avoiding ugly interleaving of getty output and boot status
6468 * There's now a system-wide CapabilityBoundingSet= option to
6469 globally reduce the set of capabilities for the
6470 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
6471 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
6472 even CAP_NET_ADMIN system-wide for secure systems.
6474 * There are now system-wide DefaultLimitXXX= options to
6475 globally change the defaults of the various resource limits
6476 for all units started by PID 1.
6478 * Harald Hoyer's systemd test suite has been integrated into
6479 systemd which allows easy testing of systemd builds in qemu
6480 and nspawn. (This is really awesome! Ask us for details!)
6482 * The fstab parser is now implemented as generator, not inside
6485 * systemctl will now warn you if .mount units generated from
6486 /etc/fstab are out of date due to changes in fstab that
6487 have not been read by systemd yet.
6489 * systemd is now suitable for usage in initrds. Dracut has
6490 already been updated to make use of this. With this in place
6491 initrds get a slight bit faster but primarily are much
6492 easier to introspect and debug since "systemctl status" in
6493 the host system can be used to introspect initrd services,
6494 and the journal from the initrd is kept around too.
6496 * systemd-delta has been added, a tool to explore differences
6497 between user/admin configuration and vendor defaults.
6499 * PrivateTmp= now affects both /tmp and /var/tmp.
6501 * Boot time status messages are now much prettier and feature
6502 proper english language. Booting up systemd has never been
6505 * Read-ahead pack files now include the inode number of all
6506 files to pre-cache. When the inode changes the pre-caching
6507 is not attempted. This should be nicer to deal with updated
6508 packages which might result in changes of read-ahead
6511 * We now temporaritly lower the kernel's read_ahead_kb variable
6512 when collecting read-ahead data to ensure the kernel's
6513 built-in read-ahead does not add noise to our measurements
6514 of necessary blocks to pre-cache.
6516 * There's now RequiresMountsFor= to add automatic dependencies
6517 for all mounts necessary for a specific file system path.
6519 * MountAuto= and SwapAuto= have been removed from
6520 system.conf. Mounting file systems at boot has to take place
6523 * nspawn now learned a new switch --uuid= to set the machine
6524 ID on the command line.
6526 * nspawn now learned the -b switch to automatically search
6529 * vt102 is now the default TERM for serial TTYs, upgraded from
6532 * systemd-logind now works on VT-less systems.
6534 * The build tree has been reorganized. The individual
6535 components now have directories of their own.
6537 * A new condition type ConditionPathIsReadWrite= is now available.
6539 * nspawn learned the new -C switch to create cgroups for the
6540 container in other hierarchies.
6542 * We now have support for hardware watchdogs, configurable in
6545 * The scheduled shutdown logic now has a public API.
6547 * We now mount /tmp as tmpfs by default, but this can be
6548 masked and /etc/fstab can override it.
6550 * Since udisks does not make use of /media anymore we are not
6551 mounting a tmpfs on it anymore.
6553 * journalctl gained a new --local switch to only interleave
6554 locally generated journal files.
6556 * We can now load the IMA policy at boot automatically.
6558 * The GTK tools have been split off into a systemd-ui.
6560 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
6561 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
6562 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
6563 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
6564 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
6565 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
6566 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
6567 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
6568 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
6573 * This is mostly a bugfix release
6575 * Support optional initialization of the machine ID from the
6576 KVM or container configured UUID.
6578 * Support immediate reboots with "systemctl reboot -ff"
6580 * Show /etc/os-release data in systemd-analyze output
6582 * Many bugfixes for the journal, including endianness fixes and
6583 ensuring that disk space enforcement works
6585 * sd-login.h is C++ compatible again
6587 * Extend the /etc/os-release format on request of the Debian
6590 * We now refuse non-UTF8 strings used in various configuration
6591 and unit files. This is done to ensure we do not pass invalid
6592 data over D-Bus or expose it elsewhere.
6594 * Register Mimo USB Screens as suitable for automatic seat
6597 * Read SELinux client context from journal clients in a race
6600 * Reorder configuration file lookup order. /etc now always
6601 overrides /run in order to allow the administrator to always
6602 and unconditionally override vendor-supplied or
6603 automatically generated data.
6605 * The various user visible bits of the journal now have man
6606 pages. We still lack man pages for the journal API calls
6609 * We now ship all man pages in HTML format again in the
6612 Contributions from: Dave Reisner, Dirk Eibach, Frederic
6613 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
6614 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
6619 * This is mostly a bugfix release
6621 * systems lacking /etc/os-release are no longer supported.
6623 * Various functionality updates to libsystemd-login.so
6625 * Track class of PAM logins to distinguish greeters from
6628 Contributions from: Kay Sievers, Lennart Poettering, Michael
6633 * This is an important bugfix release for v41.
6635 * Building man pages is now optional which should be useful
6636 for those building systemd from git but unwilling to install
6639 * Watchdog support for supervising services is now usable. In
6640 a future release support for hardware watchdogs
6641 (i.e. /dev/watchdog) will be added building on this.
6643 * Service start rate limiting is now configurable and can be
6644 turned off per service. When a start rate limit is hit a
6645 reboot can automatically be triggered.
6647 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
6649 Contributions from: Benjamin Franzke, Bill Nottingham,
6650 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
6651 Schmidt, Michał Górny, Piotr Drąg
6655 * The systemd binary is installed /usr/lib/systemd/systemd now;
6656 An existing /sbin/init symlink needs to be adapted with the
6659 * The code that loads kernel modules has been ported to invoke
6660 libkmod directly, instead of modprobe. This means we do not
6661 support systems with module-init-tools anymore.
6663 * Watchdog support is now already useful, but still not
6666 * A new kernel command line option systemd.setenv= is
6667 understood to set system wide environment variables
6668 dynamically at boot.
6670 * We now limit the set of capabilities of systemd-journald.
6672 * We now set SIGPIPE to ignore by default, since it only is
6673 useful in shell pipelines, and has little use in general
6674 code. This can be disabled with IgnoreSIPIPE=no in unit
6677 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
6678 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
6683 * This is mostly a bugfix release
6685 * We now expose the reason why a service failed in the
6686 "Result" D-Bus property.
6688 * Rudimentary service watchdog support (will be completed over
6689 the next few releases.)
6691 * When systemd forks off in order execute some service we will
6692 now immediately changes its argv[0] to reflect which process
6693 it will execute. This is useful to minimize the time window
6694 with a generic argv[0], which makes bootcharts more useful
6696 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
6697 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
6698 Mike Kazantsev, Ray Strode
6702 * This is mostly a test release, but incorporates many
6705 * New systemd-cgtop tool to show control groups by their
6708 * Linking against libacl for ACLs is optional again. If
6709 disabled, support tracking device access for active logins
6710 goes becomes unavailable, and so does access to the user
6711 journals by the respective users.
6713 * If a group "adm" exists, journal files are automatically
6714 owned by them, thus allow members of this group full access
6715 to the system journal as well as all user journals.
6717 * The journal now stores the SELinux context of the logging
6718 client for all entries.
6720 * Add C++ inclusion guards to all public headers
6722 * New output mode "cat" in the journal to print only text
6723 messages, without any meta data like date or time.
6725 * Include tiny X server wrapper as a temporary stop-gap to
6726 teach XOrg udev display enumeration. This is used by display
6727 managers such as gdm, and will go away as soon as XOrg
6728 learned native udev hotplugging for display devices.
6730 * Add new systemd-cat tool for executing arbitrary programs
6731 with STDERR/STDOUT connected to the journal. Can also act as
6732 BSD logger replacement, and does so by default.
6734 * Optionally store all locally generated coredumps in the
6735 journal along with meta data.
6737 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
6738 writing short strings to files (for usage for /sys), and for
6739 creating symlinks, character and block device nodes.
6741 * New unit file option ControlGroupPersistent= to make cgroups
6742 persistent, following the mechanisms outlined in
6743 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
6745 * Support multiple local RTCs in a sane way
6747 * No longer monopolize IO when replaying readahead data on
6748 rotating disks, since we might starve non-file-system IO to
6749 death, since fanotify() will not see accesses done by blkid,
6752 * Do not show kernel threads in systemd-cgls anymore, unless
6753 requested with new -k switch.
6755 Contributions from: Dan Horák, Kay Sievers, Lennart
6756 Poettering, Michal Schmidt
6760 * This is mostly a test release, but incorporates many
6763 * The git repository moved to:
6764 git://anongit.freedesktop.org/systemd/systemd
6765 ssh://git.freedesktop.org/git/systemd/systemd
6767 * First release with the journal
6768 http://0pointer.de/blog/projects/the-journal.html
6770 * The journal replaces both systemd-kmsg-syslogd and
6771 systemd-stdout-bridge.
6773 * New sd_pid_get_unit() API call in libsystemd-logind
6775 * Many systemadm clean-ups
6777 * Introduce remote-fs-pre.target which is ordered before all
6778 remote mounts and may be used to start services before all
6781 * Added Mageia support
6783 * Add bash completion for systemd-loginctl
6785 * Actively monitor PID file creation for daemons which exit in
6786 the parent process before having finished writing the PID
6787 file in the daemon process. Daemons which do this need to be
6788 fixed (i.e. PID file creation must have finished before the
6789 parent exits), but we now react a bit more gracefully to them.
6791 * Add colourful boot output, mimicking the well-known output
6792 of existing distributions.
6794 * New option PassCredentials= for socket units, for
6795 compatibility with a recent kernel ABI breakage.
6797 * /etc/rc.local is now hooked in via a generator binary, and
6798 thus will no longer act as synchronization point during
6801 * systemctl list-unit-files now supports --root=.
6803 * systemd-tmpfiles now understands two new commands: z, Z for
6804 relabelling files according to the SELinux database. This is
6805 useful to apply SELinux labels to specific files in /sys,
6808 * Output of SysV services is now forwarded to both the console
6809 and the journal by default, not only just the console.
6811 * New man pages for all APIs from libsystemd-login.
6813 * The build tree got reorganized and the build system is a
6814 lot more modular allowing embedded setups to specifically
6815 select the components of systemd they are interested in.
6817 * Support for Linux systems lacking the kernel VT subsystem is
6820 * configure's --with-rootdir= got renamed to
6821 --with-rootprefix= to follow the naming used by udev and
6824 * Unless specified otherwise we will now install to /usr instead
6825 of /usr/local by default.
6827 * Processes with '@' in argv[0][0] are now excluded from the
6828 final shut-down killing spree, following the logic explained
6830 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
6832 * All processes remaining in a service cgroup when we enter
6833 the START or START_PRE states are now killed with
6834 SIGKILL. That means it is no longer possible to spawn
6835 background processes from ExecStart= lines (which was never
6836 supported anyway, and bad style).
6838 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
6839 reloading of units together.
6841 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
6842 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
6843 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
6844 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
6845 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek