1 systemd System and Service Manager
5 * Support for the cpuset cgroups v2 controller has been added.
6 Processes may be restricted to specific CPUs using the new
7 AllowedCPUs= setting, and to specific memory NUMA nodes using the new
8 AllowedMemoryNodes= setting.
10 * The signal used in restart jobs (as opposed to e.g. stop jobs) may
11 now be configured using a new RestartKillSignal= setting. This
12 allows units which signals to request termination to implement
13 different behaviour when stopping in preparation for a restart.
15 * "systemctl clean" may now be used also for socket, mount, and swap
18 * systemd will also read configuration options from the EFI variable
19 SystemdOptions. This may be used to configure systemd behaviour when
20 modifying the kernel command line is inconvenient, but configuration
21 on disk is read too late, for example for the options related to
22 cgroup hierarchy setup. 'bootctl systemd-efi-options' may be used to
25 * systemd will now disable printk ratelimits in early boot. This should
26 allow us to capture more logs from the early boot phase where normal
27 storage is not available and the kernel ring buffer is used for
28 logging. Configuration on the kernel command line has higher priority
29 and overrides the systemd setting.
31 systemd programs which log to /dev/kmsg directly use internal
32 ratelimits to prevent runaway logging. (Normally this is only used
33 during early boot, so in practice this change has very little
36 * Unit files now support top level dropin directories of the form
37 <unit_type>.d/ (e.g. service.d/) that may be used to add configuration
38 that affects all corresponding unit files.
40 * systemctl gained support for 'stop --job-mode=triggering' which will
41 stop the specified unit and any units which could trigger it.
43 * Unit status display now includes units triggering and triggered by
46 * The RuntimeMaxSec= setting is now supported by scopes, not just
47 .service units. This is particularly useful for PAM sessions which
48 create a scope unit for the user login. systemd.runtime_max_sec=
49 setting may used with the pam_systemd module to limit the duration
50 of the PAM session, for example for time-limited logins.
52 * A new @pkey system call group is now defined to make it easier to
53 whitelist memory protection syscalls for containers and services
54 which need to use them.
56 * systemd-udevd: removed the 30s timeout for killing stale workers on
57 exit. systemd-udevd now waits for workers to finish. The hard-coded
58 exit timeout of 30s was too short for some large installations, where
59 driver initialization could be prematurely interrupted during initrd
60 processing if the root file system had been mounted and init was
61 preparing to switch root. If udevd is run without systemd and workers
62 are hanging while udevd receives an exit signal, udevd will now exit
63 when udev.event_timeout is reached for the last hanging worker. With
64 systemd, the exit timeout can additionally be configured using
65 TimeoutStopSec= in systemd-udevd.service.
67 * udev now provides a program (fido_id) that identifies FIDO CTAP1
68 ("U2F")/CTAP2 security tokens based on the usage declared in their
69 report and descriptor and outputs suitable environment variables.
70 This replaces the externally maintained whitelists of all known
71 security tokens that were used previously.
73 * Automatically generated autosuspend udev rules for whitelisted
74 devices have been imported from the Chromium OS project. This should
75 improve power saving with many more devices.
77 * udev gained a new "CONST{key}=value" setting that allows matching
78 against system-wide constants without forking a helper binary.
79 Currently "arch" and "virt" keys are supported.
81 * udev now opens CDROMs in non-exclusive mode when querying their
82 capabilities. This should fix issues where other programs trying to
83 use the CDROM cannot gain access to it, but carries a risk of
84 interfering with programs writing to the disk, if they did not open
85 the device in exclusive mode as they should.
87 * systemd-networkd does not create a default route for IPv4 link local
88 addressing anymore. The creation of the route was unexpected and was
89 breaking routing in various cases, but people who rely on it being
90 created implicitly will need to adjust. Such a route may be requested
91 with DefaultRouteOnDevice=yes.
93 Similarly, systemd-networkd will not assign a link-local IPv6 address
94 when IPv6 link-local routing is not enabled.
96 * Receive and transmit buffers may now be configured on links with
97 the new RxBufferSize= and TxBufferSize= settings.
99 * systemd-networkd may now advertise additional IPv6 routes. A new
100 [IPv6RoutePrefix] section with Route= and LifetimeSec= options is
103 * systemd-networkd may now configure "next hop" routes using the
104 [NextHop] section and Gateway= and Id= settings.
106 * systemd-networkd will now retain DHCP config on restarts by default
107 (but this may be overridden using the KeepConfiguration= setting).
108 The default for SendRelease= has been changed to true.
110 * The DHCPv4 client now uses the OPTION_INFORMATION_REFRESH_TIME option
111 received from the server.
113 The client will use the received SIP server list if UseSIP=yes is
116 The client may be configured to request specific options from the
117 server using a new RequestOptions= setting.
119 The client may be configured to send arbitrary options to the server
120 using a new SendOption= setting.
122 A new IPServiceType= setting has been added to configure the "IP
123 service type" value used by the client.
125 * The DHCPv6 client learnt a new PrefixDelegationHint= option to
126 request prefix hints in the DHCPv6 solicitation.
128 * The DHCPv4 server may be configured to send arbitrary options using
129 a new SendOption= setting.
131 * The DHCPv4 server may now be configured to emit SIP server list using
132 the new EmitSIP= and SIP= settings.
134 * systemd-networkd and networkctl may now renew DHCP leases on demand.
135 networkctl has a new 'networkctl renew' verb.
137 * systemd-networkd may now reconfigure links on demand. networkctl
138 gained two new verbs: "reload" will reload the configuration, and
139 "reconfigure DEVICE…" will reconfigure one or more devices.
141 * .network files may now match on SSID and BSSID of a wireless network,
142 i.e. the access point name and hardware address using the new SSID=
143 and BSSID= options. networkctl will display the current SSID and
144 BSSID for wireless links.
146 .network files may also match on the wireless network type using the
147 new WLANInterfaceType= option.
149 * systemd-networkd now includes default configuration that enables
150 link-local addressing when connected to an ad-hoc wireless network.
152 * systemd-networkd may configure the Traffic Control queueing
153 disciplines in the kernel using the new
154 [TrafficControlQueueingDiscipline] section and Parent=,
155 NetworkEmulatorDelaySec=, NetworkEmulatorDelayJitterSec=,
156 NetworkEmulatorPacketLimit=, NetworkEmulatorLossRate=,
157 NetworkEmulatorDuplicateRate= settings.
159 * systemd-tmpfiles gained a new w+ setting to append to files.
161 * systemd-analyze dump will now report when the memory configuration in
162 the kernel does not match what systemd has configured (usually,
163 because some external program has modified the kernel configuration
166 * systemd-analyze gained a new --base-time= switch instructs the
167 'calendar' verb to resolve times relative to that timestamp instead
170 * journalctl --update-catalog now produces deterministic output (making
171 reproducible image builds easier).
173 * A new devicetree-overlay setting is now documented in the Boot Loader
176 * The default value of the WatchdogSec= setting used in systemd
177 services (the ones bundled with the project itself) may be set at
178 configuration time using the -Dservice-watchdog= setting. If set to
179 empty, the watchdogs will be disabled.
181 * systemd-resolved validates IP addresses in certificates now when GnuTLS
184 * libcryptsetup >= 2.0.1 is now required.
186 * A configuration option -Duser-path= may be used to override the $PATH
187 used by the user service manager. The default is again to use the same
188 path as the system manager.
190 * The systemd-id128 tool gained a new switch "-u" (or "--uuid") for
191 outputting the 128bit IDs in UUID format (i.e. in the "canonical
194 * Service units gained a new sandboxing option ProtectKernelLogs= which
195 makes sure the program cannot get direct access to the kernel log
196 buffer anymore, i.e. the syslog() system call (not to be confused
197 with the API of the same name in libc, which is not affected), the
198 /proc/kmsg and /dev/kmsg nodes and the CAP_SYSLOG capability are made
199 inaccessible to the service. It's recommended to enable this setting
200 for all services that should not be able to read from or write to the
201 kernel log buffer, which are probably almost all.
203 Contributions from: Aaron Plattner, Alcaro, Anita Zhang, Balint Reczey,
204 Bastien Nocera, Baybal Ni, Benjamin Bouvier, Benjamin Gilbert, Carlo
205 Teubner, cbzxt, Chen Qi, Chris Down, Christian Rebischke, Claudio
206 Zumbo, ClydeByrdIII, crashfistfight, Cyprien Laplace, Daniel Edgecumbe,
207 Daniel Gorbea, Daniel Rusek, Daniel Stuart, Dan Streetman, David
208 Pedersen, David Tardon, Dimitri John Ledkov, Dominique Martinet, Donald
209 A. Cupp Jr, Evgeny Vereshchagin, Fabian Henneke, Filipe Brandenburger,
210 Franck Bui, Frantisek Sumsal, Georg Müller, Hans de Goede, Haochen
211 Tong, HATAYAMA Daisuke, Iwan Timmer, Jan Janssen, Jan Kundrát, Jan
212 Synacek, Jan Tojnar, Jay Strict, Jérémy Rosen, Jóhann B. Guðmundsson,
213 Jonas Jelten, Jonas Thelemann, Justin Trudell, J. Xing, Kai-Heng Feng,
214 Kenneth D'souza, Kevin Becker, Kevin Kuehler, Lennart Poettering,
215 Léonard Gérard, Lorenz Bauer, Luca Boccassi, Maciej Stanczew, Mario
216 Limonciello, Marko Myllynen, Mark Stosberg, Martin Wilck, matthiasroos,
217 Michael Biebl, Michael Olbrich, Michael Tretter, Michal Sekletar,
218 Michal Sekletár, Michal Suchanek, Mike Gilbert, Mike Kazantsev, Nicolas
219 Douma, nikolas, Norbert Lange, pan93412, Pascal de Bruijn, Paul Menzel,
220 Pavel Hrdina, Peter Wu, Philip Withnall, Piotr Drąg, Rafael Fontenelle,
221 Renaud Métrich, Riccardo Schirone, RoadrunnerWMC, Ronan Pigott, Ryan
222 Attard, Sebastian Wick, Serge, Siddharth Chandrasekara, Steve Ramage,
223 Steve Traylen, Susant Sahani, Thibault Nélis, Tim Teichmann, Tom
224 Fitzhenry, Tommy J, Torsten Hilbrich, Vito Caputo, ypf791, Yu Watanabe,
225 Zach Smith, Zbigniew Jędrzejewski-Szmek
231 * This release enables unprivileged programs (i.e. requiring neither
232 setuid nor file capabilities) to send ICMP Echo (i.e. ping) requests
233 by turning on the "net.ipv4.ping_group_range" sysctl of the Linux
234 kernel for the whole UNIX group range, i.e. all processes. This
235 change should be reasonably safe, as the kernel support for it was
236 specifically implemented to allow safe access to ICMP Echo for
237 processes lacking any privileges. If this is not desirable, it can be
238 disabled again by setting the parameter to "1 0".
240 * Previously, filters defined with SystemCallFilter= would have the
241 effect that any calling of an offending system call would terminate
242 the calling thread. This behaviour never made much sense, since
243 killing individual threads of unsuspecting processes is likely to
244 create more problems than it solves. With this release the default
245 action changed from killing the thread to killing the whole
246 process. For this to work correctly both a kernel version (>= 4.14)
247 and a libseccomp version (>= 2.4.0) supporting this new seccomp
248 action is required. If an older kernel or libseccomp is used the old
249 behaviour continues to be used. This change does not affect any
250 services that have no system call filters defined, or that use
251 SystemCallErrorNumber= (and thus see EPERM or another error instead
252 of being killed when calling an offending system call). Note that
253 systemd documentation always claimed that the whole process is
254 killed. With this change behaviour is thus adjusted to match the
257 * On 64 bit systems, the "kernel.pid_max" sysctl is now bumped to
258 4194304 by default, i.e. the full 22bit range the kernel allows, up
259 from the old 16bit range. This should improve security and
260 robustness, as PID collisions are made less likely (though certainly
261 still possible). There are rumours this might create compatibility
262 problems, though at this moment no practical ones are known to
263 us. Downstream distributions are hence advised to undo this change in
264 their builds if they are concerned about maximum compatibility, but
265 for everybody else we recommend leaving the value bumped. Besides
266 improving security and robustness this should also simplify things as
267 the maximum number of allowed concurrent tasks was previously bounded
268 by both "kernel.pid_max" and "kernel.threads-max" and now effectively
269 only a single knob is left ("kernel.threads-max"). There have been
270 concerns that usability is affected by this change because larger PID
271 numbers are harder to type, but we believe the change from 5 digits
272 to 7 digits doesn't hamper usability.
274 * MemoryLow= and MemoryMin= gained hierarchy-aware counterparts,
275 DefaultMemoryLow= and DefaultMemoryMin=, which can be used to
276 hierarchically set default memory protection values for a particular
277 subtree of the unit hierarchy.
279 * Memory protection directives can now take a value of zero, allowing
280 explicit opting out of a default value propagated by an ancestor.
282 * systemd now defaults to the "unified" cgroup hierarchy setup during
283 build-time, i.e. -Ddefault-hierarchy=unified is now the build-time
284 default. Previously, -Ddefault-hierarchy=hybrid was the default. This
285 change reflects the fact that cgroupsv2 support has matured
286 substantially in both systemd and in the kernel, and is clearly the
287 way forward. Downstream production distributions might want to
288 continue to use -Ddefault-hierarchy=hybrid (or even =legacy) for
289 their builds as unfortunately the popular container managers have not
290 caught up with the kernel API changes.
292 * Man pages are not built by default anymore (html pages were already
293 disabled by default), to make development builds quicker. When
294 building systemd for a full installation with documentation, meson
295 should be called with -Dman=true and/or -Dhtml=true as appropriate.
296 The default was changed based on the assumption that quick one-off or
297 repeated development builds are much more common than full optimized
298 builds for installation, and people need to pass various other
299 options to when doing "proper" builds anyway, so the gain from making
300 development builds quicker is bigger than the one time disruption for
303 Two scripts are created in the *build* directory to generate and
304 preview man and html pages on demand, e.g.:
306 build/man/man systemctl
307 build/man/html systemd.index
309 * libidn2 is used by default if both libidn2 and libidn are installed.
310 Please use -Dlibidn=true if libidn is preferred.
312 * The D-Bus "wire format" of the CPUAffinity= attribute is changed on
313 big-endian machines. Before, bytes were written and read in native
314 machine order as exposed by the native libc __cpu_mask interface.
315 Now, little-endian order is always used (CPUs 0–7 are described by
316 bits 0–7 in byte 0, CPUs 8–15 are described by byte 1, and so on).
317 This change fixes D-Bus calls that cross endianness boundary.
319 The presentation format used for CPUAffinity= by "systemctl show" and
320 "systemd-analyze dump" is changed to present CPU indices instead of
321 the raw __cpu_mask bitmask. For example, CPUAffinity=0-1 would be
322 shown as CPUAffinity=03000000000000000000000000000… (on
323 little-endian) or CPUAffinity=00000000000000300000000000000… (on
324 64-bit big-endian), and is now shown as CPUAffinity=0-1, matching the
325 input format. The maximum integer that will be printed in the new
326 format is 8191 (four digits), while the old format always used a very
327 long number (with the length varying by architecture), so they can be
328 unambiguously distinguished.
330 * /usr/sbin/halt.local is no longer supported. Implementation in
331 distributions was inconsistent and it seems this functionality was
334 To replace this functionality, users should:
335 - either define a new unit and make it a dependency of final.target
336 (systemctl add-wants final.target my-halt-local.service)
337 - or move the shutdown script to /usr/lib/systemd/system-shutdown/
338 and ensure that it accepts "halt", "poweroff", "reboot", and
339 "kexec" as an argument, see the description in systemd-shutdown(8).
341 * When a [Match] section in .link or .network file is empty (contains
342 no match patterns), a warning will be emitted. Please add any "match
343 all" pattern instead, e.g. OriginalName=* or Name=* in case all
344 interfaces should really be matched.
346 * A new setting NUMAPolicy= may be used to set process memory
347 allocation policy. This setting can be specified in
348 /etc/systemd/system.conf and hence will set the default policy for
349 PID1. The default policy can be overridden on a per-service
350 basis. The related setting NUMAMask= is used to specify NUMA node
351 mask that should be associated with the selected policy.
353 * PID 1 will now listen to Out-Of-Memory (OOM) events the kernel
354 generates when processes it manages are reaching their memory limits,
355 and will place their units in a special state, and optionally kill or
358 * The service manager will now expose bus properties for the IO
359 resources used by units. This information is also shown in "systemctl
360 status" now (for services that have IOAccounting=yes set). Moreover,
361 the IO accounting data is included in the resource log message
362 generated whenever a unit stops.
364 * Units may now configure an explicit time-out to wait for when killed
365 with SIGABRT, for example when a service watchdog is hit. Previously,
366 the regular TimeoutStopSec= time-out was applied in this case too —
367 now a separate time-out may be set using TimeoutAbortSec=.
369 * Services may now send a special WATCHDOG=trigger message with
370 sd_notify() to trigger an immediate "watchdog missed" event, and thus
371 trigger service termination. This is useful both for testing watchdog
372 handling, but also for defining error paths in services, that shall
373 be handled the same way as watchdog events.
375 * There are two new per-unit settings IPIngressFilterPath= and
376 IPEgressFilterPath= which allow configuration of a BPF program
377 (usually by specifying a path to a program uploaded to /sys/fs/bpf/)
378 to apply to the IP packet ingress/egress path of all processes of a
379 unit. This is useful to allow running systemd services with BPF
380 programs set up externally.
382 * systemctl gained a new "clean" verb for removing the state, cache,
383 runtime or logs directories of a service while it is terminated. The
384 new verb may also be used to remove the state maintained on disk for
385 timer units that have Persistent= configured.
387 * During the last phase of shutdown systemd will now automatically
388 increase the log level configured in the "kernel.printk" sysctl so
389 that any relevant loggable events happening during late shutdown are
390 made visible. Previously, loggable events happening so late during
391 shutdown were generally lost if the "kernel.printk" sysctl was set to
392 high thresholds, as regular logging daemons are terminated at that
393 time and thus nothing is written to disk.
395 * If processes terminated during the last phase of shutdown do not exit
396 quickly systemd will now show their names after a short time, to make
397 debugging easier. After a longer time-out they are forcibly killed,
400 * journalctl (and the other tools that display logs) will now highlight
401 warnings in yellow (previously, both LOG_NOTICE and LOG_WARNING where
402 shown in bright bold, now only LOG_NOTICE is). Moreover, audit logs
403 are now shown in blue color, to separate them visually from regular
404 logs. References to configuration files are now turned into clickable
405 links on terminals that support that.
407 * systemd-journald will now stop logging to /var/log/journal during
408 shutdown when /var/ is on a separate mount, so that it can be
409 unmounted safely during shutdown.
411 * systemd-resolved gained support for a new 'strict' DNS-over-TLS mode.
413 * systemd-resolved "Cache=" configuration option in resolved.conf has
414 been extended to also accept the 'no-negative' value. Previously,
415 only a boolean option was allowed (yes/no), having yes as the
416 default. If this option is set to 'no-negative', negative answers are
417 not cached while the old cache heuristics are used positive answers.
418 The default remains unchanged.
420 * The predictable naming scheme for network devices now supports
421 generating predictable names for "netdevsim" devices.
423 Moreover, the "en" prefix was dropped from the ID_NET_NAME_ONBOARD
426 Those two changes form a new net.naming-policy-scheme= entry.
427 Distributions which want to preserve naming stability may want to set
428 the -Ddefault-net-naming-scheme= configuration option.
430 * systemd-networkd now supports MACsec, nlmon, IPVTAP and Xfrm
433 * systemd-networkd's bridge FDB support now allows configuration of a
434 destination address for each entry (Destination=), as well as the
435 VXLAN VNI (VNI=), as well as an option to declare what an entry is
436 associated with (AssociatedWith=).
438 * systemd-networkd's DHCPv4 support now understands a new MaxAttempts=
439 option for configuring the maximum number of DHCP lease requests. It
440 also learnt a new BlackList= option for blacklisting DHCP servers (a
441 similar setting has also been added to the IPv6 RA client), as well
442 as a SendRelease= option for configuring whether to send a DHCP
443 RELEASE message when terminating.
445 * systemd-networkd's DHCPv4 and DHCPv6 stacks can now be configured
446 separately in the [DHCPv4] and [DHCPv6] sections.
448 * systemd-networkd's DHCP support will now optionally create an
449 implicit host route to the DNS server specified in the DHCP lease, in
450 addition to the routes listed explicitly in the lease. This should
451 ensure that in multi-homed systems DNS traffic leaves the systems on
452 the interface that acquired the DNS server information even if other
453 routes such as default routes exist. This behaviour may be turned on
454 with the new RoutesToDNS= option.
456 * systemd-networkd's VXLAN support gained a new option
457 GenericProtocolExtension= for enabling VXLAN Generic Protocol
458 Extension support, as well as IPDoNotFragment= for setting the IP
459 "Don't fragment" bit on outgoing packets. A similar option has been
460 added to the GENEVE support.
462 * In systemd-networkd's [Route] section you may now configure
463 FastOpenNoCookie= for configuring per-route TCP fast-open support, as
464 well as TTLPropagate= for configuring Label Switched Path (LSP) TTL
465 propagation. The Type= setting now supports local, broadcast,
466 anycast, multicast, any, xresolve routes, too.
468 * systemd-networkd's [Network] section learnt a new option
469 DefaultRouteOnDevice= for automatically configuring a default route
470 onto the network device.
472 * systemd-networkd's bridging support gained two new options ProxyARP=
473 and ProxyARPWifi= for configuring proxy ARP behaviour as well as
474 MulticastRouter= for configuring multicast routing behaviour. A new
475 option MulticastIGMPVersion= may be used to change bridge's multicast
476 Internet Group Management Protocol (IGMP) version.
478 * systemd-networkd's FooOverUDP support gained the ability to configure
479 local and peer IP addresses via Local= and Peer=. A new option
480 PeerPort= may be used to configure the peer's IP port.
482 * systemd-networkd's TUN support gained a new setting VnetHeader= for
483 tweaking Generic Segment Offload support.
485 * networkctl gained a new "delete" command for removing virtual network
486 devices, as well as a new "--stats" switch for showing device
489 * networkd.conf gained a new setting SpeedMeter= and
490 SpeedMeterIntervalSec=, to measure bitrate of network interfaces. The
491 measured speed may be shown by 'networkctl status'.
493 * "networkctl status" now displays MTU and queue lengths, and more
494 detailed information about VXLAN and bridge devices.
496 * systemd-networkd's .network and .link files gained a new Property=
497 setting in the [Match] section, to match against devices with
498 specific udev properties.
500 * systemd-networkd's tunnel support gained a new option
501 AssignToLoopback= for selecting whether to use the loopback device
502 "lo" as underlying device.
504 * systemd-networkd's MACAddress= setting in the [Neighbor] section has
505 been renamed to LinkLayerAddress=, and it now allows configuration of
508 * systemd-networkd's handling of the kernel's disable_ipv6 sysctl is
509 simplified: systemd-networkd will disable the sysctl (enable IPv6) if
510 IPv6 configuration (static or DHCPv6) was found for a given
511 interface. It will not touch the sysctl otherwise.
513 * The order of entries is $PATH used by the user manager instance was
514 changed to put bin/ entries before the corresponding sbin/ entries.
515 It is recommended to not rely on this order, and only ever have one
516 binary with a given name in the system paths under /usr.
518 * A new tool systemd-network-generator has been added that may generate
519 .network, .netdev and .link files from IP configuration specified on
520 the kernel command line in the format used by Dracut.
522 * The CriticalConnection= setting in .network files is now deprecated,
523 and replaced by a new KeepConfiguration= setting which allows more
524 detailed configuration of the IP configuration to keep in place.
526 * systemd-analyze gained a few new verbs:
528 - "systemd-analyze timestamp" parses and converts timestamps. This is
529 similar to the existing "systemd-analyze calendar" command which
530 does the same for recurring calendar events.
532 - "systemd-analyze timespan" parses and converts timespans (i.e.
533 durations as opposed to points in time).
535 - "systemd-analyze condition" will parse and test ConditionXYZ=
538 - "systemd-analyze exit-status" will parse and convert exit status
539 codes to their names and back.
541 - "systemd-analyze unit-files" will print a list of all unit
542 file paths and unit aliases.
544 * SuccessExitStatus=, RestartPreventExitStatus=, and
545 RestartForceExitStatus= now accept exit status names (e.g. "DATAERR"
546 is equivalent to "65"). Those exit status name mappings may be
547 displayed with the systemd-analyze exit-status verb describe above.
549 * systemd-logind now exposes a per-session SetBrightness() bus call,
550 which may be used to securely change the brightness of a kernel
551 brightness device, if it belongs to the session's seat. By using this
552 call unprivileged clients can make changes to "backlight" and "leds"
553 devices securely with strict requirements on session membership.
554 Desktop environments may use this to generically make brightness
555 changes to such devices without shipping private SUID binaries or
556 udev rules for that purpose.
558 * "udevadm info" gained a --wait-for-initialization switch to wait for
559 a device to be initialized.
561 * systemd-hibernate-resume-generator will now look for resumeflags= on
562 the kernel command line, which is similar to rootflags= and may be
563 used to configure device timeout for the hibernation device.
565 * sd-event learnt a new API call sd_event_source_disable_unref() for
566 disabling and unref'ing an event source in a single function. A
567 related call sd_event_source_disable_unrefp() has been added for use
568 with gcc's cleanup extension.
570 * The sd-id128.h public API gained a new definition
571 SD_ID128_UUID_FORMAT_STR for formatting a 128bit ID in UUID format
574 * "busctl introspect" gained a new switch --xml-interface for dumping
575 XML introspection data unmodified.
577 * PID 1 may now show the unit name instead of the unit description
578 string in its status output during boot. This may be configured in
579 the StatusUnitFormat= setting in /etc/systemd/system.conf or the
580 kernel command line option systemd.status_unit_format=.
582 * PID 1 now understands a new option KExecWatchdogSec= in
583 /etc/systemd/system.conf to set a watchdog timeout for kexec reboots.
584 Previously watchdog functionality was only available for regular
585 reboots. The new setting defaults to off, because we don't know in
586 the general case if the watchdog will be reset after kexec (some
587 drivers do reset it, but not all), and the new userspace might not be
588 configured to handle the watchdog.
590 Moreover, the old ShutdownWatchdogSec= setting has been renamed to
591 RebootWatchdogSec= to more clearly communicate what it is about. The
592 old name is still accepted for compatibility.
594 * The systemd.debug_shell kernel command line option now optionally
595 takes a tty name to spawn the debug shell on, which allows a
596 different tty to be selected than the built-in default.
598 * Service units gained a new ExecCondition= setting which will run
599 before ExecStartPre= and either continue execution of the unit (for
600 clean exit codes), stop execution without marking the unit failed
601 (for exit codes 1 through 254), or stop execution and fail the unit
602 (for exit code 255 or abnormal termination).
604 * A new service systemd-pstore.service has been added that pulls data
605 from /sys/fs/pstore/ and saves it to /var/lib/pstore for later
608 * timedatectl gained new verbs for configuring per-interface NTP
609 service configuration for systemd-timesyncd.
611 * "localectl list-locales" won't list non-UTF-8 locales anymore. It's
612 2019. (You can set non-UTF-8 locales though, if you know their name.)
614 * If variable assignments in sysctl.d/ files are prefixed with "-" any
615 failures to apply them are now ignored.
617 * systemd-random-seed.service now optionally credits entropy when
618 applying the seed to the system. Set $SYSTEMD_RANDOM_SEED_CREDIT to
619 true for the service to enable this behaviour, but please consult the
620 documentation first, since this comes with a couple of caveats.
622 * systemd-random-seed.service is now a synchronization point for full
623 initialization of the kernel's entropy pool. Services that require
624 /dev/urandom to be correctly initialized should be ordered after this
627 * The systemd-boot boot loader has been updated to optionally maintain
628 a random seed file in the EFI System Partition (ESP). During the boot
629 phase, this random seed is read and updated with a new seed
630 cryptographically derived from it. Another derived seed is passed to
631 the OS. The latter seed is then credited to the kernel's entropy pool
632 very early during userspace initialization (from PID 1). This allows
633 systems to boot up with a fully initialized kernel entropy pool from
634 earliest boot on, and thus entirely removes all entropy pool
635 initialization delays from systems using systemd-boot. Special care
636 is taken to ensure different seeds are derived on system images
637 replicated to multiple systems. "bootctl status" will show whether
638 a seed was received from the boot loader.
640 * bootctl gained two new verbs:
642 - "bootctl random-seed" will generate the file in ESP and an EFI
643 variable to allow a random seed to be passed to the OS as described
646 - "bootctl is-installed" checks whether systemd-boot is currently
649 * bootctl will warn if it detects that boot entries are misconfigured
650 (for example if the kernel image was removed without purging the
653 * A new document has been added describing systemd's use and support
654 for the kernel's entropy pool subsystem:
656 https://systemd.io/RANDOM_SEEDS
658 * When the system is hibernated the swap device to write the
659 hibernation image to is now automatically picked from all available
660 swap devices, preferring the swap device with the highest configured
661 priority over all others, and picking the device with the most free
662 space if there are multiple devices with the highest priority.
664 * /etc/crypttab support has learnt a new keyfile-timeout= per-device
665 option that permits selecting the timout how long to wait for a
666 device with an encryption key before asking for the password.
668 * IOWeight= has learnt to properly set the IO weight when using the
669 BFQ scheduler officially found in kernels 5.0+.
671 * A new mailing list has been created for reporting of security issues:
672 systemd-security@redhat.com. For mode details, see
673 https://systemd.io/CONTRIBUTING#security-vulnerability-reports.
675 Contributions from: Aaron Barany, Adrian Bunk, Alan Jenkins, Albrecht
676 Lohofener, Andrej Valek, Anita Zhang, Arian van Putten, Balint Reczey,
677 Bastien Nocera, Ben Boeckel, Benjamin Robin, camoz, Chen Qi, Chris
678 Chiu, Chris Down, Christian Kellner, Clinton Roy, Connor Reeder, Daniel
679 Black, Daniele Medri, Dan Streetman, Dave Reisner, Dave Ross, David
680 Art, David Tardon, Debarshi Ray, Dimitri John Ledkov, Dominick Grift,
681 Donald Buczek, Douglas Christman, Eric DeVolder, EtherGraf, Evgeny
682 Vereshchagin, Feldwor, Felix Riemann, Florian Dollinger, Francesco
683 Pennica, Franck Bui, Frantisek Sumsal, Franz Pletz, frederik, Hans
684 de Goede, Iago López Galeiras, Insun Pyo, Ivan Shapovalov, Iwan Timmer,
685 Jack, Jakob Unterwurzacher, Jan Chren, Jan Klötzke, Jan Losinski, Jan
686 Pokorný, Jan Synacek, Jan-Michael Brummer, Jeka Pats, Jeremy Soller,
687 Jérémy Rosen, Jiri Pirko, Joe Lin, Joerg Behrmann, Joe Richey, Jóhann
688 B. Guðmundsson, Johannes Christ, Johannes Schmitz, Jonathan Rouleau,
689 Jorge Niedbalski, Kai Krakow, Kai Lüke, Karel Zak, Kashyap Chamarthy,
690 Krayushkin Konstantin, Lennart Poettering, Lubomir Rintel, Luca
691 Boccassi, Luís Ferreira, Marc-André Lureau, Markus Felten, Martin Pitt,
692 Matthew Leeds, Mattias Jernberg, Michael Biebl, Michael Olbrich,
693 Michael Prokop, Michael Stapelberg, Michael Zhivich, Michal Koutný,
694 Michal Sekletar, Mike Gilbert, Milan Broz, Miroslav Lichvar, mpe85,
695 Mr-Foo, Network Silence, Oliver Harley, pan93412, Paul Menzel, pEJipE,
696 Peter A. Bigot, Philip Withnall, Piotr Drąg, Rafael Fontenelle, Roberto
697 Santalla, Ronan Pigott, root, RussianNeuroMancer, Sebastian Jennen,
698 shinygold, Shreyas Behera, Simon Schricker, Susant Sahani, Thadeu Lima
699 de Souza Cascardo, Theo Ouzhinski, Thiebaud Weksteen, Thomas Haller,
700 Thomas Weißschuh, Tomas Mraz, Tommi Rantala, Topi Miettinen, VD-Lycos,
701 ven, Wieland Hoffmann, William A. Kennington III, William Wold, Xi
702 Ruoyao, Yuri Chornoivan, Yu Watanabe, Zach Smith, Zbigniew
703 Jędrzejewski-Szmek, Zhang Xianwei
705 – Camerino, 2019-09-03
709 * In .link files, MACAddressPolicy=persistent (the default) is changed
710 to cover more devices. For devices like bridges, tun, tap, bond, and
711 similar interfaces that do not have other identifying information,
712 the interface name is used as the basis for persistent seed for MAC
713 and IPv4LL addresses. The way that devices that were handled
714 previously is not changed, and this change is about covering more
715 devices then previously by the "persistent" policy.
717 MACAddressPolicy=random may be used to force randomized MACs and
718 IPv4LL addresses for a device if desired.
720 Hint: the log output from udev (at debug level) was enhanced to
721 clarify what policy is followed and which attributes are used.
722 `SYSTEMD_LOG_LEVEL=debug udevadm test-builtin net_setup_link /sys/class/net/<name>`
723 may be used to view this.
725 Hint: if a bridge interface is created without any slaves, and gains
726 a slave later, then now the bridge does not inherit slave's MAC.
727 To inherit slave's MAC, for example, create the following file:
729 # /etc/systemd/network/98-bridge-inherit-mac.link
734 MACAddressPolicy=none
737 * The .device units generated by systemd-fstab-generator and other
738 generators do not automatically pull in the corresponding .mount unit
739 as a Wants= dependency. This means that simply plugging in the device
740 will not cause the mount unit to be started automatically. But please
741 note that the mount unit may be started for other reasons, in
742 particular if it is part of local-fs.target, and any unit which
743 (transitively) depends on local-fs.target is started.
745 * networkctl list/status/lldp now accept globbing wildcards for network
746 interface names to match against all existing interfaces.
748 * The $PIDFILE environment variable is set to point the absolute path
749 configured with PIDFile= for processes of that service.
751 * The fallback DNS server list was augmented with Cloudflare public DNS
752 servers. Use `-Ddns-servers=` to set a different fallback.
754 * A new special target usb-gadget.target will be started automatically
755 when a USB Device Controller is detected (which means that the system
756 is a USB peripheral).
758 * A new unit setting CPUQuotaPeriodSec= assigns the time period
759 relatively to which the CPU time quota specified by CPUQuota= is
762 * A new unit setting ProtectHostname= may be used to prevent services
763 from modifying hostname information (even if they otherwise would
764 have privileges to do so).
766 * A new unit setting NetworkNamespacePath= may be used to specify a
767 namespace for service or socket units through a path referring to a
768 Linux network namespace pseudo-file.
770 * The PrivateNetwork= setting and JoinsNamespaceOf= dependencies now
771 have an effect on .socket units: when used the listening socket is
772 created within the configured network namespace instead of the host
775 * ExecStart= command lines in unit files may now be prefixed with ':'
776 in which case environment variable substitution is
777 disabled. (Supported for the other ExecXYZ= settings, too.)
779 * .timer units gained two new boolean settings OnClockChange= and
780 OnTimezoneChange= which may be used to also trigger a unit when the
781 system clock is changed or the local timezone is
782 modified. systemd-run has been updated to make these options easily
783 accessible from the command line for transient timers.
785 * Two new conditions for units have been added: ConditionMemory= may be
786 used to conditionalize a unit based on installed system
787 RAM. ConditionCPUs= may be used to conditionalize a unit based on
790 * The @default system call filter group understood by SystemCallFilter=
791 has been updated to include the new rseq() system call introduced in
794 * A new time-set.target has been added that indicates that the system
795 time has been set from a local source (possibly imprecise). The
796 existing time-sync.target is stronger and indicates that the time has
797 been synchronized with a precise external source. Services where
798 approximate time is sufficient should use the new target.
800 * "systemctl start" (and related commands) learnt a new
801 --show-transaction option. If specified brief information about all
802 jobs queued because of the requested operation is shown.
804 * systemd-networkd recognizes a new operation state 'enslaved', used
805 (instead of 'degraded' or 'carrier') for interfaces which form a
806 bridge, bond, or similar, and an new 'degraded-carrier' operational
807 state used for the bond or bridge master interface when one of the
808 enslaved devices is not operational.
810 * .network files learnt the new IgnoreCarrierLoss= option for leaving
811 networks configured even if the carrier is lost.
813 * The RequiredForOnline= setting in .network files may now specify a
814 minimum operational state required for the interface to be considered
815 "online" by systemd-networkd-wait-online. Related to this
816 systemd-networkd-wait-online gained a new option --operational-state=
817 to configure the same, and its --interface= option was updated to
818 optionally also take an operational state specific for an interface.
820 * systemd-networkd-wait-online gained a new setting --any for waiting
821 for only one of the requested interfaces instead of all of them.
823 * systemd-networkd now implements L2TP tunnels.
825 * Two new .network settings UseAutonomousPrefix= and UseOnLinkPrefix=
826 may be used to cause autonomous and onlink prefixes received in IPv6
827 Router Advertisements to be ignored.
829 * New MulticastFlood=, NeighborSuppression=, and Learning= .network
830 file settings may be used to tweak bridge behaviour.
832 * The new TripleSampling= option in .network files may be used to
833 configure CAN triple sampling.
835 * A new .netdev settings PrivateKeyFile= and PresharedKeyFile= may be
836 used to point to private or preshared key for a WireGuard interface.
838 * /etc/crypttab now supports the same-cpu-crypt and
839 submit-from-crypt-cpus options to tweak encryption work scheduling
842 * systemd-tmpfiles will now take a BSD file lock before operating on a
843 contents of directory. This may be used to temporarily exclude
844 directories from aging by taking the same lock (useful for example
845 when extracting a tarball into /tmp or /var/tmp as a privileged user,
846 which might create files with really old timestamps, which
847 nevertheless should not be deleted). For further details, see:
849 https://systemd.io/TEMPORARY_DIRECTORIES
851 * systemd-tmpfiles' h line type gained support for the
852 FS_PROJINHERIT_FL ('P') file attribute (introduced in kernel 4.5),
853 controlling project quota inheritance.
855 * sd-boot and bootctl now implement support for an Extended Boot Loader
856 (XBOOTLDR) partition, that is intended to be mounted to /boot, in
857 addition to the ESP partition mounted to /efi or /boot/efi.
858 Configuration file fragments, kernels, initrds and other EFI images
859 to boot will be loaded from both the ESP and XBOOTLDR partitions.
860 The XBOOTLDR partition was previously described by the Boot Loader
861 Specification, but implementation was missing in sd-boot. Support for
862 this concept allows using the sd-boot boot loader in more
863 conservative scenarios where the boot loader itself is placed in the
864 ESP but the kernels to boot (and their metadata) in a separate
867 * A system may now be booted with systemd.volatile=overlay on the
868 kernel command line, which causes the root file system to be set up
869 an overlayfs mount combining the root-only root directory with a
870 writable tmpfs. In this setup, the underlying root device is not
871 modified, and any changes are lost at reboot.
873 * Similar, systemd-nspawn can now boot containers with a volatile
874 overlayfs root with the new --volatile=overlay switch.
876 * systemd-nspawn can now consume OCI runtime bundles using a new
877 --oci-bundle= option. This implementation is fully usable, with most
878 features in the specification implemented, but since this a lot of
879 new code and functionality, this feature should most likely not
880 be used in production yet.
882 * systemd-nspawn now supports various options described by the OCI
883 runtime specification on the command-line and in .nspawn files:
884 --inaccessible=/Inaccessible= may be used to mask parts of the file
885 system tree, --console=/--pipe may be used to configure how standard
886 input, output, and error are set up.
888 * busctl learned the `emit` verb to generate D-Bus signals.
890 * systemd-analyze cat-config may be used to gather and display
891 configuration spread over multiple files, for example system and user
892 presets, tmpfiles.d, sysusers.d, udev rules, etc.
894 * systemd-analyze calendar now takes an optional new parameter
895 --iterations= which may be used to show a maximum number of iterations
896 the specified expression will elapse next.
898 * The sd-bus C API gained support for naming method parameters in the
901 * systemd-logind gained D-Bus APIs to specify the "reboot parameter"
902 the reboot() system call expects.
904 * journalctl learnt a new --cursor-file= option that points to a file
905 from which a cursor should be loaded in the beginning and to which
906 the updated cursor should be stored at the end.
908 * ACRN hypervisor and Windows Subsystem for Linux (WSL) are now
909 detected by systemd-detect-virt (and may also be used in
910 ConditionVirtualization=).
912 * The behaviour of systemd-logind may now be modified with environment
913 variables $SYSTEMD_REBOOT_TO_FIRMWARE_SETUP,
914 $SYSTEMD_REBOOT_TO_BOOT_LOADER_MENU, and
915 $SYSTEMD_REBOOT_TO_BOOT_LOADER_ENTRY. They cause logind to either
916 skip the relevant operation completely (when set to false), or to
917 create a flag file in /run/systemd (when set to true), instead of
918 actually commencing the real operation when requested. The presence
919 of /run/systemd/reboot-to-firmware-setup,
920 /run/systemd/reboot-to-boot-loader-menu, and
921 /run/systemd/reboot-to-boot-loader-entry, may be used by alternative
922 boot loader implementations to replace some steps logind performs
923 during reboot with their own operations.
925 * systemctl can be used to request a reboot into the boot loader menu
926 or a specific boot loader entry with the new --boot-load-menu= and
927 --boot-loader-entry= options to a reboot command. (This requires a
928 boot loader that supports this, for example sd-boot.)
930 * kernel-install will no longer unconditionally create the output
931 directory (e.g. /efi/<machine-id>/<kernel-version>) for boot loader
932 snippets, but will do only if the machine-specific parent directory
933 (i.e. /efi/<machine-id>/) already exists. bootctl has been modified
934 to create this parent directory during sd-boot installation.
936 This makes it easier to use kernel-install with plugins which support
937 a different layout of the bootloader partitions (for example grub2).
939 * During package installation (with `ninja install`), we would create
940 symlinks for getty@tty1.service, systemd-networkd.service,
941 systemd-networkd.socket, systemd-resolved.service,
942 remote-cryptsetup.target, remote-fs.target,
943 systemd-networkd-wait-online.service, and systemd-timesyncd.service
944 in /etc, as if `systemctl enable` was called for those units, to make
945 the system usable immediately after installation. Now this is not
946 done anymore, and instead calling `systemctl preset-all` is
947 recommended after the first installation of systemd.
949 * A new boolean sandboxing option RestrictSUIDSGID= has been added that
950 is built on seccomp. When turned on creation of SUID/SGID files is
953 * The NoNewPrivileges= and the new RestrictSUIDSGID= options are now
954 implied if DynamicUser= is turned on for a service. This hardens
955 these services, so that they neither can benefit from nor create
956 SUID/SGID executables. This is a minor compatibility breakage, given
957 that when DynamicUser= was first introduced SUID/SGID behaviour was
958 unaffected. However, the security benefit of these two options is
959 substantial, and the setting is still relatively new, hence we opted
960 to make it mandatory for services with dynamic users.
962 Contributions from: Adam Jackson, Alexander Tsoy, Andrey Yashkin,
963 Andrzej Pietrasiewicz, Anita Zhang, Balint Reczey, Beniamino Galvani,
964 Ben Iofel, Benjamin Berg, Benjamin Dahlhoff, Chris, Chris Morin,
965 Christopher Wong, Claudius Ellsel, Clemens Gruber, dana, Daniel Black,
966 Davide Cavalca, David Michael, David Rheinsberg, emersion, Evgeny
967 Vereshchagin, Filipe Brandenburger, Franck Bui, Frantisek Sumsal,
968 Giacinto Cifelli, Hans de Goede, Hugo Kindel, Ignat Korchagin, Insun
969 Pyo, Jan Engelhardt, Jonas Dorel, Jonathan Lebon, Jonathon Kowalski,
970 Jörg Sommer, Jörg Thalheim, Jussi Pakkanen, Kai-Heng Feng, Lennart
971 Poettering, Lubomir Rintel, Luís Ferreira, Martin Pitt, Matthias
972 Klumpp, Michael Biebl, Michael Niewöhner, Michael Olbrich, Michal
973 Sekletar, Mike Lothian, Paul Menzel, Piotr Drąg, Riccardo Schirone,
974 Robin Elvedi, Roman Kulikov, Ronald Tschalär, Ross Burton, Ryan
975 Gonzalez, Sebastian Krzyszkowiak, Stephane Chazelas, StKob, Susant
976 Sahani, Sylvain Plantefève, Szabolcs Fruhwald, Taro Yamada, Theo
977 Ouzhinski, Thomas Haller, Tobias Jungel, Tom Yan, Tony Asleson, Topi
978 Miettinen, unixsysadmin, Van Laser, Vesa Jääskeläinen, Yu, Li-Yu,
979 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
985 * The default locale can now be configured at compile time. Otherwise,
986 a suitable default will be selected automatically (one of C.UTF-8,
989 * The version string shown by systemd and other tools now includes the
990 git commit hash when built from git. An override may be specified
991 during compilation, which is intended to be used by distributions to
992 include the package release information.
994 * systemd-cat can now filter standard input and standard error streams
995 for different syslog priorities using the new --stderr-priority=
998 * systemd-journald and systemd-journal-remote reject entries which
999 contain too many fields (CVE-2018-16865) and set limits on the
1000 process' command line length (CVE-2018-16864).
1002 * $DBUS_SESSION_BUS_ADDRESS environment variable is set by pam_systemd
1005 * A new network device NamePolicy "keep" is implemented for link files,
1006 and used by default in 99-default.link (the fallback configuration
1007 provided by systemd). With this policy, if the network device name
1008 was already set by userspace, the device will not be renamed again.
1009 This matches the naming scheme that was implemented before
1010 systemd-240. If naming-scheme < 240 is specified, the "keep" policy
1011 is also enabled by default, even if not specified. Effectively, this
1012 means that if naming-scheme >= 240 is specified, network devices will
1013 be renamed according to the configuration, even if they have been
1014 renamed already, if "keep" is not specified as the naming policy in
1015 the .link file. The 99-default.link file provided by systemd includes
1016 "keep" for backwards compatibility, but it is recommended for user
1017 installed .link files to *not* include it.
1019 The "kernel" policy, which keeps kernel names declared to be
1020 "persistent", now works again as documented.
1022 * kernel-install script now optionally takes the paths to one or more
1023 initrd files, and passes them to all plugins.
1025 * The mincore() system call has been dropped from the @system-service
1026 system call filter group, as it is pretty exotic and may potentially
1027 used for side-channel attacks.
1029 * -fPIE is dropped from compiler and linker options. Please specify
1030 -Db_pie=true option to meson to build position-independent
1031 executables. Note that the meson option is supported since meson-0.49.
1033 * The fs.protected_regular and fs.protected_fifos sysctls, which were
1034 added in Linux 4.19 to make some data spoofing attacks harder, are
1035 now enabled by default. While this will hopefully improve the
1036 security of most installations, it is technically a backwards
1037 incompatible change; to disable these sysctls again, place the
1038 following lines in /etc/sysctl.d/60-protected.conf or a similar file:
1040 fs.protected_regular = 0
1041 fs.protected_fifos = 0
1043 Note that the similar hardlink and symlink protection has been
1044 enabled since v199, and may be disabled likewise.
1046 * The files read from the EnvironmentFile= setting in unit files now
1047 parse backslashes inside quotes literally, matching the behaviour of
1050 * udevadm trigger, udevadm control, udevadm settle and udevadm monitor
1051 now automatically become NOPs when run in a chroot() environment.
1053 * The tmpfiles.d/ "C" line type will now copy directory trees not only
1054 when the destination is so far missing, but also if it already exists
1055 as a directory and is empty. This is useful to cater for systems
1056 where directory trees are put together from multiple separate mount
1057 points but otherwise empty.
1059 * A new function sd_bus_close_unref() (and the associated
1060 sd_bus_close_unrefp()) has been added to libsystemd, that combines
1061 sd_bus_close() and sd_bus_unref() in one.
1063 * udevadm control learnt a new option for --ping for testing whether a
1064 systemd-udevd instance is running and reacting.
1066 * udevadm trigger learnt a new option for --wait-daemon for waiting
1067 systemd-udevd daemon to be initialized.
1069 Contributions from: Aaron Plattner, Alberts Muktupāvels, Alex Mayer,
1070 Ayman Bagabas, Beniamino Galvani, Burt P, Chris Down, Chris Lamb, Chris
1071 Morin, Christian Hesse, Claudius Ellsel, dana, Daniel Axtens, Daniele
1072 Medri, Dave Reisner, David Santamaría Rogado, Diego Canuhe, Dimitri
1073 John Ledkov, Evgeny Vereshchagin, Fabrice Fontaine, Filipe
1074 Brandenburger, Franck Bui, Frantisek Sumsal, govwin, Hans de Goede,
1075 James Hilliard, Jan Engelhardt, Jani Uusitalo, Jan Janssen, Jan
1076 Synacek, Jonathan McDowell, Jonathan Roemer, Jonathon Kowalski, Joost
1077 Heitbrink, Jörg Thalheim, Lance, Lennart Poettering, Louis Taylor,
1078 Lucas Werkmeister, Mantas Mikulėnas, Marc-Antoine Perennou,
1079 marvelousblack, Michael Biebl, Michael Sloan, Michal Sekletar, Mike
1080 Auty, Mike Gilbert, Mikhail Kasimov, Neil Brown, Niklas Hambüchen,
1081 Patrick Williams, Paul Seyfert, Peter Hutterer, Philip Withnall, Roger
1082 James, Ronnie P. Thomas, Ryan Gonzalez, Sam Morris, Stephan Edel,
1083 Stephan Gerhold, Susant Sahani, Taro Yamada, Thomas Haller, Topi
1084 Miettinen, YiFei Zhu, YmrDtnJu, YunQiang Su, Yu Watanabe, Zbigniew
1085 Jędrzejewski-Szmek, zsergeant77, Дамјан Георгиевски
1087 — Berlin, 2019-02-14
1091 * NoNewPrivileges=yes has been set for all long-running services
1092 implemented by systemd. Previously, this was problematic due to
1093 SELinux (as this would also prohibit the transition from PID1's label
1094 to the service's label). This restriction has since been lifted, but
1095 an SELinux policy update is required.
1096 (See e.g. https://github.com/fedora-selinux/selinux-policy/pull/234.)
1098 * DynamicUser=yes is dropped from systemd-networkd.service,
1099 systemd-resolved.service and systemd-timesyncd.service, which was
1100 enabled in v239 for systemd-networkd.service and systemd-resolved.service,
1101 and since v236 for systemd-timesyncd.service. The users and groups
1102 systemd-network, systemd-resolve and systemd-timesync are created
1103 by systemd-sysusers again. Distributors or system administrators
1104 may need to create these users and groups if they not exist (or need
1105 to re-enable DynamicUser= for those units) while upgrading systemd.
1106 Also, the clock file for systemd-timesyncd may need to move from
1107 /var/lib/private/systemd/timesync/clock to /var/lib/systemd/timesync/clock.
1109 * When unit files are loaded from disk, previously systemd would
1110 sometimes (depending on the unit loading order) load units from the
1111 target path of symlinks in .wants/ or .requires/ directories of other
1112 units. This meant that unit could be loaded from different paths
1113 depending on whether the unit was requested explicitly or as a
1114 dependency of another unit, not honouring the priority of directories
1115 in search path. It also meant that it was possible to successfully
1116 load and start units which are not found in the unit search path, as
1117 long as they were requested as a dependency and linked to from
1118 .wants/ or .requires/. The target paths of those symlinks are not
1119 used for loading units anymore and the unit file must be found in
1122 * A new service type has been added: Type=exec. It's very similar to
1123 Type=simple but ensures the service manager will wait for both fork()
1124 and execve() of the main service binary to complete before proceeding
1125 with follow-up units. This is primarily useful so that the manager
1126 propagates any errors in the preparation phase of service execution
1127 back to the job that requested the unit to be started. For example,
1128 consider a service that has ExecStart= set to a file system binary
1129 that doesn't exist. With Type=simple starting the unit would be
1130 considered instantly successful, as only fork() has to complete
1131 successfully and the manager does not wait for execve(), and hence
1132 its failure is seen "too late". With the new Type=exec service type
1133 starting the unit will fail, as the manager will wait for the
1134 execve() and notice its failure, which is then propagated back to the
1137 NOTE: with the next release 241 of systemd we intend to change the
1138 systemd-run tool to default to Type=exec for transient services
1139 started by it. This should be mostly safe, but in specific corner
1140 cases might result in problems, as the systemd-run tool will then
1141 block on NSS calls (such as user name look-ups due to User=) done
1142 between the fork() and execve(), which under specific circumstances
1143 might cause problems. It is recommended to specify "-p Type=simple"
1144 explicitly in the few cases where this applies. For regular,
1145 non-transient services (i.e. those defined with unit files on disk)
1146 we will continue to default to Type=simple.
1148 * The Linux kernel's current default RLIMIT_NOFILE resource limit for
1149 userspace processes is set to 1024 (soft) and 4096
1150 (hard). Previously, systemd passed this on unmodified to all
1151 processes it forked off. With this systemd release the hard limit
1152 systemd passes on is increased to 512K, overriding the kernel's
1153 defaults and substantially increasing the number of simultaneous file
1154 descriptors unprivileged userspace processes can allocate. Note that
1155 the soft limit remains at 1024 for compatibility reasons: the
1156 traditional UNIX select() call cannot deal with file descriptors >=
1157 1024 and increasing the soft limit globally might thus result in
1158 programs unexpectedly allocating a high file descriptor and thus
1159 failing abnormally when attempting to use it with select() (of
1160 course, programs shouldn't use select() anymore, and prefer
1161 poll()/epoll, but the call unfortunately remains undeservedly popular
1162 at this time). This change reflects the fact that file descriptor
1163 handling in the Linux kernel has been optimized in more recent
1164 kernels and allocating large numbers of them should be much cheaper
1165 both in memory and in performance than it used to be. Programs that
1166 want to take benefit of the increased limit have to "opt-in" into
1167 high file descriptors explicitly by raising their soft limit. Of
1168 course, when they do that they must acknowledge that they cannot use
1169 select() anymore (and neither can any shared library they use — or
1170 any shared library used by any shared library they use and so on).
1171 Which default hard limit is most appropriate is of course hard to
1172 decide. However, given reports that ~300K file descriptors are used
1173 in real-life applications we believe 512K is sufficiently high as new
1174 default for now. Note that there are also reports that using very
1175 high hard limits (e.g. 1G) is problematic: some software allocates
1176 large arrays with one element for each potential file descriptor
1177 (Java, …) — a high hard limit thus triggers excessively large memory
1178 allocations in these applications. Hopefully, the new default of 512K
1179 is a good middle ground: higher than what real-life applications
1180 currently need, and low enough for avoid triggering excessively large
1181 allocations in problematic software. (And yes, somebody should fix
1184 * The fs.nr_open and fs.file-max sysctls are now automatically bumped
1185 to the highest possible values, as separate accounting of file
1186 descriptors is no longer necessary, as memcg tracks them correctly as
1187 part of the memory accounting anyway. Thus, from the four limits on
1188 file descriptors currently enforced (fs.file-max, fs.nr_open,
1189 RLIMIT_NOFILE hard, RLIMIT_NOFILE soft) we turn off the first two,
1190 and keep only the latter two. A set of build-time options
1191 (-Dbump-proc-sys-fs-file-max=false and -Dbump-proc-sys-fs-nr-open=false)
1192 has been added to revert this change in behaviour, which might be
1193 an option for systems that turn off memcg in the kernel.
1195 * When no /etc/locale.conf file exists (and hence no locale settings
1196 are in place), systemd will now use the "C.UTF-8" locale by default,
1197 and set LANG= to it. This locale is supported by various
1198 distributions including Fedora, with clear indications that upstream
1199 glibc is going to make it available too. This locale enables UTF-8
1200 mode by default, which appears appropriate for 2018.
1202 * The "net.ipv4.conf.all.rp_filter" sysctl will now be set to 2 by
1203 default. This effectively switches the RFC3704 Reverse Path filtering
1204 from Strict mode to Loose mode. This is more appropriate for hosts
1205 that have multiple links with routes to the same networks (e.g.
1206 a client with a Wi-Fi and Ethernet both connected to the internet).
1208 Consult the kernel documentation for details on this sysctl:
1209 https://www.kernel.org/doc/Documentation/networking/ip-sysctl.txt
1211 * CPUAccounting=yes no longer enables the CPU controller when using
1212 kernel 4.15+ and the unified cgroup hierarchy, as required accounting
1213 statistics are now provided independently from the CPU controller.
1215 * Support for disabling a particular cgroup controller within a sub-tree
1216 has been added through the DisableControllers= directive.
1218 * cgroup_no_v1=all on the kernel command line now also implies
1219 using the unified cgroup hierarchy, unless one explicitly passes
1220 systemd.unified_cgroup_hierarchy=0 on the kernel command line.
1222 * The new "MemoryMin=" unit file property may now be used to set the
1223 memory usage protection limit of processes invoked by the unit. This
1224 controls the cgroup v2 memory.min attribute. Similarly, the new
1225 "IODeviceLatencyTargetSec=" property has been added, wrapping the new
1226 cgroup v2 io.latency cgroup property for configuring per-service I/O
1229 * systemd now supports the cgroup v2 devices BPF logic, as counterpart
1230 to the cgroup v1 "devices" cgroup controller.
1232 * systemd-escape now is able to combine --unescape with --template. It
1233 also learnt a new option --instance for extracting and unescaping the
1234 instance part of a unit name.
1236 * sd-bus now provides the sd_bus_message_readv() which is similar to
1237 sd_bus_message_read() but takes a va_list object. The pair
1238 sd_bus_set_method_call_timeout() and sd_bus_get_method_call_timeout()
1239 has been added for configuring the default method call timeout to
1240 use. sd_bus_error_move() may be used to efficiently move the contents
1241 from one sd_bus_error structure to another, invalidating the
1242 source. sd_bus_set_close_on_exit() and sd_bus_get_close_on_exit() may
1243 be used to control whether a bus connection object is automatically
1244 flushed when an sd-event loop is exited.
1246 * When processing classic BSD syslog log messages, journald will now
1247 save the original time-stamp string supplied in the new
1248 SYSLOG_TIMESTAMP= journal field. This permits consumers to
1249 reconstruct the original BSD syslog message more correctly.
1251 * StandardOutput=/StandardError= in service files gained support for
1252 new "append:…" parameters, for connecting STDOUT/STDERR of a service
1253 to a file, and appending to it.
1255 * The signal to use as last step of killing of unit processes is now
1256 configurable. Previously it was hard-coded to SIGKILL, which may now
1257 be overridden with the new KillSignal= setting. Note that this is the
1258 signal used when regular termination (i.e. SIGTERM) does not suffice.
1259 Similarly, the signal used when aborting a program in case of a
1260 watchdog timeout may now be configured too (WatchdogSignal=).
1262 * The XDG_SESSION_DESKTOP environment variable may now be configured in
1263 the pam_systemd argument line, using the new desktop= switch. This is
1264 useful to initialize it properly from a display manager without
1265 having to touch C code.
1267 * Most configuration options that previously accepted percentage values
1268 now also accept permille values with the '‰' suffix (instead of '%').
1270 * systemd-resolved may now optionally use OpenSSL instead of GnuTLS for
1273 * systemd-resolved's configuration file resolved.conf gained a new
1274 option ReadEtcHosts= which may be used to turn off processing and
1275 honoring /etc/hosts entries.
1277 * The "--wait" switch may now be passed to "systemctl
1278 is-system-running", in which case the tool will synchronously wait
1279 until the system finished start-up.
1281 * hostnamed gained a new bus call to determine the DMI product UUID.
1283 * On x86-64 systemd will now prefer using the RDRAND processor
1284 instruction over /dev/urandom whenever it requires randomness that
1285 neither has to be crypto-grade nor should be reproducible. This
1286 should substantially reduce the amount of entropy systemd requests
1287 from the kernel during initialization on such systems, though not
1288 reduce it to zero. (Why not zero? systemd still needs to allocate
1289 UUIDs and such uniquely, which require high-quality randomness.)
1291 * networkd gained support for Foo-Over-UDP, ERSPAN and ISATAP
1292 tunnels. It also gained a new option ForceDHCPv6PDOtherInformation=
1293 for forcing the "Other Information" bit in IPv6 RA messages. The
1294 bonding logic gained four new options AdActorSystemPriority=,
1295 AdUserPortKey=, AdActorSystem= for configuring various 802.3ad
1296 aspects, and DynamicTransmitLoadBalancing= for enabling dynamic
1297 shuffling of flows. The tunnel logic gained a new
1298 IPv6RapidDeploymentPrefix= option for configuring IPv6 Rapid
1299 Deployment. The policy rule logic gained four new options IPProtocol=,
1300 SourcePort= and DestinationPort=, InvertRule=. The bridge logic gained
1301 support for the MulticastToUnicast= option. networkd also gained
1302 support for configuring static IPv4 ARP or IPv6 neighbor entries.
1304 * .preset files (as read by 'systemctl preset') may now be used to
1305 instantiate services.
1307 * /etc/crypttab now understands the sector-size= option to configure
1308 the sector size for an encrypted partition.
1310 * Key material for encrypted disks may now be placed on a formatted
1311 medium, and referenced from /etc/crypttab by the UUID of the file
1312 system, followed by "=" suffixed by the path to the key file.
1314 * The "collect" udev component has been removed without replacement, as
1315 it is neither used nor maintained.
1317 * When the RuntimeDirectory=, StateDirectory=, CacheDirectory=,
1318 LogsDirectory=, ConfigurationDirectory= settings are used in a
1319 service the executed processes will now receive a set of environment
1320 variables containing the full paths of these directories.
1321 Specifically, RUNTIME_DIRECTORY=, STATE_DIRECTORY, CACHE_DIRECTORY,
1322 LOGS_DIRECTORY, CONFIGURATION_DIRECTORY are now set if these options
1323 are used. Note that these options may be used multiple times per
1324 service in which case the resulting paths will be concatenated and
1325 separated by colons.
1327 * Predictable interface naming has been extended to cover InfiniBand
1328 NICs. They will be exposed with an "ib" prefix.
1330 * tmpfiles.d/ line types may now be suffixed with a '-' character, in
1331 which case the respective line failing is ignored.
1333 * .link files may now be used to configure the equivalent to the
1334 "ethtool advertise" commands.
1336 * The sd-device.h and sd-hwdb.h APIs are now exported, as an
1337 alternative to libudev.h. Previously, the latter was just an internal
1338 wrapper around the former, but now these two APIs are exposed
1341 * sd-id128.h gained a new function sd_id128_get_boot_app_specific()
1342 which calculates an app-specific boot ID similar to how
1343 sd_id128_get_machine_app_specific() generates an app-specific machine
1346 * A new tool systemd-id128 has been added that can be used to determine
1347 and generate various 128bit IDs.
1349 * /etc/os-release gained two new standardized fields DOCUMENTATION_URL=
1352 * systemd-hibernate-resume-generator will now honor the "noresume"
1353 kernel command line option, in which case it will bypass resuming
1354 from any hibernated image.
1356 * The systemd-sleep.conf configuration file gained new options
1357 AllowSuspend=, AllowHibernation=, AllowSuspendThenHibernate=,
1358 AllowHybridSleep= for prohibiting specific sleep modes even if the
1359 kernel exports them.
1361 * portablectl is now officially supported and has thus moved to
1364 * bootctl learnt the two new commands "set-default" and "set-oneshot"
1365 for setting the default boot loader item to boot to (either
1366 persistently or only for the next boot). This is currently only
1367 compatible with sd-boot, but may be implemented on other boot loaders
1368 too, that follow the boot loader interface. The updated interface is
1369 now documented here:
1371 https://systemd.io/BOOT_LOADER_INTERFACE
1373 * A new kernel command line option systemd.early_core_pattern= is now
1374 understood which may be used to influence the core_pattern PID 1
1375 installs during early boot.
1377 * busctl learnt two new options -j and --json= for outputting method
1378 call replies, properties and monitoring output in JSON.
1380 * journalctl's JSON output now supports simple ANSI coloring as well as
1381 a new "json-seq" mode for generating RFC7464 output.
1383 * Unit files now support the %g/%G specifiers that resolve to the UNIX
1384 group/GID of the service manager runs as, similar to the existing
1385 %u/%U specifiers that resolve to the UNIX user/UID.
1387 * systemd-logind learnt a new global configuration option
1388 UserStopDelaySec= that may be set in logind.conf. It specifies how
1389 long the systemd --user instance shall remain started after a user
1390 logs out. This is useful to speed up repetitive re-connections of the
1391 same user, as it means the user's service manager doesn't have to be
1392 stopped/restarted on each iteration, but can be reused between
1393 subsequent options. This setting defaults to 10s. systemd-logind also
1394 exports two new properties on its Manager D-Bus objects indicating
1395 whether the system's lid is currently closed, and whether the system
1398 * systemd gained support for a generic boot counting logic, which
1399 generically permits automatic reverting to older boot loader entries
1400 if newer updated ones don't work. The boot loader side is implemented
1401 in sd-boot, but is kept open for other boot loaders too. For details
1404 https://systemd.io/AUTOMATIC_BOOT_ASSESSMENT
1406 * The SuccessAction=/FailureAction= unit file settings now learnt two
1407 new parameters: "exit" and "exit-force", which result in immediate
1408 exiting of the service manager, and are only useful in systemd --user
1409 and container environments.
1411 * Unit files gained support for a pair of options
1412 FailureActionExitStatus=/SuccessActionExitStatus= for configuring the
1413 exit status to use as service manager exit status when
1414 SuccessAction=/FailureAction= is set to exit or exit-force.
1416 * A pair of LogRateLimitIntervalSec=/LogRateLimitBurst= per-service
1417 options may now be used to configure the log rate limiting applied by
1418 journald per-service.
1420 * systemd-analyze gained a new verb "timespan" for parsing and
1421 normalizing time span values (i.e. strings like "5min 7s 8us").
1423 * systemd-analyze also gained a new verb "security" for analyzing the
1424 security and sand-boxing settings of services in order to determine an
1425 "exposure level" for them, indicating whether a service would benefit
1426 from more sand-boxing options turned on for them.
1428 * "systemd-analyze syscall-filter" will now also show system calls
1429 supported by the local kernel but not included in any of the defined
1432 * .nspawn files now understand the Ephemeral= setting, matching the
1433 --ephemeral command line switch.
1435 * sd-event gained the new APIs sd_event_source_get_floating() and
1436 sd_event_source_set_floating() for controlling whether a specific
1437 event source is "floating", i.e. destroyed along with the even loop
1440 * Unit objects on D-Bus gained a new "Refs" property that lists all
1441 clients that currently have a reference on the unit (to ensure it is
1444 * The JoinControllers= option in system.conf is no longer supported, as
1445 it didn't work correctly, is hard to support properly, is legacy (as
1446 the concept only exists on cgroup v1) and apparently wasn't used.
1448 * Journal messages that are generated whenever a unit enters the failed
1449 state are now tagged with a unique MESSAGE_ID. Similarly, messages
1450 generated whenever a service process exits are now made recognizable,
1451 too. A tagged message is also emitted whenever a unit enters the
1452 "dead" state on success.
1454 * systemd-run gained a new switch --working-directory= for configuring
1455 the working directory of the service to start. A shortcut -d is
1456 equivalent, setting the working directory of the service to the
1457 current working directory of the invoking program. The new --shell
1458 (or just -S) option has been added for invoking the $SHELL of the
1459 caller as a service, and implies --pty --same-dir --wait --collect
1460 --service-type=exec. Or in other words, "systemd-run -S" is now the
1461 quickest way to quickly get an interactive in a fully clean and
1462 well-defined system service context.
1464 * machinectl gained a new verb "import-fs" for importing an OS tree
1465 from a directory. Moreover, when a directory or tarball is imported
1466 and single top-level directory found with the OS itself below the OS
1467 tree is automatically mangled and moved one level up.
1469 * systemd-importd will no longer set up an implicit btrfs loop-back
1470 file system on /var/lib/machines. If one is already set up, it will
1471 continue to be used.
1473 * A new generator "systemd-run-generator" has been added. It will
1474 synthesize a unit from one or more program command lines included in
1475 the kernel command line. This is very useful in container managers
1478 # systemd-nspawn -i someimage.raw -b systemd.run='"some command line"'
1480 This will run "systemd-nspawn" on an image, invoke the specified
1481 command line and immediately shut down the container again, returning
1482 the command line's exit code.
1484 * The block device locking logic is now documented:
1486 https://systemd.io/BLOCK_DEVICE_LOCKING
1488 * loginctl and machinectl now optionally output the various tables in
1489 JSON using the --output= switch. It is our intention to add similar
1490 support to systemctl and all other commands.
1492 * udevadm's query and trigger verb now optionally take a .device unit
1495 * systemd-udevd's network naming logic now understands a new
1496 net.naming-scheme= kernel command line switch, which may be used to
1497 pick a specific version of the naming scheme. This helps stabilizing
1498 interface names even as systemd/udev are updated and the naming logic
1501 * sd-id128.h learnt two new auxiliary helpers: sd_id128_is_allf() and
1502 SD_ID128_ALLF to test if a 128bit ID is set to all 0xFF bytes, and to
1503 initialize one to all 0xFF.
1505 * After loading the SELinux policy systemd will now recursively relabel
1506 all files and directories listed in
1507 /run/systemd/relabel-extra.d/*.relabel (which should be simple
1508 newline separated lists of paths) in addition to the ones it already
1509 implicitly relabels in /run, /dev and /sys. After the relabelling is
1510 completed the *.relabel files (and /run/systemd/relabel-extra.d/) are
1511 removed. This is useful to permit initrds (i.e. code running before
1512 the SELinux policy is in effect) to generate files in the host
1513 filesystem safely and ensure that the correct label is applied during
1514 the transition to the host OS.
1516 * KERNEL API BREAKAGE: Linux kernel 4.18 changed behaviour regarding
1517 mknod() handling in user namespaces. Previously mknod() would always
1518 fail with EPERM in user namespaces. Since 4.18 mknod() will succeed
1519 but device nodes generated that way cannot be opened, and attempts to
1520 open them result in EPERM. This breaks the "graceful fallback" logic
1521 in systemd's PrivateDevices= sand-boxing option. This option is
1522 implemented defensively, so that when systemd detects it runs in a
1523 restricted environment (such as a user namespace, or an environment
1524 where mknod() is blocked through seccomp or absence of CAP_SYS_MKNOD)
1525 where device nodes cannot be created the effect of PrivateDevices= is
1526 bypassed (following the logic that 2nd-level sand-boxing is not
1527 essential if the system systemd runs in is itself already sand-boxed
1528 as a whole). This logic breaks with 4.18 in container managers where
1529 user namespacing is used: suddenly PrivateDevices= succeeds setting
1530 up a private /dev/ file system containing devices nodes — but when
1531 these are opened they don't work.
1533 At this point it is recommended that container managers utilizing
1534 user namespaces that intend to run systemd in the payload explicitly
1535 block mknod() with seccomp or similar, so that the graceful fallback
1538 We are very sorry for the breakage and the requirement to change
1539 container configurations for newer kernels. It's purely caused by an
1540 incompatible kernel change. The relevant kernel developers have been
1541 notified about this userspace breakage quickly, but they chose to
1544 * PermissionsStartOnly= setting is deprecated (but is still supported
1545 for backwards compatibility). The same functionality is provided by
1546 the more flexible "+", "!", and "!!" prefixes to ExecStart= and other
1549 * $DBUS_SESSION_BUS_ADDRESS environment variable is not set by
1550 pam_systemd anymore.
1552 * The naming scheme for network devices was changed to always rename
1553 devices, even if they were already renamed by userspace. The "kernel"
1554 policy was changed to only apply as a fallback, if no other naming
1557 * The requirements to build systemd is bumped to meson-0.46 and
1560 Contributions from: afg, Alan Jenkins, Aleksei Timofeyev, Alexander
1561 Filippov, Alexander Kurtz, Alexey Bogdanenko, Andreas Henriksson,
1562 Andrew Jorgensen, Anita Zhang, apnix-uk, Arkan49, Arseny Maslennikov,
1563 asavah, Asbjørn Apeland, aszlig, Bastien Nocera, Ben Boeckel, Benedikt
1564 Morbach, Benjamin Berg, Bruce Zhang, Carlo Caione, Cedric Viou, Chen
1565 Qi, Chris Chiu, Chris Down, Chris Morin, Christian Rebischke, Claudius
1566 Ellsel, Colin Guthrie, dana, Daniel, Daniele Medri, Daniel Kahn
1567 Gillmor, Daniel Rusek, Daniel van Vugt, Dariusz Gadomski, Dave Reisner,
1568 David Anderson, Davide Cavalca, David Leeds, David Malcolm, David
1569 Strauss, David Tardon, Dimitri John Ledkov, Dmitry Torokhov, dj-kaktus,
1570 Dongsu Park, Elias Probst, Emil Soleyman, Erik Kooistra, Ervin Peters,
1571 Evgeni Golov, Evgeny Vereshchagin, Fabrice Fontaine, Faheel Ahmad,
1572 Faizal Luthfi, Felix Yan, Filipe Brandenburger, Franck Bui, Frank
1573 Schaefer, Frantisek Sumsal, Gautier Husson, Gianluca Boiano, Giuseppe
1574 Scrivano, glitsj16, Hans de Goede, Harald Hoyer, Harry Mallon, Harshit
1575 Jain, Helmut Grohne, Henry Tung, Hui Yiqun, imayoda, Insun Pyo, Iwan
1576 Timmer, Jan Janssen, Jan Pokorný, Jan Synacek, Jason A. Donenfeld,
1577 javitoom, Jérémy Nouhaud, Jeremy Su, Jiuyang Liu, João Paulo Rechi
1578 Vita, Joe Hershberger, Joe Rayhawk, Joerg Behrmann, Joerg Steffens,
1579 Jonas Dorel, Jon Ringle, Josh Soref, Julian Andres Klode, Jun Bo Bi,
1580 Jürg Billeter, Keith Busch, Khem Raj, Kirill Marinushkin, Larry
1581 Bernstone, Lennart Poettering, Lion Yang, Li Song, Lorenz
1582 Hübschle-Schneider, Lubomir Rintel, Lucas Werkmeister, Ludwin Janvier,
1583 Lukáš Nykrýn, Luke Shumaker, mal, Marc-Antoine Perennou, Marcin
1584 Skarbek, Marco Trevisan (Treviño), Marian Cepok, Mario Hros, Marko
1585 Myllynen, Markus Grimm, Martin Pitt, Martin Sobotka, Martin Wilck,
1586 Mathieu Trudel-Lapierre, Matthew Leeds, Michael Biebl, Michael Olbrich,
1587 Michael 'pbone' Pobega, Michael Scherer, Michal Koutný, Michal
1588 Sekletar, Michal Soltys, Mike Gilbert, Mike Palmer, Muhammet Kara, Neal
1589 Gompa, Neil Brown, Network Silence, Niklas Tibbling, Nikolas Nyby,
1590 Nogisaka Sadata, Oliver Smith, Patrik Flykt, Pavel Hrdina, Paweł
1591 Szewczyk, Peter Hutterer, Piotr Drąg, Ray Strode, Reinhold Mueller,
1592 Renaud Métrich, Roman Gushchin, Ronny Chevalier, Rubén Suárez Alvarez,
1593 Ruixin Bao, RussianNeuroMancer, Ryutaroh Matsumoto, Saleem Rashid, Sam
1594 Morris, Samuel Morris, Sandy Carter, scootergrisen, Sébastien Bacher,
1595 Sergey Ptashnick, Shawn Landden, Shengyao Xue, Shih-Yuan Lee
1596 (FourDollars), Silvio Knizek, Sjoerd Simons, Stasiek Michalski, Stephen
1597 Gallagher, Steven Allen, Steve Ramage, Susant Sahani, Sven Joachim,
1598 Sylvain Plantefève, Tanu Kaskinen, Tejun Heo, Thiago Macieira, Thomas
1599 Blume, Thomas Haller, Thomas H. P. Andersen, Tim Ruffing, TJ, Tobias
1600 Jungel, Todd Walton, Tommi Rantala, Tomsod M, Tony Novak, Tore
1601 Anderson, Trevonn, Victor Laskurain, Victor Tapia, Violet Halo, Vojtech
1602 Trefny, welaq, William A. Kennington III, William Douglas, Wyatt Ward,
1603 Xiang Fan, Xi Ruoyao, Xuanwo, Yann E. Morin, YmrDtnJu, Yu Watanabe,
1604 Zbigniew Jędrzejewski-Szmek, Zhang Xianwei, Zsolt Dollenstein
1606 — Warsaw, 2018-12-21
1610 * NETWORK INTERFACE DEVICE NAMING CHANGES: systemd-udevd's "net_id"
1611 builtin will name network interfaces differently than in previous
1612 versions for virtual network interfaces created with SR-IOV and NPAR
1613 and for devices where the PCI network controller device does not have
1614 a slot number associated.
1616 SR-IOV virtual devices are now named based on the name of the parent
1617 interface, with a suffix of "v<N>", where <N> is the virtual device
1618 number. Previously those virtual devices were named as if completely
1621 The ninth and later NPAR virtual devices will be named following the
1622 scheme used for the first eight NPAR partitions. Previously those
1623 devices were not renamed and the kernel default (eth<n>) was used.
1625 "net_id" will also generate names for PCI devices where the PCI
1626 network controller device does not have an associated slot number
1627 itself, but one of its parents does. Previously those devices were
1628 not renamed and the kernel default (eth<n>) was used.
1630 * AF_INET and AF_INET6 are dropped from RestrictAddressFamilies= in
1631 systemd-logind.service. Since v235, IPAddressDeny=any has been set to
1632 the unit. So, it is expected that the default behavior of
1633 systemd-logind is not changed. However, if distribution packagers or
1634 administrators disabled or modified IPAddressDeny= setting by a
1635 drop-in config file, then it may be necessary to update the file to
1636 re-enable AF_INET and AF_INET6 to support network user name services,
1639 * When the RestrictNamespaces= unit property is specified multiple
1640 times, then the specified types are merged now. Previously, only the
1641 last assignment was used. So, if distribution packagers or
1642 administrators modified the setting by a drop-in config file, then it
1643 may be necessary to update the file.
1645 * When OnFailure= is used in combination with Restart= on a service
1646 unit, then the specified units will no longer be triggered on
1647 failures that result in restarting. Previously, the specified units
1648 would be activated each time the unit failed, even when the unit was
1649 going to be restarted automatically. This behaviour contradicted the
1650 documentation. With this release the code is adjusted to match the
1653 * systemd-tmpfiles will now print a notice whenever it encounters
1654 tmpfiles.d/ lines referencing the /var/run/ directory. It will
1655 recommend reworking them to use the /run/ directory instead (for
1656 which /var/run/ is simply a symlinked compatibility alias). This way
1657 systemd-tmpfiles can properly detect line conflicts and merge lines
1658 referencing the same file by two paths, without having to access
1661 * systemctl disable/unmask/preset/preset-all cannot be used with
1662 --runtime. Previously this was allowed, but resulted in unintuitive
1663 behaviour that wasn't useful. systemctl disable/unmask will now undo
1664 both runtime and persistent enablement/masking, i.e. it will remove
1665 any relevant symlinks both in /run and /etc.
1667 * Note that all long-running system services shipped with systemd will
1668 now default to a system call whitelist (rather than a blacklist, as
1669 before). In particular, systemd-udevd will now enforce one too. For
1670 most cases this should be safe, however downstream distributions
1671 which disabled sandboxing of systemd-udevd (specifically the
1672 MountFlags= setting), might want to disable this security feature
1673 too, as the default whitelisting will prohibit all mount, swap,
1674 reboot and clock changing operations from udev rules.
1676 * sd-boot acquired new loader configuration settings to optionally turn
1677 off Windows and MacOS boot partition discovery as well as
1678 reboot-into-firmware menu items. It is also able to pick a better
1679 screen resolution for HiDPI systems, and now provides loader
1680 configuration settings to change the resolution explicitly.
1682 * systemd-resolved now supports DNS-over-TLS. It's still
1683 turned off by default, use DNSOverTLS=opportunistic to turn it on in
1684 resolved.conf. We intend to make this the default as soon as couple
1685 of additional techniques for optimizing the initial latency caused by
1686 establishing a TLS/TCP connection are implemented.
1688 * systemd-resolved.service and systemd-networkd.service now set
1689 DynamicUser=yes. The users systemd-resolve and systemd-network are
1690 not created by systemd-sysusers anymore.
1692 NOTE: This has a chance of breaking nss-ldap and similar NSS modules
1693 that embed a network facing module into any process using getpwuid()
1694 or related call: the dynamic allocation of the user ID for
1695 systemd-resolved.service means the service manager has to check NSS
1696 if the user name is already taken when forking off the service. Since
1697 the user in the common case won't be defined in /etc/passwd the
1698 lookup is likely to trigger nss-ldap which in turn might use NSS to
1699 ask systemd-resolved for hostname lookups. This will hence result in
1700 a deadlock: a user name lookup in order to start
1701 systemd-resolved.service will result in a host name lookup for which
1702 systemd-resolved.service needs to be started already. There are
1703 multiple ways to work around this problem: pre-allocate the
1704 "systemd-resolve" user on such systems, so that nss-ldap won't be
1705 triggered; or use a different NSS package that doesn't do networking
1706 in-process but provides a local asynchronous name cache; or configure
1707 the NSS package to avoid lookups for UIDs in the range `pkg-config
1708 systemd --variable=dynamicuidmin` … `pkg-config systemd
1709 --variable=dynamicuidmax`, so that it does not consider itself
1710 authoritative for the same UID range systemd allocates dynamic users
1713 * The systemd-resolve tool has been renamed to resolvectl (it also
1714 remains available under the old name, for compatibility), and its
1715 interface is now verb-based, similar in style to the other <xyz>ctl
1716 tools, such as systemctl or loginctl.
1718 * The resolvectl/systemd-resolve tool also provides 'resolvconf'
1719 compatibility. It may be symlinked under the 'resolvconf' name, in
1720 which case it will take arguments and input compatible with the
1721 Debian and FreeBSD resolvconf tool.
1723 * Support for suspend-then-hibernate has been added, i.e. a sleep mode
1724 where the system initially suspends, and after a timeout resumes and
1727 * networkd's ClientIdentifier= now accepts a new option "duid-only". If
1728 set the client will only send a DUID as client identifier.
1730 * The nss-systemd glibc NSS module will now enumerate dynamic users and
1731 groups in effect. Previously, it could resolve UIDs/GIDs to user
1732 names/groups and vice versa, but did not support enumeration.
1734 * journald's Compress= configuration setting now optionally accepts a
1735 byte threshold value. All journal objects larger than this threshold
1736 will be compressed, smaller ones will not. Previously this threshold
1737 was not configurable and set to 512.
1739 * A new system.conf setting NoNewPrivileges= is now available which may
1740 be used to turn off acquisition of new privileges system-wide
1741 (i.e. set Linux' PR_SET_NO_NEW_PRIVS for PID 1 itself, and thus also
1742 for all its children). Note that turning this option on means setuid
1743 binaries and file system capabilities lose their special powers.
1744 While turning on this option is a big step towards a more secure
1745 system, doing so is likely to break numerous pre-existing UNIX tools,
1746 in particular su and sudo.
1748 * A new service systemd-time-sync-wait.service has been added. If
1749 enabled it will delay the time-sync.target unit at boot until time
1750 synchronization has been received from the network. This
1751 functionality is useful on systems lacking a local RTC or where it is
1752 acceptable that the boot process shall be delayed by external network
1755 * When hibernating, systemd will now inform the kernel of the image
1756 write offset, on kernels new enough to support this. This means swap
1757 files should work for hibernation now.
1759 * When loading unit files, systemd will now look for drop-in unit files
1760 extensions in additional places. Previously, for a unit file name
1761 "foo-bar-baz.service" it would look for dropin files in
1762 "foo-bar-baz.service.d/*.conf". Now, it will also look in
1763 "foo-bar-.service.d/*.conf" and "foo-.service.d/", i.e. at the
1764 service name truncated after all inner dashes. This scheme allows
1765 writing drop-ins easily that apply to a whole set of unit files at
1766 once. It's particularly useful for mount and slice units (as their
1767 naming is prefix based), but is also useful for service and other
1768 units, for packages that install multiple unit files at once,
1769 following a strict naming regime of beginning the unit file name with
1770 the package's name. Two new specifiers are now supported in unit
1771 files to match this: %j and %J are replaced by the part of the unit
1772 name following the last dash.
1774 * Unit files and other configuration files that support specifier
1775 expansion now understand another three new specifiers: %T and %V will
1776 resolve to /tmp and /var/tmp respectively, or whatever temporary
1777 directory has been set for the calling user. %E will expand to either
1778 /etc (for system units) or $XDG_CONFIG_HOME (for user units).
1780 * The ExecStart= lines of unit files are no longer required to
1781 reference absolute paths. If non-absolute paths are specified the
1782 specified binary name is searched within the service manager's
1783 built-in $PATH, which may be queried with 'systemd-path
1784 search-binaries-default'. It's generally recommended to continue to
1785 use absolute paths for all binaries specified in unit files.
1787 * Units gained a new load state "bad-setting", which is used when a
1788 unit file was loaded, but contained fatal errors which prevent it
1789 from being started (for example, a service unit has been defined
1790 lacking both ExecStart= and ExecStop= lines).
1792 * coredumpctl's "gdb" verb has been renamed to "debug", in order to
1793 support alternative debuggers, for example lldb. The old name
1794 continues to be available however, for compatibility reasons. Use the
1795 new --debugger= switch or the $SYSTEMD_DEBUGGER environment variable
1796 to pick an alternative debugger instead of the default gdb.
1798 * systemctl and the other tools will now output escape sequences that
1799 generate proper clickable hyperlinks in various terminal emulators
1800 where useful (for example, in the "systemctl status" output you can
1801 now click on the unit file name to quickly open it in the
1802 editor/viewer of your choice). Note that not all terminal emulators
1803 support this functionality yet, but many do. Unfortunately, the
1804 "less" pager doesn't support this yet, hence this functionality is
1805 currently automatically turned off when a pager is started (which
1806 happens quite often due to auto-paging). We hope to remove this
1807 limitation as soon as "less" learns these escape sequences. This new
1808 behaviour may also be turned off explicitly with the $SYSTEMD_URLIFY
1809 environment variable. For details on these escape sequences see:
1810 https://gist.github.com/egmontkob/eb114294efbcd5adb1944c9f3cb5feda
1812 * networkd's .network files now support a new IPv6MTUBytes= option for
1813 setting the MTU used by IPv6 explicitly as well as a new MTUBytes=
1814 option in the [Route] section to configure the MTU to use for
1815 specific routes. It also gained support for configuration of the DHCP
1816 "UserClass" option through the new UserClass= setting. It gained
1817 three new options in the new [CAN] section for configuring CAN
1818 networks. The MULTICAST and ALLMULTI interface flags may now be
1819 controlled explicitly with the new Multicast= and AllMulticast=
1822 * networkd will now automatically make use of the kernel's route
1823 expiration feature, if it is available.
1825 * udevd's .link files now support setting the number of receive and
1826 transmit channels, using the RxChannels=, TxChannels=,
1827 OtherChannels=, CombinedChannels= settings.
1829 * Support for UDPSegmentationOffload= has been removed, given its
1830 limited support in hardware, and waning software support.
1832 * networkd's .netdev files now support creating "netdevsim" interfaces.
1834 * PID 1 learnt a new bus call GetUnitByControlGroup() which may be used
1835 to query the unit belonging to a specific kernel control group.
1837 * systemd-analyze gained a new verb "cat-config", which may be used to
1838 dump the contents of any configuration file, with all its matching
1839 drop-in files added in, and honouring the usual search and masking
1840 logic applied to systemd configuration files. For example use
1841 "systemd-analyze cat-config systemd/system.conf" to get the complete
1842 system configuration file of systemd how it would be loaded by PID 1
1843 itself. Similar to this, various tools such as systemd-tmpfiles or
1844 systemd-sysusers, gained a new option "--cat-config", which does the
1845 corresponding operation for their own configuration settings. For
1846 example, "systemd-tmpfiles --cat-config" will now output the full
1847 list of tmpfiles.d/ lines in place.
1849 * timedatectl gained three new verbs: "show" shows bus properties of
1850 systemd-timedated, "timesync-status" shows the current NTP
1851 synchronization state of systemd-timesyncd, and "show-timesync"
1852 shows bus properties of systemd-timesyncd.
1854 * systemd-timesyncd gained a bus interface on which it exposes details
1857 * A new environment variable $SYSTEMD_TIMEDATED_NTP_SERVICES is now
1858 understood by systemd-timedated. It takes a colon-separated list of
1859 unit names of NTP client services. The list is used by
1860 "timedatectl set-ntp".
1862 * systemd-nspawn gained a new --rlimit= switch for setting initial
1863 resource limits for the container payload. There's a new switch
1864 --hostname= to explicitly override the container's hostname. A new
1865 --no-new-privileges= switch may be used to control the
1866 PR_SET_NO_NEW_PRIVS flag for the container payload. A new
1867 --oom-score-adjust= switch controls the OOM scoring adjustment value
1868 for the payload. The new --cpu-affinity= switch controls the CPU
1869 affinity of the container payload. The new --resolv-conf= switch
1870 allows more detailed control of /etc/resolv.conf handling of the
1871 container. Similarly, the new --timezone= switch allows more detailed
1872 control of /etc/localtime handling of the container.
1874 * systemd-detect-virt gained a new --list switch, which will print a
1875 list of all currently known VM and container environments.
1877 * Support for "Portable Services" has been added, see
1878 doc/PORTABLE_SERVICES.md for details. Currently, the support is still
1879 experimental, but this is expected to change soon. Reflecting this
1880 experimental state, the "portablectl" binary is not installed into
1881 /usr/bin yet. The binary has to be called with the full path
1882 /usr/lib/systemd/portablectl instead.
1884 * journalctl's and systemctl's -o switch now knows a new log output
1885 mode "with-unit". The output it generates is very similar to the
1886 regular "short" mode, but displays the unit name instead of the
1887 syslog tag for each log line. Also, the date is shown with timezone
1888 information. This mode is probably more useful than the classic
1889 "short" output mode for most purposes, except where pixel-perfect
1890 compatibility with classic /var/log/messages formatting is required.
1892 * A new --dump-bus-properties switch has been added to the systemd
1893 binary, which may be used to dump all supported D-Bus properties.
1894 (Options which are still supported, but are deprecated, are *not*
1897 * sd-bus gained a set of new calls:
1898 sd_bus_slot_set_floating()/sd_bus_slot_get_floating() may be used to
1899 enable/disable the "floating" state of a bus slot object,
1900 i.e. whether the slot object pins the bus it is allocated for into
1901 memory or if the bus slot object gets disconnected when the bus goes
1902 away. sd_bus_open_with_description(),
1903 sd_bus_open_user_with_description(),
1904 sd_bus_open_system_with_description() may be used to allocate bus
1905 objects and set their description string already during allocation.
1907 * sd-event gained support for watching inotify events from the event
1908 loop, in an efficient way, sharing inotify handles between multiple
1909 users. For this a new function sd_event_add_inotify() has been added.
1911 * sd-event and sd-bus gained support for calling special user-supplied
1912 destructor functions for userdata pointers associated with
1913 sd_event_source, sd_bus_slot, and sd_bus_track objects. For this new
1914 functions sd_bus_slot_set_destroy_callback,
1915 sd_bus_slot_get_destroy_callback, sd_bus_track_set_destroy_callback,
1916 sd_bus_track_get_destroy_callback,
1917 sd_event_source_set_destroy_callback,
1918 sd_event_source_get_destroy_callback have been added.
1920 * The "net.ipv4.tcp_ecn" sysctl will now be turned on by default.
1922 * PID 1 will now automatically reschedule .timer units whenever the
1923 local timezone changes. (They previously got rescheduled
1924 automatically when the system clock changed.)
1926 * New documentation has been added to document cgroups delegation,
1927 portable services and the various code quality tools we have set up:
1929 https://github.com/systemd/systemd/blob/master/docs/CGROUP_DELEGATION.md
1930 https://github.com/systemd/systemd/blob/master/docs/PORTABLE_SERVICES.md
1931 https://github.com/systemd/systemd/blob/master/docs/CODE_QUALITY.md
1933 * The Boot Loader Specification has been added to the source tree.
1935 https://github.com/systemd/systemd/blob/master/docs/BOOT_LOADER_SPECIFICATION.md
1937 While moving it into our source tree we have updated it and further
1938 changes are now accepted through the usual github PR workflow.
1940 * pam_systemd will now look for PAM userdata fields systemd.memory_max,
1941 systemd.tasks_max, systemd.cpu_weight, systemd.io_weight set by
1942 earlier PAM modules. The data in these fields is used to initialize
1943 the session scope's resource properties. Thus external PAM modules
1944 may now configure per-session limits, for example sourced from
1945 external user databases.
1947 * socket units with Accept=yes will now maintain a "refused" counter in
1948 addition to the existing "accepted" counter, counting connections
1949 refused due to the enforced limits.
1951 * The "systemd-path search-binaries-default" command may now be use to
1952 query the default, built-in $PATH PID 1 will pass to the services it
1955 * A new unit file setting PrivateMounts= has been added. It's a boolean
1956 option. If enabled the unit's processes are invoked in their own file
1957 system namespace. Note that this behaviour is also implied if any
1958 other file system namespacing options (such as PrivateTmp=,
1959 PrivateDevices=, ProtectSystem=, …) are used. This option is hence
1960 primarily useful for services that do not use any of the other file
1961 system namespacing options. One such service is systemd-udevd.service
1962 where this is now used by default.
1964 * ConditionSecurity= gained a new value "uefi-secureboot" that is true
1965 when the system is booted in UEFI "secure mode".
1967 * A new unit "system-update-pre.target" is added, which defines an
1968 optional synchronization point for offline system updates, as
1969 implemented by the pre-existing "system-update.target" unit. It
1970 allows ordering services before the service that executes the actual
1971 update process in a generic way.
1973 * Systemd now emits warnings whenever .include syntax is used.
1975 Contributions from: Adam Duskett, Alan Jenkins, Alessandro Casale,
1976 Alexander Kurtz, Alex Gartrell, Anssi Hannula, Arnaud Rebillout, Brian
1977 J. Murrell, Bruno Vernay, Chris Lamb, Chris Lesiak, Christian Brauner,
1978 Christian Hesse, Christian Rebischke, Colin Guthrie, Daniel Dao, Daniel
1979 Lin, Danylo Korostil, Davide Cavalca, David Tardon, Dimitri John
1980 Ledkov, Dmitriy Geels, Douglas Christman, Elia Geretto, emelenas, Emil
1981 Velikov, Evgeny Vereshchagin, Felipe Sateler, Feng Sun, Filipe
1982 Brandenburger, Franck Bui, futpib, Giuseppe Scrivano, Guillem Jover,
1983 guixxx, Hannes Reinecke, Hans de Goede, Harald Hoyer, Henrique Dante de
1984 Almeida, Hiram van Paassen, Ian Miell, Igor Gnatenko, Ivan Shapovalov,
1985 Iwan Timmer, James Cowgill, Jan Janssen, Jan Synacek, Jared Kazimir,
1986 Jérémy Rosen, João Paulo Rechi Vita, Joost Heitbrink, Jui-Chi Ricky
1987 Liang, Jürg Billeter, Kai-Heng Feng, Karol Augustin, Kay Sievers,
1988 Krzysztof Nowicki, Lauri Tirkkonen, Lennart Poettering, Leonard König,
1989 Long Li, Luca Boccassi, Lucas Werkmeister, Marcel Hoppe, Marc
1990 Kleine-Budde, Mario Limonciello, Martin Jansa, Martin Wilck, Mathieu
1991 Malaterre, Matteo F. Vescovi, Matthew McGinn, Matthias-Christian Ott,
1992 Michael Biebl, Michael Olbrich, Michael Prokop, Michal Koutný, Michal
1993 Sekletar, Mike Gilbert, Mikhail Kasimov, Milan Broz, Milan Pässler,
1994 Mladen Pejaković, Muhammet Kara, Nicolas Boichat, Omer Katz, Paride
1995 Legovini, Paul Menzel, Paul Milliken, Pavel Hrdina, Peter A. Bigot,
1996 Peter D'Hoye, Peter Hutterer, Peter Jones, Philip Sequeira, Philip
1997 Withnall, Piotr Drąg, Radostin Stoyanov, Ricardo Salveti de Araujo,
1998 Ronny Chevalier, Rosen Penev, Rubén Suárez Alvarez, Ryan Gonzalez,
1999 Salvo Tomaselli, Sebastian Reichel, Sergey Ptashnick, Sergio Lindo
2000 Mansilla, Stefan Schweter, Stephen Hemminger, Stuart Hayes, Susant
2001 Sahani, Sylvain Plantefève, Thomas H. P. Andersen, Tobias Jungel,
2002 Tomasz Torcz, Vito Caputo, Will Dietz, Will Thompson, Wim van Mourik,
2003 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
2005 — Berlin, 2018-06-22
2009 * The MemoryAccounting= unit property now defaults to on. After
2010 discussions with the upstream control group maintainers we learnt
2011 that the negative impact of cgroup memory accounting on current
2012 kernels is finally relatively minimal, so that it should be safe to
2013 enable this by default without affecting system performance. Besides
2014 memory accounting only task accounting is turned on by default, all
2015 other forms of resource accounting (CPU, IO, IP) remain off for now,
2016 because it's not clear yet that their impact is small enough to move
2017 from opt-in to opt-out. We recommend downstreams to leave memory
2018 accounting on by default if kernel 4.14 or higher is primarily
2019 used. On very resource constrained systems or when support for old
2020 kernels is a necessity, -Dmemory-accounting-default=false can be used
2021 to revert this change.
2023 * rpm scriptlets to update the udev hwdb and rules (%udev_hwdb_update,
2024 %udev_rules_update) and the journal catalog (%journal_catalog_update)
2025 from the upgrade scriptlets of individual packages now do nothing.
2026 Transfiletriggers have been added which will perform those updates
2027 once at the end of the transaction.
2029 Similar transfiletriggers have been added to execute any sysctl.d
2030 and binfmt.d rules. Thus, it should be unnecessary to provide any
2031 scriptlets to execute this configuration from package installation
2034 * systemd-sysusers gained a mode where the configuration to execute is
2035 specified on the command line, but this configuration is not executed
2036 directly, but instead it is merged with the configuration on disk,
2037 and the result is executed. This is useful for package installation
2038 scripts which want to create the user before installing any files on
2039 disk (in case some of those files are owned by that user), while
2040 still allowing local admin overrides.
2042 This functionality is exposed to rpm scriptlets through a new
2043 %sysusers_create_package macro. Old %sysusers_create and
2044 %sysusers_create_inline macros are deprecated.
2046 A transfiletrigger for sysusers.d configuration is now installed,
2047 which means that it should be unnecessary to call systemd-sysusers from
2048 package installation scripts, unless the package installs any files
2049 owned by those newly-created users, in which case
2050 %sysusers_create_package should be used.
2052 * Analogous change has been done for systemd-tmpfiles: it gained a mode
2053 where the command-line configuration is merged with the configuration
2054 on disk. This is exposed as the new %tmpfiles_create_package macro,
2055 and %tmpfiles_create is deprecated. A transfiletrigger is installed
2056 for tmpfiles.d, hence it should be unnecessary to call systemd-tmpfiles
2057 from package installation scripts.
2059 * sysusers.d configuration for a user may now also specify the group
2060 number, in addition to the user number ("u username 123:456"), or
2061 without the user number ("u username -:456").
2063 * Configution items for systemd-sysusers can now be specified as
2064 positional arguments when the new --inline switch is used.
2066 * The login shell of users created through sysusers.d may now be
2067 specified (previously, it was always /bin/sh for root and
2068 /sbin/nologin for other users).
2070 * systemd-analyze gained a new --global switch to look at global user
2071 configuration. It also gained a unit-paths verb to list the unit load
2072 paths that are compiled into systemd (which can be used with
2073 --systemd, --user, or --global).
2075 * udevadm trigger gained a new --settle/-w option to wait for any
2076 triggered events to finish (but just those, and not any other events
2077 which are triggered meanwhile).
2079 * The action that systemd-logind takes when the lid is closed and the
2080 machine is connected to external power can now be configured using
2081 HandleLidSwitchExternalPower= in logind.conf. Previously, this action
2082 was determined by HandleLidSwitch=, and, for backwards compatibility,
2083 is still is, if HandleLidSwitchExternalPower= is not explicitly set.
2085 * journalctl will periodically call sd_journal_process() to make it
2086 resilient against inotify queue overruns when journal files are
2087 rotated very quickly.
2089 * Two new functions in libsystemd — sd_bus_get_n_queued_read and
2090 sd_bus_get_n_queued_write — may be used to check the number of
2091 pending bus messages.
2093 * systemd gained a new
2094 org.freedesktop.systemd1.Manager.AttachProcessesToUnit dbus call
2095 which can be used to migrate foreign processes to scope and service
2096 units. The primary user for this new API is systemd itself: the
2097 systemd --user instance uses this call of the systemd --system
2098 instance to migrate processes if it itself gets the request to
2099 migrate processes and the kernel refuses this due to access
2100 restrictions. Thanks to this "systemd-run --scope --user …" works
2101 again in pure cgroup v2 environments when invoked from the user
2104 * A new TemporaryFileSystem= setting can be used to mask out part of
2105 the real file system tree with tmpfs mounts. This may be combined
2106 with BindPaths= and BindReadOnlyPaths= to hide files or directories
2107 not relevant to the unit, while still allowing some paths lower in
2108 the tree to be accessed.
2110 ProtectHome=tmpfs may now be used to hide user home and runtime
2111 directories from units, in a way that is mostly equivalent to
2112 "TemporaryFileSystem=/home /run/user /root".
2114 * Non-service units are now started with KeyringMode=shared by default.
2115 This means that mount and swapon and other mount tools have access
2116 to keys in the main keyring.
2118 * /sys/fs/bpf is now mounted automatically.
2120 * QNX virtualization is now detected by systemd-detect-virt and may
2121 be used in ConditionVirtualization=.
2123 * IPAccounting= may now be enabled also for slice units.
2125 * A new -Dsplit-bin= build configuration switch may be used to specify
2126 whether bin and sbin directories are merged, or if they should be
2127 included separately in $PATH and various listings of executable
2128 directories. The build configuration scripts will try to autodetect
2129 the proper values of -Dsplit-usr= and -Dsplit-bin= based on build
2130 system, but distributions are encouraged to configure this
2133 * A new -Dok-color= build configuration switch may be used to change
2134 the colour of "OK" status messages.
2136 * UPGRADE ISSUE: serialization of units using JoinsNamespaceOf= with
2137 PrivateNetwork=yes was buggy in previous versions of systemd. This
2138 means that after the upgrade and daemon-reexec, any such units must
2141 * INCOMPATIBILITY: as announced in the NEWS for 237, systemd-tmpfiles
2142 will not exclude read-only files owned by root from cleanup.
2144 Contributions from: Alan Jenkins, Alexander F Rødseth, Alexis Jeandet,
2145 Andika Triwidada, Andrei Gherzan, Ansgar Burchardt, antizealot1337,
2146 Batuhan Osman Taşkaya, Beniamino Galvani, Bill Yodlowsky, Caio Marcelo
2147 de Oliveira Filho, CuBiC, Daniele Medri, Daniel Mouritzen, Daniel
2148 Rusek, Davide Cavalca, Dimitri John Ledkov, Douglas Christman, Evgeny
2149 Vereshchagin, Faalagorn, Filipe Brandenburger, Franck Bui, futpib,
2150 Giacomo Longo, Gunnar Hjalmarsson, Hans de Goede, Hermann Gausterer,
2151 Iago López Galeiras, Jakub Filak, Jan Synacek, Jason A. Donenfeld,
2152 Javier Martinez Canillas, Jérémy Rosen, Lennart Poettering, Lucas
2153 Werkmeister, Mao Huang, Marco Gulino, Michael Biebl, Michael Vogt,
2154 MilhouseVH, Neal Gompa (ニール・ゴンパ), Oleander Reis, Olof Mogren,
2155 Patrick Uiterwijk, Peter Hutterer, Peter Portante, Piotr Drąg, Robert
2156 Antoni Buj Gelonch, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
2157 Fowler, SjonHortensius, snorreflorre, Susant Sahani, Sylvain
2158 Plantefève, Thomas Blume, Thomas Haller, Vito Caputo, Yu Watanabe,
2159 Zbigniew Jędrzejewski-Szmek, Марко М. Костић (Marko M. Kostić)
2161 — Warsaw, 2018-03-05
2165 * Some keyboards come with a zoom see-saw or rocker which until now got
2166 mapped to the Linux "zoomin/out" keys in hwdb. However, these
2167 keycodes are not recognized by any major desktop. They now produce
2168 Up/Down key events so that they can be used for scrolling.
2170 * INCOMPATIBILITY: systemd-tmpfiles' "f" lines changed behaviour
2171 slightly: previously, if an argument was specified for lines of this
2172 type (i.e. the right-most column was set) this string was appended to
2173 existing files each time systemd-tmpfiles was run. This behaviour was
2174 different from what the documentation said, and not particularly
2175 useful, as repeated systemd-tmpfiles invocations would not be
2176 idempotent and grow such files without bounds. With this release
2177 behaviour has been altered to match what the documentation says:
2178 lines of this type only have an effect if the indicated files don't
2179 exist yet, and only then the argument string is written to the file.
2181 * FUTURE INCOMPATIBILITY: In systemd v238 we intend to slightly change
2182 systemd-tmpfiles behaviour: previously, read-only files owned by root
2183 were always excluded from the file "aging" algorithm (i.e. the
2184 automatic clean-up of directories like /tmp based on
2185 atime/mtime/ctime). We intend to drop this restriction, and age files
2186 by default even when owned by root and read-only. This behaviour was
2187 inherited from older tools, but there have been requests to remove
2188 it, and it's not obvious why this restriction was made in the first
2189 place. Please speak up now, if you are aware of software that reqires
2190 this behaviour, otherwise we'll remove the restriction in v238.
2192 * A new environment variable $SYSTEMD_OFFLINE is now understood by
2193 systemctl. It takes a boolean argument. If on, systemctl assumes it
2194 operates on an "offline" OS tree, and will not attempt to talk to the
2195 service manager. Previously, this mode was implicitly enabled if a
2196 chroot() environment was detected, and this new environment variable
2197 now provides explicit control.
2199 * .path and .socket units may now be created transiently, too.
2200 Previously only service, mount, automount and timer units were
2201 supported as transient units. The systemd-run tool has been updated
2202 to expose this new functionality, you may hence use it now to bind
2203 arbitrary commands to path or socket activation on-the-fly from the
2204 command line. Moreover, almost all properties are now exposed for the
2205 unit types that already supported transient operation.
2207 * The systemd-mount command gained support for a new --owner= parameter
2208 which takes a user name, which is then resolved and included in uid=
2209 and gid= mount options string of the file system to mount.
2211 * A new unit condition ConditionControlGroupController= has been added
2212 that checks whether a specific cgroup controller is available.
2214 * Unit files, udev's .link files, and systemd-networkd's .netdev and
2215 .network files all gained support for a new condition
2216 ConditionKernelVersion= for checking against specific kernel
2219 * In systemd-networkd, the [IPVLAN] section in .netdev files gained
2220 support for configuring device flags in the Flags= setting. In the
2221 same files, the [Tunnel] section gained support for configuring
2222 AllowLocalRemote=. The [Route] section in .network files gained
2223 support for configuring InitialCongestionWindow=,
2224 InitialAdvertisedReceiveWindow= and QuickAck=. The [DHCP] section now
2225 understands RapidCommit=.
2227 * systemd-networkd's DHCPv6 support gained support for Prefix
2230 * sd-bus gained support for a new "watch-bind" feature. When this
2231 feature is enabled, an sd_bus connection may be set up to connect to
2232 an AF_UNIX socket in the file system as soon as it is created. This
2233 functionality is useful for writing early-boot services that
2234 automatically connect to the system bus as soon as it is started,
2235 without ugly time-based polling. systemd-networkd and
2236 systemd-resolved have been updated to make use of this
2237 functionality. busctl exposes this functionality in a new
2238 --watch-bind= command line switch.
2240 * sd-bus will now optionally synthesize a local "Connected" signal as
2241 soon as a D-Bus connection is set up fully. This message mirrors the
2242 already existing "Disconnected" signal which is synthesized when the
2243 connection is terminated. This signal is generally useful but
2244 particularly handy in combination with the "watch-bind" feature
2245 described above. Synthesizing of this message has to be requested
2246 explicitly through the new API call sd_bus_set_connected_signal(). In
2247 addition a new call sd_bus_is_ready() has been added that checks
2248 whether a connection is fully set up (i.e. between the "Connected" and
2249 "Disconnected" signals).
2251 * sd-bus gained two new calls sd_bus_request_name_async() and
2252 sd_bus_release_name_async() for asynchronously registering bus
2253 names. Similar, there is now sd_bus_add_match_async() for installing
2254 a signal match asynchronously. All of systemd's own services have
2255 been updated to make use of these calls. Doing these operations
2256 asynchronously has two benefits: it reduces the risk of deadlocks in
2257 case of cyclic dependencies between bus services, and it speeds up
2258 service initialization since synchronization points for bus
2259 round-trips are removed.
2261 * sd-bus gained two new calls sd_bus_match_signal() and
2262 sd_bus_match_signal_async(), which are similar to sd_bus_add_match()
2263 and sd_bus_add_match_async() but instead of taking a D-Bus match
2264 string take match fields as normal function parameters.
2266 * sd-bus gained two new calls sd_bus_set_sender() and
2267 sd_bus_message_set_sender() for setting the sender name of outgoing
2268 messages (either for all outgoing messages or for just one specific
2269 one). These calls are only useful in direct connections as on
2270 brokered connections the broker fills in the sender anyway,
2271 overwriting whatever the client filled in.
2273 * sd-event gained a new pseudo-handle that may be specified on all API
2274 calls where an "sd_event*" object is expected: SD_EVENT_DEFAULT. When
2275 used this refers to the default event loop object of the calling
2276 thread. Note however that this does not implicitly allocate one —
2277 which has to be done prior by using sd_event_default(). Similarly
2278 sd-bus gained three new pseudo-handles SD_BUS_DEFAULT,
2279 SD_BUS_DEFAULT_USER, SD_BUS_DEFAULT_SYSTEM that may be used to refer
2280 to the default bus of the specified type of the calling thread. Here
2281 too this does not implicitly allocate bus connection objects, this
2282 has to be done prior with sd_bus_default() and friends.
2284 * sd-event gained a new call pair
2285 sd_event_source_{get|set}_io_fd_own(). This may be used to request
2286 automatic closure of the file descriptor an IO event source watches
2287 when the event source is destroyed.
2289 * systemd-networkd gained support for natively configuring WireGuard
2292 * In previous versions systemd synthesized user records both for the
2293 "nobody" (UID 65534) and "root" (UID 0) users in nss-systemd and
2294 internally. In order to simplify distribution-wide renames of the
2295 "nobody" user (like it is planned in Fedora: nfsnobody → nobody), a
2296 new transitional flag file has been added: if
2297 /etc/systemd/dont-synthesize-nobody exists synthesizing of the 65534
2298 user and group record within the systemd codebase is disabled.
2300 * systemd-notify gained a new --uid= option for selecting the source
2301 user/UID to use for notification messages sent to the service
2304 * journalctl gained a new --grep= option to list only entries in which
2305 the message matches a certain pattern. By default matching is case
2306 insensitive if the pattern is lowercase, and case sensitive
2307 otherwise. Option --case-sensitive=yes|no can be used to override
2308 this an specify case sensitivity or case insensitivity.
2310 * There's now a "systemd-analyze service-watchdogs" command for printing
2311 the current state of the service runtime watchdog, and optionally
2312 enabling or disabling the per-service watchdogs system-wide if given a
2313 boolean argument (i.e. the concept you configure in WatchdogSec=), for
2314 debugging purposes. There's also a kernel command line option
2315 systemd.service_watchdogs= for controlling the same.
2317 * Two new "log-level" and "log-target" options for systemd-analyze were
2318 added that merge the now deprecated get-log-level, set-log-level and
2319 get-log-target, set-log-target pairs. The deprecated options are still
2320 understood for backwards compatibility. The two new options print the
2321 current value when no arguments are given, and set them when a
2322 level/target is given as an argument.
2324 * sysusers.d's "u" lines now optionally accept both a UID and a GID
2325 specification, separated by a ":" character, in order to create users
2326 where UID and GID do not match.
2328 Contributions from: Adam Duskett, Alan Jenkins, Alexander Kuleshov,
2329 Alexis Deruelle, Andrew Jeddeloh, Armin Widegreen, Batuhan Osman
2330 Taşkaya, Björn Esser, bleep_blop, Bruce A. Johnson, Chris Down, Clinton
2331 Roy, Colin Walters, Daniel Rusek, Dimitri John Ledkov, Dmitry Rozhkov,
2332 Evgeny Vereshchagin, Ewout van Mansom, Felipe Sateler, Franck Bui,
2333 Frantisek Sumsal, George Gaydarov, Gianluca Boiano, Hans-Christian
2334 Noren Egtvedt, Hans de Goede, Henrik Grindal Bakken, Jan Alexander
2335 Steffens, Jan Klötzke, Jason A. Donenfeld, jdkbx, Jérémy Rosen,
2336 Jerónimo Borque, John Lin, John Paul Herold, Jonathan Rudenberg, Jörg
2337 Thalheim, Ken (Bitsko) MacLeod, Larry Bernstone, Lennart Poettering,
2338 Lucas Werkmeister, Maciej S. Szmigiero, Marek Čermák, Martin Pitt,
2339 Mathieu Malaterre, Matthew Thode, Matthias-Christian Ott, Max Harmathy,
2340 Michael Biebl, Michael Vogt, Michal Koutný, Michal Sekletar, Michał
2341 Szczepański, Mike Gilbert, Nathaniel McCallum, Nicolas Chauvet, Olaf
2342 Hering, Olivier Schwander, Patrik Flykt, Paul Cercueil, Peter Hutterer,
2343 Piotr Drąg, Raphael Vogelgsang, Reverend Homer, Robert Kolchmeyer,
2344 Samuel Dionne-Riel, Sergey Ptashnick, Shawn Landden, Susant Sahani,
2345 Sylvain Plantefève, Thomas H. P. Andersen, Thomas Huth, Tomasz
2346 Bachorski, Vladislav Vishnyakov, Wieland Hoffmann, Yu Watanabe, Zachary
2347 Winnerman, Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски, Дилян
2354 * The modprobe.d/ drop-in for the bonding.ko kernel module introduced
2355 in v235 has been extended to also set the dummy.ko module option
2356 numdummies=0, preventing the kernel from automatically creating
2357 dummy0. All dummy interfaces must now be explicitly created.
2359 * Unknown '%' specifiers in configuration files are now rejected. This
2360 applies to units and tmpfiles.d configuration. Any percent characters
2361 that are followed by a letter or digit that are not supposed to be
2362 interpreted as the beginning of a specifier should be escaped by
2363 doubling ("%%"). (So "size=5%" is still accepted, as well as
2364 "size=5%,foo=bar", but not "LABEL=x%y%z" since %y and %z are not
2365 valid specifiers today.)
2367 * systemd-resolved now maintains a new dynamic
2368 /run/systemd/resolve/stub-resolv.conf compatibility file. It is
2369 recommended to make /etc/resolv.conf a symlink to it. This file
2370 points at the systemd-resolved stub DNS 127.0.0.53 resolver and
2371 includes dynamically acquired search domains, achieving more correct
2372 DNS resolution by software that bypasses local DNS APIs such as NSS.
2374 * The "uaccess" udev tag has been dropped from /dev/kvm and
2375 /dev/dri/renderD*. These devices now have the 0666 permissions by
2376 default (but this may be changed at build-time). /dev/dri/renderD*
2377 will now be owned by the "render" group along with /dev/kfd.
2379 * "DynamicUser=yes" has been enabled for systemd-timesyncd.service,
2380 systemd-journal-gatewayd.service and
2381 systemd-journal-upload.service. This means "nss-systemd" must be
2382 enabled in /etc/nsswitch.conf to ensure the UIDs assigned to these
2383 services are resolved properly.
2385 * In /etc/fstab two new mount options are now understood:
2386 x-systemd.makefs and x-systemd.growfs. The former has the effect that
2387 the configured file system is formatted before it is mounted, the
2388 latter that the file system is resized to the full block device size
2389 after it is mounted (i.e. if the file system is smaller than the
2390 partition it resides on, it's grown). This is similar to the fsck
2391 logic in /etc/fstab, and pulls in systemd-makefs@.service and
2392 systemd-growfs@.service as necessary, similar to
2393 systemd-fsck@.service. Resizing is currently only supported on ext4
2396 * In systemd-networkd, the IPv6 RA logic now optionally may announce
2397 DNS server and domain information.
2399 * Support for the LUKS2 on-disk format for encrypted partitions has
2400 been added. This requires libcryptsetup2 during compilation and
2403 * The systemd --user instance will now signal "readiness" when its
2404 basic.target unit has been reached, instead of when the run queue ran
2405 empty for the first time.
2407 * Tmpfiles.d with user configuration are now also supported.
2408 systemd-tmpfiles gained a new --user switch, and snippets placed in
2409 ~/.config/user-tmpfiles.d/ and corresponding directories will be
2410 executed by systemd-tmpfiles --user running in the new
2411 systemd-tmpfiles-setup.service and systemd-tmpfiles-clean.service
2412 running in the user session.
2414 * Unit files and tmpfiles.d snippets learnt three new % specifiers:
2415 %S resolves to the top-level state directory (/var/lib for the system
2416 instance, $XDG_CONFIG_HOME for the user instance), %C resolves to the
2417 top-level cache directory (/var/cache for the system instance,
2418 $XDG_CACHE_HOME for the user instance), %L resolves to the top-level
2419 logs directory (/var/log for the system instance,
2420 $XDG_CONFIG_HOME/log/ for the user instance). This matches the
2421 existing %t specifier, that resolves to the top-level runtime
2422 directory (/run for the system instance, and $XDG_RUNTIME_DIR for the
2425 * journalctl learnt a new parameter --output-fields= for limiting the
2426 set of journal fields to output in verbose and JSON output modes.
2428 * systemd-timesyncd's configuration file gained a new option
2429 RootDistanceMaxSec= for setting the maximum root distance of servers
2430 it'll use, as well as the new options PollIntervalMinSec= and
2431 PollIntervalMaxSec= to tweak the minimum and maximum poll interval.
2433 * bootctl gained a new command "list" for listing all available boot
2434 menu items on systems that follow the boot loader specification.
2436 * systemctl gained a new --dry-run switch that shows what would be done
2437 instead of doing it, and is currently supported by the shutdown and
2440 * ConditionSecurity= can now detect the TOMOYO security module.
2442 * Unit file [Install] sections are now also respected in unit drop-in
2443 files. This is intended to be used by drop-ins under /usr/lib/.
2445 * systemd-firstboot may now also set the initial keyboard mapping.
2447 * Udev "changed" events for devices which are exposed as systemd
2448 .device units are now propagated to units specified in
2449 ReloadPropagatedFrom= as reload requests.
2451 * If a udev device has a SYSTEMD_WANTS= property containing a systemd
2452 unit template name (i.e. a name in the form of 'foobar@.service',
2453 without the instance component between the '@' and - the '.'), then
2454 the escaped sysfs path of the device is automatically used as the
2457 * SystemCallFilter= in unit files has been extended so that an "errno"
2458 can be specified individually for each system call. Example:
2459 SystemCallFilter=~uname:EILSEQ.
2461 * The cgroup delegation logic has been substantially updated. Delegate=
2462 now optionally takes a list of controllers (instead of a boolean, as
2463 before), which lists the controllers to delegate at least.
2465 * The networkd DHCPv6 client now implements the FQDN option (RFC 4704).
2467 * A new LogLevelMax= setting configures the maximum log level any
2468 process of the service may log at (i.e. anything with a lesser
2469 priority than what is specified is automatically dropped). A new
2470 LogExtraFields= setting allows configuration of additional journal
2471 fields to attach to all log records generated by any of the unit's
2474 * New StandardInputData= and StandardInputText= settings along with the
2475 new option StandardInput=data may be used to configure textual or
2476 binary data that shall be passed to the executed service process via
2477 standard input, encoded in-line in the unit file.
2479 * StandardInput=, StandardOutput= and StandardError= may now be used to
2480 connect stdin/stdout/stderr of executed processes directly with a
2481 file or AF_UNIX socket in the file system, using the new "file:" option.
2483 * A new unit file option CollectMode= has been added, that allows
2484 tweaking the garbage collection logic for units. It may be used to
2485 tell systemd to garbage collect units that have failed automatically
2486 (normally it only GCs units that exited successfully). systemd-run
2487 and systemd-mount expose this new functionality with a new -G option.
2489 * "machinectl bind" may now be used to bind mount non-directories
2490 (i.e. regularfiles, devices, fifos, sockets).
2492 * systemd-analyze gained a new verb "calendar" for validating and
2493 testing calendar time specifications to use for OnCalendar= in timer
2494 units. Besides validating the expression it will calculate the next
2495 time the specified expression would elapse.
2497 * In addition to the pre-existing FailureAction= unit file setting
2498 there's now SuccessAction=, for configuring a shutdown action to
2499 execute when a unit completes successfully. This is useful in
2500 particular inside containers that shall terminate after some workload
2501 has been completed. Also, both options are now supported for all unit
2502 types, not just services.
2504 * networkds's IP rule support gained two new options
2505 IncomingInterface= and OutgoingInterface= for configuring the incoming
2506 and outgoing interfaces of configured rules. systemd-networkd also
2507 gained support for "vxcan" network devices.
2509 * networkd gained a new setting RequiredForOnline=, taking a
2510 boolean. If set, systemd-wait-online will take it into consideration
2511 when determining that the system is up, otherwise it will ignore the
2512 interface for this purpose.
2514 * The sd_notify() protocol gained support for a new operation: with
2515 FDSTOREREMOVE=1 file descriptors may be removed from the per-service
2516 store again, ahead of POLLHUP or POLLERR when they are removed
2519 * A new document doc/UIDS-GIDS.md has been added to the source tree,
2520 that documents the UID/GID range and assignment assumptions and
2521 requirements of systemd.
2523 * The watchdog device PID 1 will ping may now be configured through the
2524 WatchdogDevice= configuration file setting, or by setting the
2525 systemd.watchdog_service= kernel commandline option.
2527 * systemd-resolved's gained support for registering DNS-SD services on
2528 the local network using MulticastDNS. Services may either be
2529 registered by dropping in a .dnssd file in /etc/systemd/dnssd/ (or
2530 the same dir below /run, /usr/lib), or through its D-Bus API.
2532 * The sd_notify() protocol can now with EXTEND_TIMEOUT_USEC=microsecond
2533 extend the effective start, runtime, and stop time. The service must
2534 continue to send EXTEND_TIMEOUT_USEC within the period specified to
2535 prevent the service manager from making the service as timedout.
2537 * systemd-resolved's DNSSEC support gained support for RFC 8080
2538 (Ed25519 keys and signatures).
2540 * The systemd-resolve command line tool gained a new set of options
2541 --set-dns=, --set-domain=, --set-llmnr=, --set-mdns=, --set-dnssec=,
2542 --set-nta= and --revert to configure per-interface DNS configuration
2543 dynamically during runtime. It's useful for pushing DNS information
2544 into systemd-resolved from DNS hook scripts that various interface
2545 managing software supports (such as pppd).
2547 * systemd-nspawn gained a new --network-namespace-path= command line
2548 option, which may be used to make a container join an existing
2549 network namespace, by specifying a path to a "netns" file.
2551 Contributions from: Alan Jenkins, Alan Robertson, Alessandro Ghedini,
2552 Andrew Jeddeloh, Antonio Rojas, Ari, asavah, bleep_blop, Carsten
2553 Strotmann, Christian Brauner, Christian Hesse, Clinton Roy, Collin
2554 Eggert, Cong Wang, Daniel Black, Daniel Lockyer, Daniel Rusek, Dimitri
2555 John Ledkov, Dmitry Rozhkov, Dongsu Park, Edward A. James, Evgeny
2556 Vereshchagin, Florian Klink, Franck Bui, Gwendal Grignou, Hans de
2557 Goede, Harald Hoyer, Hristo Venev, Iago López Galeiras, Ikey Doherty,
2558 Jakub Wilk, Jérémy Rosen, Jiahui Xie, John Lin, José Bollo, Josef
2559 Andersson, juga0, Krzysztof Nowicki, Kyle Walker, Lars Karlitski, Lars
2560 Kellogg-Stedman, Lauri Tirkkonen, Lennart Poettering, Lubomir Rintel,
2561 Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn, Lukáš Říha, Lukasz
2562 Rubaszewski, Maciej S. Szmigiero, Mantas Mikulėnas, Marcus Folkesson,
2563 Martin Steuer, Mathieu Trudel-Lapierre, Matija Skala,
2564 Matthias-Christian Ott, Max Resch, Michael Biebl, Michael Vogt, Michal
2565 Koutný, Michal Sekletar, Mike Gilbert, Muhammet Kara, Neil Brown, Olaf
2566 Hering, Ondrej Kozina, Patrik Flykt, Patryk Kocielnik, Peter Hutterer,
2567 Piotr Drąg, Razvan Cojocaru, Robin McCorkell, Roland Hieber, Saran
2568 Tunyasuvunakool, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
2569 Arlott, Simon Peeters, Stanislav Angelovič, Stefan Agner, Susant
2570 Sahani, Sylvain Plantefève, Thomas Blume, Thomas Haller, Tiago Salem
2571 Herrmann, Tinu Weber, Tom Stellard, Topi Miettinen, Torsten Hilbrich,
2572 Vito Caputo, Vladislav Vishnyakov, WaLyong Cho, Yu Watanabe, Zbigniew
2573 Jędrzejewski-Szmek, Zeal Jagannatha
2575 — Berlin, 2017-12-14
2579 * INCOMPATIBILITY: systemd-logind.service and other long-running
2580 services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP
2581 communication with the outside. This generally improves security of
2582 the system, and is in almost all cases a safe and good choice, as
2583 these services do not and should not provide any network-facing
2584 functionality. However, systemd-logind uses the glibc NSS API to
2585 query the user database. This creates problems on systems where NSS
2586 is set up to directly consult network services for user database
2587 lookups. In particular, this creates incompatibilities with the
2588 "nss-nis" module, which attempts to directly contact the NIS/YP
2589 network servers it is configured for, and will now consistently
2590 fail. In such cases, it is possible to turn off IP sandboxing for
2591 systemd-logind.service (set IPAddressDeny= in its [Service] section
2592 to the empty string, via a .d/ unit file drop-in). Downstream
2593 distributions might want to update their nss-nis packaging to include
2594 such a drop-in snippet, accordingly, to hide this incompatibility
2595 from the user. Another option is to make use of glibc's nscd service
2596 to proxy such network requests through a privilege-separated, minimal
2597 local caching daemon, or to switch to more modern technologies such
2598 sssd, whose NSS hook-ups generally do not involve direct network
2599 access. In general, we think it's definitely time to question the
2600 implementation choices of nss-nis, i.e. whether it's a good idea
2601 today to embed a network-facing loadable module into all local
2602 processes that need to query the user database, including the most
2603 trivial and benign ones, such as "ls". For more details about
2604 IPAddressDeny= see below.
2606 * A new modprobe.d drop-in is now shipped by default that sets the
2607 bonding module option max_bonds=0. This overrides the kernel default,
2608 to avoid conflicts and ambiguity as to whether or not bond0 should be
2609 managed by systemd-networkd or not. This resolves multiple issues
2610 with bond0 properties not being applied, when bond0 is configured
2611 with systemd-networkd. Distributors may choose to not package this,
2612 however in that case users will be prevented from correctly managing
2613 bond0 interface using systemd-networkd.
2615 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
2616 which print the logging level and target of the system manager. They
2617 complement the existing "set-log-level" and "set-log-target" verbs
2618 used to change those values.
2620 * journald.conf gained a new boolean setting ReadKMsg= which defaults
2621 to on. If turned off kernel log messages will not be read by
2622 systemd-journald or included in the logs. It also gained a new
2623 setting LineMax= for configuring the maximum line length in
2624 STDOUT/STDERR log streams. The new default for this value is 48K, up
2625 from the previous hardcoded 2048.
2627 * A new unit setting RuntimeDirectoryPreserve= has been added, which
2628 allows more detailed control of what to do with a runtime directory
2629 configured with RuntimeDirectory= (i.e. a directory below /run or
2630 $XDG_RUNTIME_DIR) after a unit is stopped.
2632 * The RuntimeDirectory= setting for units gained support for creating
2633 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
2634 one top-level directory.
2636 * Units gained new options StateDirectory=, CacheDirectory=,
2637 LogsDirectory= and ConfigurationDirectory= which are closely related
2638 to RuntimeDirectory= but manage per-service directories below
2639 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
2640 possible to write unit files which when activated automatically gain
2641 properly owned service specific directories in these locations, thus
2642 making unit files self-contained and increasing compatibility with
2643 stateless systems and factory reset where /etc or /var are
2644 unpopulated at boot. Matching these new settings there's also
2645 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
2646 ConfigurationDirectoryMode= for configuring the access mode of these
2647 directories. These settings are particularly useful in combination
2648 with DynamicUser=yes as they provide secure, properly-owned,
2649 writable, and stateful locations for storage, excluded from the
2650 sandbox that such services live in otherwise.
2652 * Automake support has been removed from this release. systemd is now
2655 * systemd-journald will now aggressively cache client metadata during
2656 runtime, speeding up log write performance under pressure. This comes
2657 at a small price though: as much of the metadata is read
2658 asynchronously from /proc/ (and isn't implicitly attached to log
2659 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
2660 metadata stored alongside a log entry might be slightly
2661 out-of-date. Previously it could only be slightly newer than the log
2662 message. The time window is small however, and given that the kernel
2663 is unlikely to be improved anytime soon in this regard, this appears
2666 * nss-myhostname/systemd-resolved will now by default synthesize an
2667 A/AAAA resource record for the "_gateway" hostname, pointing to the
2668 current default IP gateway. Previously it did that for the "gateway"
2669 name, hampering adoption, as some distributions wanted to leave that
2670 host name open for local use. The old behaviour may still be
2671 requested at build time.
2673 * systemd-networkd's [Address] section in .network files gained a new
2674 Scope= setting for configuring the IP address scope. The [Network]
2675 section gained a new boolean setting ConfigureWithoutCarrier= that
2676 tells systemd-networkd to ignore link sensing when configuring the
2677 device. The [DHCP] section gained a new Anonymize= boolean option for
2678 turning on a number of options suggested in RFC 7844. A new
2679 [RoutingPolicyRule] section has been added for configuring the IP
2680 routing policy. The [Route] section has gained support for a new
2681 Type= setting which permits configuring
2682 blackhole/unreachable/prohibit routes.
2684 * The [VRF] section in .netdev files gained a new Table= setting for
2685 configuring the routing table to use. The [Tunnel] section gained a
2686 new Independent= boolean field for configuring tunnels independent of
2687 an underlying network interface. The [Bridge] section gained a new
2688 GroupForwardMask= option for configuration of propagation of link
2689 local frames between bridge ports.
2691 * The WakeOnLan= setting in .link files gained support for a number of
2692 new modes. A new TCP6SegmentationOffload= setting has been added for
2693 configuring TCP/IPv6 hardware segmentation offload.
2695 * The IPv6 RA sender implementation may now optionally send out RDNSS
2696 and RDNSSL records to supply DNS configuration to peers.
2698 * systemd-nspawn gained support for a new --system-call-filter= command
2699 line option for adding and removing entries in the default system
2700 call filter it applies. Moreover systemd-nspawn has been changed to
2701 implement a system call whitelist instead of a blacklist.
2703 * systemd-run gained support for a new --pipe command line option. If
2704 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
2705 are directly passed on to the activated transient service
2706 executable. This allows invoking arbitrary processes as systemd
2707 services (for example to take benefit of dependency management,
2708 accounting management, resource management or log management that is
2709 done automatically for services) — while still allowing them to be
2710 integrated in a classic UNIX shell pipeline.
2712 * When a service sends RELOAD=1 via sd_notify() and reload propagation
2713 using ReloadPropagationTo= is configured, a reload is now propagated
2714 to configured units. (Previously this was only done on explicitly
2715 requested reloads, using "systemctl reload" or an equivalent
2718 * For each service unit a restart counter is now kept: it is increased
2719 each time the service is restarted due to Restart=, and may be
2720 queried using "systemctl show -p NRestarts …".
2722 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
2723 @signal and @timer have been added, for usage with SystemCallFilter=
2724 in unit files and the new --system-call-filter= command line option
2725 of systemd-nspawn (see above).
2727 * ExecStart= lines in unit files gained two new modifiers: when a
2728 command line is prefixed with "!" the command will be executed as
2729 configured, except for the credentials applied by
2730 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
2731 "+", but does still apply namespacing options unlike "+". There's
2732 also "!!" now, which is mostly identical, but becomes a NOP on
2733 systems that support ambient capabilities. This is useful to write
2734 unit files that work with ambient capabilities where possible but
2735 automatically fall back to traditional privilege dropping mechanisms
2736 on systems where this is not supported.
2738 * ListenNetlink= settings in socket units now support RDMA netlink
2741 * A new unit file setting LockPersonality= has been added which permits
2742 locking down the chosen execution domain ("personality") of a service
2745 * A new special target "getty-pre.target" has been added, which is
2746 ordered before all text logins, and may be used to order services
2747 before textual logins acquire access to the console.
2749 * systemd will now attempt to load the virtio-rng.ko kernel module very
2750 early on if a VM environment supporting this is detected. This should
2751 improve entropy during early boot in virtualized environments.
2753 * A _netdev option is now supported in /etc/crypttab that operates in a
2754 similar way as the same option in /etc/fstab: it permits configuring
2755 encrypted devices that need to be ordered after the network is up.
2756 Following this logic, two new special targets
2757 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
2758 added that are to cryptsetup.target what remote-fs.target and
2759 remote-fs-pre.target are to local-fs.target.
2761 * Service units gained a new UnsetEnvironment= setting which permits
2762 unsetting specific environment variables for services that are
2763 normally passed to it (for example in order to mask out locale
2764 settings for specific services that can't deal with it).
2766 * Units acquired a new boolean option IPAccounting=. When turned on, IP
2767 traffic accounting (packet count as well as byte count) is done for
2768 the service, and shown as part of "systemctl status" or "systemd-run
2771 * Service units acquired two new options IPAddressAllow= and
2772 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
2773 for configuring a simple IP access control list for all sockets of
2774 the unit. These options are available also on .slice and .socket
2775 units, permitting flexible access list configuration for individual
2776 services as well as groups of services (as defined by a slice unit),
2777 including system-wide. Note that IP ACLs configured this way are
2778 enforced on every single IPv4 and IPv6 socket created by any process
2779 of the service unit, and apply to ingress as well as egress traffic.
2781 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
2782 structured log message is generated each time the unit is stopped,
2783 containing information about the consumed resources of this
2786 * A new setting KeyringMode= has been added to unit files, which may be
2787 used to control how the kernel keyring is set up for executed
2790 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
2791 "systemctl kexec" and "systemctl exit" are now always asynchronous in
2792 behaviour (that is: these commands return immediately after the
2793 operation was enqueued instead of waiting for the operation to
2794 complete). Previously, "systemctl poweroff" and "systemctl reboot"
2795 were asynchronous on systems using systemd-logind (i.e. almost
2796 always, and like they were on sysvinit), and the other three commands
2797 were unconditionally synchronous. With this release this is cleaned
2798 up, and callers will see the same asynchronous behaviour on all
2799 systems for all five operations.
2801 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
2804 * .timer units now accept calendar specifications in other timezones
2805 than UTC or the local timezone.
2807 * The tmpfiles snippet var.conf has been changed to create
2808 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
2809 the "utmp" group already, and it appears to be generally understood
2810 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
2811 databases. Previously this was implemented correctly for all these
2812 databases excepts btmp, which has been opened up like this now
2813 too. Note that while the other databases are world-readable
2814 (i.e. 0644), btmp is not and remains more restrictive.
2816 * The systemd-resolve tool gained a new --reset-server-features
2817 switch. When invoked like this systemd-resolved will forget
2818 everything it learnt about the features supported by the configured
2819 upstream DNS servers, and restarts the feature probing logic on the
2820 next resolver look-up for them at the highest feature level
2823 * The status dump systemd-resolved sends to the logs upon receiving
2824 SIGUSR1 now also includes information about all DNS servers it is
2825 configured to use, and the features levels it probed for them.
2827 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
2828 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
2829 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
2830 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
2831 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
2832 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
2833 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
2834 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
2835 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
2836 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
2837 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
2838 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
2839 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
2840 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
2841 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
2842 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
2843 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
2844 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
2845 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
2846 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
2848 — Berlin, 2017-10-06
2852 * Meson is now supported as build system in addition to Automake. It is
2853 our plan to remove Automake in one of our next releases, so that
2854 Meson becomes our exclusive build system. Hence, please start using
2855 the Meson build system in your downstream packaging. There's plenty
2856 of documentation around how to use Meson, the extremely brief
2859 ./autogen.sh && ./configure && make && sudo make install
2863 meson build && ninja -C build && sudo ninja -C build install
2865 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
2866 which permits configuring a timeout on the time a job is
2867 running. This is particularly useful for setting timeouts on jobs for
2870 * Unit files gained two new options ConditionUser= and ConditionGroup=
2871 for conditionalizing units based on the identity of the user/group
2872 running a systemd user instance.
2874 * systemd-networkd now understands a new FlowLabel= setting in the
2875 [VXLAN] section of .network files, as well as a Priority= in
2876 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
2877 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
2878 gained support for configuration of GENEVE links, and IPv6 address
2879 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
2881 * .link files now understand a new Port= setting.
2883 * systemd-networkd's DHCP support gained support for DHCP option 119
2884 (domain search list).
2886 * systemd-networkd gained support for serving IPv6 address ranges using
2887 the Router Advertisement protocol. The new .network configuration
2888 section [IPv6Prefix] may be used to configure the ranges to
2889 serve. This is implemented based on a new, minimal, native server
2890 implementation of RA.
2892 * journalctl's --output= switch gained support for a new parameter
2893 "short-iso-precise" for a mode where timestamps are shown as precise
2896 * systemd-udevd's "net_id" builtin may now generate stable network
2897 interface names from IBM PowerVM VIO devices as well as ACPI platform
2900 * MulticastDNS support in systemd-resolved may now be explicitly
2901 enabled/disabled using the new MulticastDNS= configuration file
2904 * systemd-resolved may now optionally use libidn2 instead of the libidn
2905 for processing internationalized domain names. Support for libidn2
2906 should be considered experimental and should not be enabled by
2909 * "machinectl pull-tar" and related call may now do verification of
2910 downloaded images using SUSE-style .sha256 checksum files in addition
2911 to the already existing support for validating using Ubuntu-style
2914 * sd-bus gained support for a new sd_bus_message_appendv() call which
2915 is va_list equivalent of sd_bus_message_append().
2917 * sd-boot gained support for validating images using SHIM/MOK.
2919 * The SMACK code learnt support for "onlycap".
2921 * systemd-mount --umount is now much smarter in figuring out how to
2922 properly unmount a device given its mount or device path.
2924 * The code to call libnss_dns as a fallback from libnss_resolve when
2925 the communication with systemd-resolved fails was removed. This
2926 fallback was redundant and interfered with the [!UNAVAIL=return]
2927 suffix. See nss-resolve(8) for the recommended configuration.
2929 * systemd-logind may now be restarted without losing state. It stores
2930 the file descriptors for devices it manages in the system manager
2931 using the FDSTORE= mechanism. Please note that further changes in
2932 other components may be required to make use of this (for example
2933 Xorg has code to listen for stops of systemd-logind and terminate
2934 itself when logind is stopped or restarted, in order to avoid using
2935 stale file descriptors for graphical devices, which is now
2936 counterproductive and must be reverted in order for restarts of
2937 systemd-logind to be safe. See
2938 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
2940 * All kernel-install plugins are called with the environment variable
2941 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
2942 /etc/machine-id. If the machine ID could not be determined,
2943 $KERNEL_INSTALL_MACHINE_ID will be empty. Plugins should not put
2944 anything in the entry directory (passed as the second argument) if
2945 $KERNEL_INSTALL_MACHINE_ID is empty. For backwards compatibility, a
2946 temporary directory is passed as the entry directory and removed
2947 after all the plugins exit.
2949 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
2950 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
2951 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
2952 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
2953 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
2954 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
2955 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
2956 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
2957 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
2958 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
2959 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
2960 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
2961 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
2962 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
2963 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
2964 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
2965 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
2966 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
2967 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
2968 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
2969 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
2970 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
2971 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
2972 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
2973 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
2974 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
2975 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
2976 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
2977 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
2980 — Berlin, 2017-07-12
2984 * The "hybrid" control group mode has been modified to improve
2985 compatibility with "legacy" cgroups-v1 setups. Specifically, the
2986 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
2987 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
2988 cgroups-v1 hierarchy), the only externally visible change being that
2989 the cgroups-v2 hierarchy is also mounted, to
2990 /sys/fs/cgroup/unified. This should provide a large degree of
2991 compatibility with "legacy" cgroups-v1, while taking benefit of the
2992 better management capabilities of cgroups-v2.
2994 * The default control group setup mode may be selected both a boot-time
2995 via a set of kernel command line parameters (specifically:
2996 systemd.unified_cgroup_hierarchy= and
2997 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
2998 default selected on the configure command line
2999 (--with-default-hierarchy=). The upstream default is "hybrid"
3000 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
3001 this will change in a future systemd version to be "unified" (pure
3002 cgroups-v2 mode). The third option for the compile time option is
3003 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
3004 distributions to default to "hybrid" mode for release distributions,
3005 starting with v233. We recommend "unified" for development
3006 distributions (specifically: distributions such as Fedora's rawhide)
3007 as that's where things are headed in the long run. Use "legacy" for
3008 greatest stability and compatibility only.
3010 * Note one current limitation of "unified" and "hybrid" control group
3011 setup modes: the kernel currently does not permit the systemd --user
3012 instance (i.e. unprivileged code) to migrate processes between two
3013 disconnected cgroup subtrees, even if both are managed and owned by
3014 the user. This effectively means "systemd-run --user --scope" doesn't
3015 work when invoked from outside of any "systemd --user" service or
3016 scope. Specifically, it is not supported from session scopes. We are
3017 working on fixing this in a future systemd version. (See #3388 for
3018 further details about this.)
3020 * DBus policy files are now installed into /usr rather than /etc. Make
3021 sure your system has dbus >= 1.9.18 running before upgrading to this
3022 version, or override the install path with --with-dbuspolicydir= .
3024 * All python scripts shipped with systemd (specifically: the various
3025 tests written in Python) now require Python 3.
3027 * systemd unit tests can now run standalone (without the source or
3028 build directories), and can be installed into /usr/lib/systemd/tests/
3029 with 'make install-tests'.
3031 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
3032 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
3035 * Support for the %c, %r, %R specifiers in unit files has been
3036 removed. Specifiers are not supposed to be dependent on configuration
3037 in the unit file itself (so that they resolve the same regardless
3038 where used in the unit files), but these specifiers were influenced
3039 by the Slice= option.
3041 * The shell invoked by debug-shell.service now defaults to /bin/sh in
3042 all cases. If distributions want to use a different shell for this
3043 purpose (for example Fedora's /sbin/sushell) they need to specify
3044 this explicitly at configure time using --with-debug-shell=.
3046 * The confirmation spawn prompt has been reworked to offer the
3049 (c)ontinue, proceed without asking anymore
3050 (D)ump, show the state of the unit
3051 (f)ail, don't execute the command and pretend it failed
3053 (i)nfo, show a short summary of the unit
3054 (j)obs, show jobs that are in progress
3055 (s)kip, don't execute the command and pretend it succeeded
3056 (y)es, execute the command
3058 The 'n' choice for the confirmation spawn prompt has been removed,
3059 because its meaning was confusing.
3061 The prompt may now also be redirected to an alternative console by
3062 specifying the console as parameter to systemd.confirm_spawn=.
3064 * Services of Type=notify require a READY=1 notification to be sent
3065 during startup. If no such message is sent, the service now fails,
3066 even if the main process exited with a successful exit code.
3068 * Services that fail to start up correctly now always have their
3069 ExecStopPost= commands executed. Previously, they'd enter "failed"
3070 state directly, without executing these commands.
3072 * The option MulticastDNS= of network configuration files has acquired
3073 an actual implementation. With MulticastDNS=yes a host can resolve
3074 names of remote hosts and reply to mDNS A and AAAA requests.
3076 * When units are about to be started an additional check is now done to
3077 ensure that all dependencies of type BindsTo= (when used in
3078 combination with After=) have been started.
3080 * systemd-analyze gained a new verb "syscall-filter" which shows which
3081 system call groups are defined for the SystemCallFilter= unit file
3082 setting, and which system calls they contain.
3084 * A new system call filter group "@filesystem" has been added,
3085 consisting of various file system related system calls. Group
3086 "@reboot" has been added, covering reboot, kexec and shutdown related
3087 calls. Finally, group "@swap" has been added covering swap
3088 configuration related calls.
3090 * A new unit file option RestrictNamespaces= has been added that may be
3091 used to restrict access to the various process namespace types the
3092 Linux kernel provides. Specifically, it may be used to take away the
3093 right for a service unit to create additional file system, network,
3094 user, and other namespaces. This sandboxing option is particularly
3095 relevant due to the high amount of recently discovered namespacing
3096 related vulnerabilities in the kernel.
3098 * systemd-udev's .link files gained support for a new AutoNegotiation=
3099 setting for configuring Ethernet auto-negotiation.
3101 * systemd-networkd's .network files gained support for a new
3102 ListenPort= setting in the [DHCP] section to explicitly configure the
3103 UDP client port the DHCP client shall listen on.
3105 * .network files gained a new Unmanaged= boolean setting for explicitly
3106 excluding one or more interfaces from management by systemd-networkd.
3108 * The systemd-networkd ProxyARP= option has been renamed to
3109 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
3110 renamed to ReduceARPProxy=. The old names continue to be available
3113 * systemd-networkd gained support for configuring IPv6 Proxy NDP
3114 addresses via the new IPv6ProxyNDPAddress= .network file setting.
3116 * systemd-networkd's bonding device support gained support for two new
3117 configuration options ActiveSlave= and PrimarySlave=.
3119 * The various options in the [Match] section of .network files gained
3120 support for negative matching.
3122 * New systemd-specific mount options are now understood in /etc/fstab:
3124 x-systemd.mount-timeout= may be used to configure the maximum
3125 permitted runtime of the mount command.
3127 x-systemd.device-bound may be set to bind a mount point to its
3128 backing device unit, in order to automatically remove a mount point
3129 if its backing device is unplugged. This option may also be
3130 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
3131 on the block device, which is now automatically set for all CDROM
3132 drives, so that mounted CDs are automatically unmounted when they are
3133 removed from the drive.
3135 x-systemd.after= and x-systemd.before= may be used to explicitly
3136 order a mount after or before another unit or mount point.
3138 * Enqueued start jobs for device units are now automatically garbage
3139 collected if there are no jobs waiting for them anymore.
3141 * systemctl list-jobs gained two new switches: with --after, for every
3142 queued job the jobs it's waiting for are shown; with --before the
3143 jobs which it's blocking are shown.
3145 * systemd-nspawn gained support for ephemeral boots from disk images
3146 (or in other words: --ephemeral and --image= may now be
3147 combined). Moreover, ephemeral boots are now supported for normal
3148 directories, even if the backing file system is not btrfs. Of course,
3149 if the file system does not support file system snapshots or
3150 reflinks, the initial copy operation will be relatively expensive, but
3151 this should still be suitable for many use cases.
3153 * Calendar time specifications in .timer units now support
3154 specifications relative to the end of a month by using "~" instead of
3155 "-" as separator between month and day. For example, "*-02~03" means
3156 "the third last day in February". In addition a new syntax for
3157 repeated events has been added using the "/" character. For example,
3158 "9..17/2:00" means "every two hours from 9am to 5pm".
3160 * systemd-socket-proxyd gained a new parameter --connections-max= for
3161 configuring the maximum number of concurrent connections.
3163 * sd-id128 gained a new API for generating unique IDs for the host in a
3164 way that does not leak the machine ID. Specifically,
3165 sd_id128_get_machine_app_specific() derives an ID based on the
3166 machine ID a in well-defined, non-reversible, stable way. This is
3167 useful whenever an identifier for the host is needed but where the
3168 identifier shall not be useful to identify the system beyond the
3169 scope of the application itself. (Internally this uses HMAC-SHA256 as
3170 keyed hash function using the machine ID as input.)
3172 * NotifyAccess= gained a new supported value "exec". When set
3173 notifications are accepted from all processes systemd itself invoked,
3174 including all control processes.
3176 * .nspawn files gained support for defining overlay mounts using the
3177 Overlay= and OverlayReadOnly= options. Previously this functionality
3178 was only available on the systemd-nspawn command line.
3180 * systemd-nspawn's --bind= and --overlay= options gained support for
3181 bind/overlay mounts whose source lies within the container tree by
3182 prefixing the source path with "+".
3184 * systemd-nspawn's --bind= and --overlay= options gained support for
3185 automatically allocating a temporary source directory in /var/tmp
3186 that is removed when the container dies. Specifically, if the source
3187 directory is specified as empty string this mechanism is selected. An
3188 example usage is --overlay=+/var::/var, which creates an overlay
3189 mount based on the original /var contained in the image, overlayed
3190 with a temporary directory in the host's /var/tmp. This way changes
3191 to /var are automatically flushed when the container shuts down.
3193 * systemd-nspawn --image= option does now permit raw file system block
3194 devices (in addition to images containing partition tables, as
3197 * The disk image dissection logic in systemd-nspawn gained support for
3198 automatically setting up LUKS encrypted as well as Verity protected
3199 partitions. When a container is booted from an encrypted image the
3200 passphrase is queried at start-up time. When a container with Verity
3201 data is started, the root hash is search in a ".roothash" file
3202 accompanying the disk image (alternatively, pass the root hash via
3203 the new --root-hash= command line option).
3205 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
3206 be used to dissect disk images the same way as systemd-nspawn does
3207 it, following the Bootable Partition Specification. It may even be
3208 used to mount disk images with complex partition setups (including
3209 LUKS and Verity partitions) to a local host directory, in order to
3210 inspect them. This tool is not considered public API (yet), and is
3211 thus not installed into /usr/bin. Please do not rely on its
3212 existence, since it might go away or be changed in later systemd
3215 * A new generator "systemd-verity-generator" has been added, similar in
3216 style to "systemd-cryptsetup-generator", permitting automatic setup of
3217 Verity root partitions when systemd boots up. In order to make use of
3218 this your partition setup should follow the Discoverable Partitions
3219 Specification, and the GPT partition ID of the root file system
3220 partition should be identical to the upper 128bit of the Verity root
3221 hash. The GPT partition ID of the Verity partition protecting it
3222 should be the lower 128bit of the Verity root hash. If the partition
3223 image follows this model it is sufficient to specify a single
3224 "roothash=" kernel command line argument to both configure which root
3225 image and verity partition to use as well as the root hash for
3226 it. Note that systemd-nspawn's Verity support follows the same
3227 semantics, meaning that disk images with proper Verity data in place
3228 may be booted in containers with systemd-nspawn as well as on
3229 physical systems via the verity generator. Also note that the "mkosi"
3230 tool available at https://github.com/systemd/mkosi has been updated
3231 to generate Verity protected disk images following this scheme. In
3232 fact, it has been updated to generate disk images that optionally
3233 implement a complete UEFI SecureBoot trust chain, involving a signed
3234 kernel and initrd image that incorporates such a root hash as well as
3235 a Verity-enabled root partition.
3237 * The hardware database (hwdb) udev supports has been updated to carry
3238 accelerometer quirks.
3240 * All system services are now run with a fresh kernel keyring set up
3241 for them. The invocation ID is stored by default in it, thus
3242 providing a safe, non-overridable way to determine the invocation
3245 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
3246 options for bind mounting arbitrary paths in a service-specific
3247 way. When these options are used, arbitrary host or service files and
3248 directories may be mounted to arbitrary locations in the service's
3251 * Documentation has been added that lists all of systemd's low-level
3252 environment variables:
3254 https://github.com/systemd/systemd/blob/master/docs/ENVIRONMENT.md
3256 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
3257 whether a specific socket file descriptor matches a specified socket
3260 * systemd-firstboot has been updated to check for the
3261 systemd.firstboot= kernel command line option. It accepts a boolean
3262 and when set to false the first boot questions are skipped.
3264 * systemd-fstab-generator has been updated to check for the
3265 systemd.volatile= kernel command line option, which either takes an
3266 optional boolean parameter or the special value "state". If used the
3267 system may be booted in a "volatile" boot mode. Specifically,
3268 "systemd.volatile" is used, the root directory will be mounted as
3269 tmpfs, and only /usr is mounted from the actual root file system. If
3270 "systemd.volatile=state" is used, the root directory will be mounted
3271 as usual, but /var is mounted as tmpfs. This concept provides similar
3272 functionality as systemd-nspawn's --volatile= option, but provides it
3273 on physical boots. Use this option for implementing stateless
3274 systems, or testing systems with all state and/or configuration reset
3275 to the defaults. (Note though that many distributions are not
3276 prepared to boot up without a populated /etc or /var, though.)
3278 * systemd-gpt-auto-generator gained support for LUKS encrypted root
3279 partitions. Previously it only supported LUKS encrypted partitions
3280 for all other uses, except for the root partition itself.
3282 * Socket units gained support for listening on AF_VSOCK sockets for
3283 communication in virtualized QEMU environments.
3285 * The "configure" script gained a new option --with-fallback-hostname=
3286 for specifying the fallback hostname to use if none is configured in
3287 /etc/hostname. For example, by specifying
3288 --with-fallback-hostname=fedora it is possible to default to a
3289 hostname of "fedora" on pristine installations.
3291 * systemd-cgls gained support for a new --unit= switch for listing only
3292 the control groups of a specific unit. Similar --user-unit= has been
3293 added for listing only the control groups of a specific user unit.
3295 * systemd-mount gained a new --umount switch for unmounting a mount or
3296 automount point (and all mount/automount points below it).
3298 * systemd will now refuse full configuration reloads (via systemctl
3299 daemon-reload and related calls) unless at least 16MiB of free space
3300 are available in /run. This is a safety precaution in order to ensure
3301 that generators can safely operate after the reload completed.
3303 * A new unit file option RootImage= has been added, which has a similar
3304 effect as RootDirectory= but mounts the service's root directory from
3305 a disk image instead of plain directory. This logic reuses the same
3306 image dissection and mount logic that systemd-nspawn already uses,
3307 and hence supports any disk images systemd-nspawn supports, including
3308 those following the Discoverable Partition Specification, as well as
3309 Verity enabled images. This option enables systemd to run system
3310 services directly off disk images acting as resource bundles,
3311 possibly even including full integrity data.
3313 * A new MountAPIVFS= unit file option has been added, taking a boolean
3314 argument. If enabled /proc, /sys and /dev (collectively called the
3315 "API VFS") will be mounted for the service. This is only relevant if
3316 RootDirectory= or RootImage= is used for the service, as these mounts
3317 are of course in place in the host mount namespace anyway.
3319 * systemd-nspawn gained support for a new --pivot-root= switch. If
3320 specified the root directory within the container image is pivoted to
3321 the specified mount point, while the original root disk is moved to a
3322 different place. This option enables booting of ostree images
3323 directly with systemd-nspawn.
3325 * The systemd build scripts will no longer complain if the NTP server
3326 addresses are not changed from the defaults. Google now supports
3327 these NTP servers officially. We still recommend downstreams to
3328 properly register an NTP pool with the NTP pool project though.
3330 * coredumpctl gained a new "--reverse" option for printing the list
3331 of coredumps in reverse order.
3333 * coredumpctl will now show additional information about truncated and
3334 inaccessible coredumps, as well as coredumps that are still being
3335 processed. It also gained a new --quiet switch for suppressing
3336 additional informational message in its output.
3338 * coredumpctl gained support for only showing coredumps newer and/or
3339 older than specific timestamps, using the new --since= and --until=
3340 options, reminiscent of journalctl's options by the same name.
3342 * The systemd-coredump logic has been improved so that it may be reused
3343 to collect backtraces in non-compiled languages, for example in
3344 scripting languages such as Python.
3346 * machinectl will now show the UID shift of local containers, if user
3347 namespacing is enabled for them.
3349 * systemd will now optionally run "environment generator" binaries at
3350 configuration load time. They may be used to add environment
3351 variables to the environment block passed to services invoked. One
3352 user environment generator is shipped by default that sets up
3353 environment variables based on files dropped into /etc/environment.d
3354 and ~/.config/environment.d/.
3356 * systemd-resolved now includes the new, recently published 2017 DNSSEC
3359 * hostnamed has been updated to report a new chassis type of
3360 "convertible" to cover "foldable" laptops that can both act as a
3361 tablet and as a laptop, such as various Lenovo Yoga devices.
3363 Contributions from: Adrián López, Alexander Galanin, Alexander
3364 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
3365 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
3366 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
3367 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
3368 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
3369 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
3370 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
3371 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
3372 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
3373 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
3374 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
3375 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
3376 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
3377 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
3378 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
3379 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
3380 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
3381 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
3382 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
3383 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
3384 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
3385 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
3386 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
3387 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
3388 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
3389 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
3390 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
3393 — Berlin, 2017-03-01
3397 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
3398 RestrictAddressFamilies= enabled. These sandboxing options should
3399 generally be compatible with the various external udev call-out
3400 binaries we are aware of, however there may be exceptions, in
3401 particular when exotic languages for these call-outs are used. In
3402 this case, consider turning off these settings locally.
3404 * The new RemoveIPC= option can be used to remove IPC objects owned by
3405 the user or group of a service when that service exits.
3407 * The new ProtectKernelModules= option can be used to disable explicit
3408 load and unload operations of kernel modules by a service. In
3409 addition access to /usr/lib/modules is removed if this option is set.
3411 * ProtectSystem= option gained a new value "strict", which causes the
3412 whole file system tree with the exception of /dev, /proc, and /sys,
3413 to be remounted read-only for a service.
3415 * The new ProtectKernelTunables= option can be used to disable
3416 modification of configuration files in /sys and /proc by a service.
3417 Various directories and files are remounted read-only, so access is
3418 restricted even if the file permissions would allow it.
3420 * The new ProtectControlGroups= option can be used to disable write
3421 access by a service to /sys/fs/cgroup.
3423 * Various systemd services have been hardened with
3424 ProtectKernelTunables=yes, ProtectControlGroups=yes,
3425 RestrictAddressFamilies=.
3427 * Support for dynamically creating users for the lifetime of a service
3428 has been added. If DynamicUser=yes is specified, user and group IDs
3429 will be allocated from the range 61184..65519 for the lifetime of the
3430 service. They can be resolved using the new nss-systemd.so NSS
3431 module. The module must be enabled in /etc/nsswitch.conf. Services
3432 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
3433 any resources allocated by the service will be cleaned up when the
3434 service exits. They also have ProtectHome=read-only and
3435 ProtectSystem=strict enabled, so they are not able to make any
3436 permanent modifications to the system.
3438 * The nss-systemd module also always resolves root and nobody, making
3439 it possible to have no /etc/passwd or /etc/group files in minimal
3440 container or chroot environments.
3442 * Services may be started with their own user namespace using the new
3443 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
3444 under which the service is running are mapped. All other users are
3447 * Support for the cgroup namespace has been added to systemd-nspawn. If
3448 supported by kernel, the container system started by systemd-nspawn
3449 will have its own view of the cgroup hierarchy. This new behaviour
3450 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
3452 * The new MemorySwapMax= option can be used to limit the maximum swap
3453 usage under the unified cgroup hierarchy.
3455 * Support for the CPU controller in the unified cgroup hierarchy has
3456 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
3457 options. This controller requires out-of-tree patches for the kernel
3458 and the support is provisional.
3460 * Mount and automount units may now be created transiently
3461 (i.e. dynamically at runtime via the bus API, instead of requiring
3462 unit files in the file system).
3464 * systemd-mount is a new tool which may mount file systems – much like
3465 mount(8), optionally pulling in additional dependencies through
3466 transient .mount and .automount units. For example, this tool
3467 automatically runs fsck on a backing block device before mounting,
3468 and allows the automount logic to be used dynamically from the
3469 command line for establishing mount points. This tool is particularly
3470 useful when dealing with removable media, as it will ensure fsck is
3471 run – if necessary – before the first access and that the file system
3472 is quickly unmounted after each access by utilizing the automount
3473 logic. This maximizes the chance that the file system on the
3474 removable media stays in a clean state, and if it isn't in a clean
3475 state is fixed automatically.
3477 * LazyUnmount=yes option for mount units has been added to expose the
3478 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
3481 * /efi will be used as the mount point of the EFI boot partition, if
3482 the directory is present, and the mount point was not configured
3483 through other means (e.g. fstab). If /efi directory does not exist,
3484 /boot will be used as before. This makes it easier to automatically
3485 mount the EFI partition on systems where /boot is used for something
3488 * When operating on GPT disk images for containers, systemd-nspawn will
3489 now mount the ESP to /boot or /efi according to the same rules as PID
3490 1 running on a host. This allows tools like "bootctl" to operate
3491 correctly within such containers, in order to make container images
3492 bootable on physical systems.
3494 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
3496 * Two new user session targets have been added to support running
3497 graphical sessions under the systemd --user instance:
3498 graphical-session.target and graphical-session-pre.target. See
3499 systemd.special(7) for a description of how those targets should be
3502 * The vconsole initialization code has been significantly reworked to
3503 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
3504 support unicode keymaps. Font and keymap configuration will now be
3505 copied to all allocated virtual consoles.
3507 * FreeBSD's bhyve virtualization is now detected.
3509 * Information recorded in the journal for core dumps now includes the
3510 contents of /proc/mountinfo and the command line of the process at
3511 the top of the process hierarchy (which is usually the init process
3514 * systemd-journal-gatewayd learned the --directory= option to serve
3515 files from the specified location.
3517 * journalctl --root=… can be used to peruse the journal in the
3518 /var/log/ directories inside of a container tree. This is similar to
3519 the existing --machine= option, but does not require the container to
3522 * The hardware database has been extended to support
3523 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
3526 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
3527 specify the click rate for mice which include a horizontal wheel with
3528 a click rate that is different than the one for the vertical wheel.
3530 * systemd-run gained a new --wait option that makes service execution
3531 synchronous. (Specifically, the command will not return until the
3532 specified service binary exited.)
3534 * systemctl gained a new --wait option that causes the start command to
3535 wait until the units being started have terminated again.
3537 * A new journal output mode "short-full" has been added which displays
3538 timestamps with abbreviated English day names and adds a timezone
3539 suffix. Those timestamps include more information than the default
3540 "short" output mode, and can be passed directly to journalctl's
3541 --since= and --until= options.
3543 * /etc/resolv.conf will be bind-mounted into containers started by
3544 systemd-nspawn, if possible, so any changes to resolv.conf contents
3545 are automatically propagated to the container.
3547 * The number of instances for socket-activated services originating
3548 from a single IP address can be limited with
3549 MaxConnectionsPerSource=, extending the existing setting of
3552 * systemd-networkd gained support for vcan ("Virtual CAN") interface
3555 * .netdev and .network configuration can now be extended through
3558 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
3559 Segmentation Offload, Generic Receive Offload, Large Receive Offload
3560 can be enabled and disabled using the new UDPSegmentationOffload=,
3561 TCPSegmentationOffload=, GenericSegmentationOffload=,
3562 GenericReceiveOffload=, LargeReceiveOffload= options in the
3563 [Link] section of .link files.
3565 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
3566 Port VLAN ID can be configured for bridge devices using the new STP=,
3567 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
3568 section of .netdev files.
3570 * The route table to which routes received over DHCP or RA should be
3571 added can be configured with the new RouteTable= option in the [DHCP]
3572 and [IPv6AcceptRA] sections of .network files.
3574 * The Address Resolution Protocol can be disabled on links managed by
3575 systemd-networkd using the ARP=no setting in the [Link] section of
3578 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
3579 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
3580 encode information about the result and exit codes of the current
3581 service runtime cycle.
3583 * systemd-sysctl will now configure kernel parameters in the order
3584 they occur in the configuration files. This matches what sysctl
3585 has been traditionally doing.
3587 * kernel-install "plugins" that are executed to perform various
3588 tasks after a new kernel is added and before an old one is removed
3589 can now return a special value to terminate the procedure and
3590 prevent any later plugins from running.
3592 * Journald's SplitMode=login setting has been deprecated. It has been
3593 removed from documentation, and its use is discouraged. In a future
3594 release it will be completely removed, and made equivalent to current
3595 default of SplitMode=uid.
3597 * Storage=both option setting in /etc/systemd/coredump.conf has been
3598 removed. With fast LZ4 compression storing the core dump twice is not
3601 * The --share-system systemd-nspawn option has been replaced with an
3602 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
3603 this functionality is discouraged. In addition the variables
3604 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
3605 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
3606 individual namespaces.
3608 * "machinectl list" now shows the IP address of running containers in
3609 the output, as well as OS release information.
3611 * "loginctl list" now shows the TTY of each session in the output.
3613 * sd-bus gained new API calls sd_bus_track_set_recursive(),
3614 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
3615 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
3616 tracking objects in a "recursive" mode, where a single client can be
3617 counted multiple times, if it takes multiple references.
3619 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
3620 sd_bus_get_exit_on_disconnect(). They may be used to make a
3621 process using sd-bus automatically exit if the bus connection is
3624 * Bus clients of the service manager may now "pin" loaded units into
3625 memory, by taking an explicit reference on them. This is useful to
3626 ensure the client can retrieve runtime data about the service even
3627 after the service completed execution. Taking such a reference is
3628 available only for privileged clients and should be helpful to watch
3629 running services in a race-free manner, and in particular collect
3630 information about exit statuses and results.
3632 * The nss-resolve module has been changed to strictly return UNAVAIL
3633 when communication via D-Bus with resolved failed, and NOTFOUND when
3634 a lookup completed but was negative. This means it is now possible to
3635 neatly configure fallbacks using nsswitch.conf result checking
3636 expressions. Taking benefit of this, the new recommended
3637 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
3639 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
3641 * A new setting CtrlAltDelBurstAction= has been added to
3642 /etc/systemd/system.conf which may be used to configure the precise
3643 behaviour if the user on the console presses Ctrl-Alt-Del more often
3644 than 7 times in 2s. Previously this would unconditionally result in
3645 an expedited, immediate reboot. With this new setting the precise
3646 operation may be configured in more detail, and also turned off
3649 * In .netdev files two new settings RemoteChecksumTx= and
3650 RemoteChecksumRx= are now understood that permit configuring the
3651 remote checksumming logic for VXLAN networks.
3653 * The service manager learnt a new "invocation ID" concept for invoked
3654 services. Each runtime cycle of a service will get a new invocation
3655 ID (a 128bit random UUID) assigned that identifies the current
3656 run of the service uniquely and globally. A new invocation ID
3657 is generated each time a service starts up. The journal will store
3658 the invocation ID of a service along with any logged messages, thus
3659 making the invocation ID useful for matching the online runtime of a
3660 service with the offline log data it generated in a safe way without
3661 relying on synchronized timestamps. In many ways this new service
3662 invocation ID concept is similar to the kernel's boot ID concept that
3663 uniquely and globally identifies the runtime of each boot. The
3664 invocation ID of a service is passed to the service itself via an
3665 environment variable ($INVOCATION_ID). A new bus call
3666 GetUnitByInvocationID() has been added that is similar to GetUnit()
3667 but instead of retrieving the bus path for a unit by its name
3668 retrieves it by its invocation ID. The returned path is valid only as
3669 long as the passed invocation ID is current.
3671 * systemd-resolved gained a new "DNSStubListener" setting in
3672 resolved.conf. It either takes a boolean value or the special values
3673 "udp" and "tcp", and configures whether to enable the stub DNS
3674 listener on 127.0.0.53:53.
3676 * IP addresses configured via networkd may now carry additional
3677 configuration settings supported by the kernel. New options include:
3678 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
3679 PrefixRoute=, AutoJoin=.
3681 * The PAM configuration fragment file for "user@.service" shipped with
3682 systemd (i.e. the --user instance of systemd) has been stripped to
3683 the minimum necessary to make the system boot. Previously, it
3684 contained Fedora-specific stanzas that did not apply to other
3685 distributions. It is expected that downstream distributions add
3686 additional configuration lines, matching their needs to this file,
3687 using it only as rough template of what systemd itself needs. Note
3688 that this reduced fragment does not even include an invocation of
3689 pam_limits which most distributions probably want to add, even though
3690 systemd itself does not need it. (There's also the new build time
3691 option --with-pamconfdir=no to disable installation of the PAM
3694 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
3695 capability is now also dropped from its set (in addition to
3696 CAP_SYS_MKNOD as before).
3698 * In service unit files it is now possible to connect a specific named
3699 file descriptor with stdin/stdout/stdout of an executed service. The
3700 name may be specified in matching .socket units using the
3701 FileDescriptorName= setting.
3703 * A number of journal settings may now be configured on the kernel
3704 command line. Specifically, the following options are now understood:
3705 systemd.journald.max_level_console=,
3706 systemd.journald.max_level_store=,
3707 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
3708 systemd.journald.max_level_wall=.
3710 * "systemctl is-enabled --full" will now show by which symlinks a unit
3711 file is enabled in the unit dependency tree.
3713 * Support for VeraCrypt encrypted partitions has been added to the
3714 "cryptsetup" logic and /etc/crypttab.
3716 * systemd-detect-virt gained support for a new --private-users switch
3717 that checks whether the invoking processes are running inside a user
3718 namespace. Similar, a new special value "private-users" for the
3719 existing ConditionVirtualization= setting has been added, permitting
3720 skipping of specific units in user namespace environments.
3722 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
3723 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
3724 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
3725 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
3726 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
3727 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
3728 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
3729 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
3730 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
3731 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
3732 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
3733 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
3734 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
3735 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
3736 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
3737 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
3738 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
3739 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
3740 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
3741 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
3742 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
3743 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
3744 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
3745 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
3746 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
3747 Jędrzejewski-Szmek, Zeal Jagannatha
3749 — Santa Fe, 2016-11-03
3753 * In service units the various ExecXYZ= settings have been extended
3754 with an additional special character as first argument of the
3755 assigned value: if the character '+' is used the specified command
3756 line it will be run with full privileges, regardless of User=,
3757 Group=, CapabilityBoundingSet= and similar options. The effect is
3758 similar to the existing PermissionsStartOnly= option, but allows
3759 configuration of this concept for each executed command line
3762 * Services may now alter the service watchdog timeout at runtime by
3763 sending a WATCHDOG_USEC= message via sd_notify().
3765 * MemoryLimit= and related unit settings now optionally take percentage
3766 specifications. The percentage is taken relative to the amount of
3767 physical memory in the system (or in case of containers, the assigned
3768 amount of memory). This allows scaling service resources neatly with
3769 the amount of RAM available on the system. Similarly, systemd-logind's
3770 RuntimeDirectorySize= option now also optionally takes percentage
3773 * In similar fashion TasksMax= takes percentage values now, too. The
3774 value is taken relative to the configured maximum number of processes
3775 on the system. The per-service task maximum has been changed to 15%
3776 using this functionality. (Effectively this is an increase of 512 →
3777 4915 for service units, given the kernel's default pid_max setting.)
3779 * Calendar time specifications in .timer units now understand a ".."
3780 syntax for time ranges. Example: "4..7:10" may now be used for
3781 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
3784 * The InaccessableDirectories=, ReadOnlyDirectories= and
3785 ReadWriteDirectories= unit file settings have been renamed to
3786 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
3787 applied to all kinds of file nodes, and not just directories, with
3788 the exception of symlinks. Specifically these settings may now be
3789 used on block and character device nodes, UNIX sockets and FIFOS as
3790 well as regular files. The old names of these settings remain
3791 available for compatibility.
3793 * systemd will now log about all service processes it kills forcibly
3794 (using SIGKILL) because they remained after the clean shutdown phase
3795 of the service completed. This should help identifying services that
3796 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
3797 systemd-logind's configuration a similar log message is generated for
3798 processes killed at the end of each session due to this setting.
3800 * systemd will now set the $JOURNAL_STREAM environment variable for all
3801 services whose stdout/stderr are connected to the Journal (which
3802 effectively means by default: all services). The variable contains
3803 the device and inode number of the file descriptor used for
3804 stdout/stderr. This may be used by invoked programs to detect whether
3805 their stdout/stderr is connected to the Journal, in which case they
3806 can switch over to direct Journal communication, thus being able to
3807 pass extended, structured metadata along with their log messages. As
3808 one example, this is now used by glib's logging primitives.
3810 * When using systemd's default tmp.mount unit for /tmp, the mount point
3811 will now be established with the "nosuid" and "nodev" options. This
3812 avoids privilege escalation attacks that put traps and exploits into
3813 /tmp. However, this might cause problems if you e. g. put container
3814 images or overlays into /tmp; if you need this, override tmp.mount's
3815 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
3818 * systemd now supports the "memory" cgroup controller also on
3821 * The systemd-cgtop tool now optionally takes a control group path as
3822 command line argument. If specified, the control group list shown is
3823 limited to subgroups of that group.
3825 * The SystemCallFilter= unit file setting gained support for
3826 pre-defined, named system call filter sets. For example
3827 SystemCallFilter=@clock is now an effective way to make all clock
3828 changing-related system calls unavailable to a service. A number of
3829 similar pre-defined groups are defined. Writing system call filters
3830 for system services is simplified substantially with this new
3831 concept. Accordingly, all of systemd's own, long-running services now
3832 enable system call filtering based on this, by default.
3834 * A new service setting MemoryDenyWriteExecute= has been added, taking
3835 a boolean value. If turned on, a service may no longer create memory
3836 mappings that are writable and executable at the same time. This
3837 enhances security for services where this is enabled as it becomes
3838 harder to dynamically write and then execute memory in exploited
3839 service processes. This option has been enabled for all of systemd's
3840 own long-running services.
3842 * A new RestrictRealtime= service setting has been added, taking a
3843 boolean argument. If set the service's processes may no longer
3844 acquire realtime scheduling. This improves security as realtime
3845 scheduling may otherwise be used to easily freeze the system.
3847 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
3848 value. This may be used for requesting that the system manager inside
3849 of the container reports start-up completion to nspawn which then
3850 propagates this notification further to the service manager
3851 supervising nspawn itself. A related option NotifyReady= in .nspawn
3852 files has been added too. This functionality allows ordering of the
3853 start-up of multiple containers using the usual systemd ordering
3856 * machinectl gained a new command "stop" that is an alias for
3859 * systemd-resolved gained support for contacting DNS servers on
3860 link-local IPv6 addresses.
3862 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
3863 its caches. A method call for requesting the same operation has been
3864 added to the bus API too, and is made available via "systemd-resolve
3867 * systemd-resolve gained a new --status switch. If passed a brief
3868 summary of the used DNS configuration with per-interface information
3871 * resolved.conf gained a new Cache= boolean option, defaulting to
3872 on. If turned off local DNS caching is disabled. This comes with a
3873 performance penalty in particular when DNSSEC is enabled. Note that
3874 resolved disables its internal caching implicitly anyway, when the
3875 configured DNS server is on a host-local IP address such as ::1 or
3876 127.0.0.1, thus automatically avoiding double local caching.
3878 * systemd-resolved now listens on the local IP address 127.0.0.53:53
3879 for DNS requests. This improves compatibility with local programs
3880 that do not use the libc NSS or systemd-resolved's bus APIs for name
3881 resolution. This minimal DNS service is only available to local
3882 programs and does not implement the full DNS protocol, but enough to
3883 cover local DNS clients. A new, static resolv.conf file, listing just
3884 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
3885 now recommended to make /etc/resolv.conf a symlink to this file in
3886 order to route all DNS lookups to systemd-resolved, regardless if
3887 done via NSS, the bus API or raw DNS packets. Note that this local
3888 DNS service is not as fully featured as the libc NSS or
3889 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
3890 used to deliver link-local address information (as this implies
3891 sending a local interface index along), LLMNR/mDNS support via this
3892 interface is severely restricted. It is thus strongly recommended for
3893 all applications to use the libc NSS API or native systemd-resolved
3896 * systemd-networkd's bridge support learned a new setting
3897 VLANFiltering= for controlling VLAN filtering. Moreover a new section
3898 in .network files has been added for configuring VLAN bridging in
3899 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
3901 * systemd-networkd's IPv6 Router Advertisement code now makes use of
3902 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
3903 now be acquired without relying on DHCPv6. Two new options
3904 UseDomains= and UseDNS= have been added to configure this behaviour.
3906 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
3907 renamed IPv6AcceptRA=, without altering its behaviour. The old
3908 setting name remains available for compatibility reasons.
3910 * The systemd-networkd VTI/VTI6 tunneling support gained new options
3911 Key=, InputKey= and OutputKey=.
3913 * systemd-networkd gained support for VRF ("Virtual Routing Function")
3914 interface configuration.
3916 * "systemctl edit" may now be used to create new unit files by
3917 specifying the --force switch.
3919 * sd-event gained a new function sd_event_get_iteration() for
3920 requesting the current iteration counter of the event loop. It starts
3921 at zero and is increased by one with each event loop iteration.
3923 * A new rpm macro %systemd_ordering is provided by the macros.systemd
3924 file. It can be used in lieu of %systemd_requires in packages which
3925 don't use any systemd functionality and are intended to be installed
3926 in minimal containers without systemd present. This macro provides
3927 ordering dependencies to ensure that if the package is installed in
3928 the same rpm transaction as systemd, systemd will be installed before
3929 the scriptlets for the package are executed, allowing unit presets
3932 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
3933 been added to simplify packaging of generators.
3935 * The os-release file gained VERSION_CODENAME field for the
3936 distribution nickname (e.g. VERSION_CODENAME=woody).
3938 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
3939 can be set to disable parsing of metadata and the creation
3940 of persistent symlinks for that device.
3942 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
3943 to make them available to logged-in users has been reverted.
3945 * Much of the common code of the various systemd components is now
3946 built into an internal shared library libsystemd-shared-231.so
3947 (incorporating the systemd version number in the name, to be updated
3948 with future releases) that the components link to. This should
3949 decrease systemd footprint both in memory during runtime and on
3950 disk. Note that the shared library is not for public use, and is
3951 neither API nor ABI stable, but is likely to change with every new
3952 released update. Packagers need to make sure that binaries
3953 linking to libsystemd-shared.so are updated in step with the
3956 * Configuration for "mkosi" is now part of the systemd
3957 repository. mkosi is a tool to easily build legacy-free OS images,
3958 and is available on github: https://github.com/systemd/mkosi. If
3959 "mkosi" is invoked in the build tree a new raw OS image is generated
3960 incorporating the systemd sources currently being worked on and a
3961 clean, fresh distribution installation. The generated OS image may be
3962 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
3963 UEFI PC. This functionality is particularly useful to easily test
3964 local changes made to systemd in a pristine, defined environment. See
3965 doc/HACKING for details.
3967 * configure learned the --with-support-url= option to specify the
3968 distribution's bugtracker.
3970 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
3971 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
3972 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
3973 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
3974 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
3975 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
3976 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
3977 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
3978 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
3979 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
3980 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
3981 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
3982 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
3983 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
3984 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
3985 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
3986 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
3987 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
3988 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
3990 — Berlin, 2016-07-25
3994 * DNSSEC is now turned on by default in systemd-resolved (in
3995 "allow-downgrade" mode), but may be turned off during compile time by
3996 passing "--with-default-dnssec=no" to "configure" (and of course,
3997 during runtime with DNSSEC= in resolved.conf). We recommend
3998 downstreams to leave this on at least during development cycles and
3999 report any issues with the DNSSEC logic upstream. We are very
4000 interested in collecting feedback about the DNSSEC validator and its
4001 limitations in the wild. Note however, that DNSSEC support is
4002 probably nothing downstreams should turn on in stable distros just
4003 yet, as it might create incompatibilities with a few DNS servers and
4004 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
4005 automatically whenever we detect such incompatible setups, but there
4006 might be systems we do not cover yet. Hence: please help us testing
4007 the DNSSEC code, leave this on where you can, report back, but then
4008 again don't consider turning this on in your stable, LTS or
4009 production release just yet. (Note that you have to enable
4010 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
4011 and its DNSSEC mode for host name resolution from local
4014 * systemd-resolve conveniently resolves DANE records with the --tlsa
4015 option and OPENPGPKEY records with the --openpgp option. It also
4016 supports dumping raw DNS record data via the new --raw= switch.
4018 * systemd-logind will now by default terminate user processes that are
4019 part of the user session scope unit (session-XX.scope) when the user
4020 logs out. This behavior is controlled by the KillUserProcesses=
4021 setting in logind.conf, and the previous default of "no" is now
4022 changed to "yes". This means that user sessions will be properly
4023 cleaned up after, but additional steps are necessary to allow
4024 intentionally long-running processes to survive logout.
4026 While the user is logged in at least once, user@.service is running,
4027 and any service that should survive the end of any individual login
4028 session can be started at a user service or scope using systemd-run.
4029 systemd-run(1) man page has been extended with an example which shows
4030 how to run screen in a scope unit underneath user@.service. The same
4031 command works for tmux.
4033 After the user logs out of all sessions, user@.service will be
4034 terminated too, by default, unless the user has "lingering" enabled.
4035 To effectively allow users to run long-term tasks even if they are
4036 logged out, lingering must be enabled for them. See loginctl(1) for
4037 details. The default polkit policy was modified to allow users to
4038 set lingering for themselves without authentication.
4040 Previous defaults can be restored at compile time by the
4041 --without-kill-user-processes option to "configure".
4043 * systemd-logind gained new configuration settings SessionsMax= and
4044 InhibitorsMax=, both with a default of 8192. It will not register new
4045 user sessions or inhibitors above this limit.
4047 * systemd-logind will now reload configuration on SIGHUP.
4049 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
4050 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
4051 enable. Also, support for the "io" cgroup controller in the unified
4052 hierarchy has been added, so that the "memory", "pids" and "io" are
4053 now the controllers that are supported on the unified hierarchy.
4055 WARNING: it is not possible to use previous systemd versions with
4056 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
4057 is necessary to also update systemd in the initramfs if using the
4058 unified hierarchy. An updated SELinux policy is also required.
4060 * LLDP support has been extended, and both passive (receive-only) and
4061 active (sender) modes are supported. Passive mode ("routers-only") is
4062 enabled by default in systemd-networkd. Active LLDP mode is enabled
4063 by default for containers on the internal network. The "networkctl
4064 lldp" command may be used to list information gathered. "networkctl
4065 status" will also show basic LLDP information on connected peers now.
4067 * The IAID and DUID unique identifier sent in DHCP requests may now be
4068 configured for the system and each .network file managed by
4069 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
4071 * systemd-networkd gained support for configuring proxy ARP support for
4072 each interface, via the ProxyArp= setting in .network files. It also
4073 gained support for configuring the multicast querier feature of
4074 bridge devices, via the new MulticastQuerier= setting in .netdev
4075 files. Similarly, snooping on the IGMP traffic can be controlled
4076 via the new setting MulticastSnooping=.
4078 A new setting PreferredLifetime= has been added for addresses
4079 configured in .network file to configure the lifetime intended for an
4082 The systemd-networkd DHCP server gained the option EmitRouter=, which
4083 defaults to yes, to configure whether the DHCP Option 3 (Router)
4086 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
4087 systemd-socket-activate and installed into /usr/bin. It is now fully
4090 * systemd-journald now uses separate threads to flush changes to disk
4091 when closing journal files, thus reducing impact of slow disk I/O on
4092 logging performance.
4094 * The sd-journal API gained two new calls
4095 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
4096 can be used to open journal files using file descriptors instead of
4097 file or directory paths. sd_journal_open_container() has been
4098 deprecated, sd_journal_open_directory_fd() should be used instead
4099 with the flag SD_JOURNAL_OS_ROOT.
4101 * journalctl learned a new output mode "-o short-unix" that outputs log
4102 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
4103 UTC). It also gained support for a new --no-hostname setting to
4104 suppress the hostname column in the family of "short" output modes.
4106 * systemd-ask-password now optionally skips printing of the password to
4107 stdout with --no-output which can be useful in scripts.
4109 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
4110 (devices tagged with ID_MAKER_TOOL) are now tagged with
4111 "uaccess" and are available to logged in users.
4113 * The DeviceAllow= unit setting now supports specifiers (with "%").
4115 * "systemctl show" gained a new --value switch, which allows print a
4116 only the contents of a specific unit property, without also printing
4117 the property's name. Similar support was added to "show*" verbs
4118 of loginctl and machinectl that output "key=value" lists.
4120 * A new unit type "generated" was added for files dynamically generated
4121 by generator tools. Similarly, a new unit type "transient" is used
4122 for unit files created using the runtime API. "systemctl enable" will
4123 refuse to operate on such files.
4125 * A new command "systemctl revert" has been added that may be used to
4126 revert to the vendor version of a unit file, in case local changes
4127 have been made by adding drop-ins or overriding the unit file.
4129 * "machinectl clean" gained a new verb to automatically remove all or
4130 just hidden container images.
4132 * systemd-tmpfiles gained support for a new line type "e" for emptying
4133 directories, if they exist, without creating them if they don't.
4135 * systemd-nspawn gained support for automatically patching the UID/GIDs
4136 of the owners and the ACLs of all files and directories in a
4137 container tree to match the UID/GID user namespacing range selected
4138 for the container invocation. This mode is enabled via the new
4139 --private-users-chown switch. It also gained support for
4140 automatically choosing a free, previously unused UID/GID range when
4141 starting a container, via the new --private-users=pick setting (which
4142 implies --private-users-chown). Together, these options for the first
4143 time make user namespacing for nspawn containers fully automatic and
4144 thus deployable. The systemd-nspawn@.service template unit file has
4145 been changed to use this functionality by default.
4147 * systemd-nspawn gained a new --network-zone= switch, that allows
4148 creating ad-hoc virtual Ethernet links between multiple containers,
4149 that only exist as long as at least one container referencing them is
4150 running. This allows easy connecting of multiple containers with a
4151 common link that implements an Ethernet broadcast domain. Each of
4152 these network "zones" may be named relatively freely by the user, and
4153 may be referenced by any number of containers, but each container may
4154 only reference one of these "zones". On the lower level, this is
4155 implemented by an automatically managed bridge network interface for
4156 each zone, that is created when the first container referencing its
4157 zone is created and removed when the last one referencing its zone
4160 * The default start timeout may now be configured on the kernel command
4161 line via systemd.default_timeout_start_sec=. It was already
4162 configurable via the DefaultTimeoutStartSec= option in
4163 /etc/systemd/system.conf.
4165 * Socket units gained a new TriggerLimitIntervalSec= and
4166 TriggerLimitBurst= setting to configure a limit on the activation
4167 rate of the socket unit.
4169 * The LimitNICE= setting now optionally takes normal UNIX nice values
4170 in addition to the raw integer limit value. If the specified
4171 parameter is prefixed with "+" or "-" and is in the range -20..19 the
4172 value is understood as UNIX nice value. If not prefixed like this it
4173 is understood as raw RLIMIT_NICE limit.
4175 * Note that the effect of the PrivateDevices= unit file setting changed
4176 slightly with this release: the per-device /dev file system will be
4177 mounted read-only from this version on, and will have "noexec"
4178 set. This (minor) change of behavior might cause some (exceptional)
4179 legacy software to break, when PrivateDevices=yes is set for its
4180 service. Please leave PrivateDevices= off if you run into problems
4183 * systemd-bootchart has been split out to a separate repository:
4184 https://github.com/systemd/systemd-bootchart
4186 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
4187 merged into the kernel in its current form.
4189 * The compatibility libraries libsystemd-daemon.so,
4190 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
4191 which have been deprecated since systemd-209 have been removed along
4192 with the corresponding pkg-config files. All symbols provided by
4193 those libraries are provided by libsystemd.so.
4195 * The Capabilities= unit file setting has been removed (it is ignored
4196 for backwards compatibility). AmbientCapabilities= and
4197 CapabilityBoundingSet= should be used instead.
4199 * A new special target has been added, initrd-root-device.target,
4200 which creates a synchronization point for dependencies of the root
4201 device in early userspace. Initramfs builders must ensure that this
4202 target is now included in early userspace.
4204 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
4205 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
4206 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
4207 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
4208 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
4209 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
4210 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
4211 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
4212 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
4213 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
4214 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
4215 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
4216 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
4217 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
4218 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
4219 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
4220 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
4221 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
4222 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
4223 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
4224 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
4225 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
4226 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
4227 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
4228 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
4231 — Fairfax, 2016-05-21
4235 * The systemd-resolved DNS resolver service has gained a substantial
4236 set of new features, most prominently it may now act as a DNSSEC
4237 validating stub resolver. DNSSEC mode is currently turned off by
4238 default, but is expected to be turned on by default in one of the
4239 next releases. For now, we invite everybody to test the DNSSEC logic
4240 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
4241 service also gained a full set of D-Bus interfaces, including calls
4242 to configure DNS and DNSSEC settings per link (for use by external
4243 network management software). systemd-resolved and systemd-networkd
4244 now distinguish between "search" and "routing" domains. The former
4245 are used to qualify single-label names, the latter are used purely
4246 for routing lookups within certain domains to specific links.
4247 resolved now also synthesizes RRs for all entries from /etc/hosts.
4249 * The systemd-resolve tool (which is a client utility for
4250 systemd-resolved) has been improved considerably and is now fully
4251 supported and documented. Hence it has moved from /usr/lib/systemd to
4254 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
4257 * The coredump collection logic has been reworked: when a coredump is
4258 collected it is now written to disk, compressed and processed
4259 (including stacktrace extraction) from a new instantiated service
4260 systemd-coredump@.service, instead of directly from the
4261 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
4262 processing large coredumps can take up a substantial amount of
4263 resources and time, and this previously happened entirely outside of
4264 systemd's service supervision. With the new logic the core_pattern
4265 hook only does minimal metadata collection before passing off control
4266 to the new instantiated service, which is configured with a time
4267 limit, a nice level and other settings to minimize negative impact on
4268 the rest of the system. Also note that the new logic will honour the
4269 RLIMIT_CORE setting of the crashed process, which now allows users
4270 and processes to turn off coredumping for their processes by setting
4273 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
4274 and all forked processes by default. Previously, PID 1 would leave
4275 the setting at "0" for all processes, as set by the kernel. Note that
4276 the resource limit traditionally has no effect on the generated
4277 coredumps on the system if the /proc/sys/kernel/core_pattern hook
4278 logic is used. Since the limit is now honoured (see above) its
4279 default has been changed so that the coredumping logic is enabled by
4280 default for all processes, while allowing specific opt-out.
4282 * When the stacktrace is extracted from processes of system users, this
4283 is now done as "systemd-coredump" user, in order to sandbox this
4284 potentially security sensitive parsing operation. (Note that when
4285 processing coredumps of normal users this is done under the user ID
4286 of process that crashed, as before.) Packagers should take notice
4287 that it is now necessary to create the "systemd-coredump" system user
4288 and group at package installation time.
4290 * The systemd-activate socket activation testing tool gained support
4291 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
4292 and --seqpacket switches. It also has been extended to support both
4293 new-style and inetd-style file descriptor passing. Use the new
4294 --inetd switch to request inetd-style file descriptor passing.
4296 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
4297 variable, which takes a boolean value. If set to false, ANSI color
4298 output is disabled in the tools even when run on a terminal that
4301 * The VXLAN support in networkd now supports two new settings
4302 DestinationPort= and PortRange=.
4304 * A new systemd.machine_id= kernel command line switch has been added,
4305 that may be used to set the machine ID in /etc/machine-id if it is
4306 not initialized yet. This command line option has no effect if the
4307 file is already initialized.
4309 * systemd-nspawn gained a new --as-pid2 switch that invokes any
4310 specified command line as PID 2 rather than PID 1 in the
4311 container. In this mode PID 1 is a minimal stub init process that
4312 implements the special POSIX and Linux semantics of PID 1 regarding
4313 signal and child process management. Note that this stub init process
4314 is implemented in nspawn itself and requires no support from the
4315 container image. This new logic is useful to support running
4316 arbitrary commands in the container, as normal processes are
4317 generally not prepared to run as PID 1.
4319 * systemd-nspawn gained a new --chdir= switch for setting the current
4320 working directory for the process started in the container.
4322 * "journalctl /dev/sda" will now output all kernel log messages for
4323 specified device from the current boot, in addition to all devices
4324 that are parents of it. This should make log output about devices
4325 pretty useful, as long as kernel drivers attach enough metadata to
4326 the log messages. (The usual SATA drivers do.)
4328 * The sd-journal API gained two new calls
4329 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
4330 that report whether log data from /run or /var has been found.
4332 * journalctl gained a new switch "--fields" that prints all journal
4333 record field names currently in use in the journal. This is backed
4334 by two new sd-journal API calls sd_journal_enumerate_fields() and
4335 sd_journal_restart_fields().
4337 * Most configurable timeouts in systemd now expect an argument of
4338 "infinity" to turn them off, instead of "0" as before. The semantics
4339 from now on is that a timeout of "0" means "now", and "infinity"
4340 means "never". To maintain backwards compatibility, "0" continues to
4341 turn off previously existing timeout settings.
4343 * "systemctl reload-or-try-restart" has been renamed to "systemctl
4344 try-reload-or-restart" to clarify what it actually does: the "try"
4345 logic applies to both reloading and restarting, not just restarting.
4346 The old name continues to be accepted for compatibility.
4348 * On boot-up, when PID 1 detects that the system clock is behind the
4349 release date of the systemd version in use, the clock is now set
4350 to the latter. Previously, this was already done in timesyncd, in order
4351 to avoid running with clocks set to the various clock epochs such as
4352 1902, 1938 or 1970. With this change the logic is now done in PID 1
4353 in addition to timesyncd during early boot-up, so that it is enforced
4354 before the first process is spawned by systemd. Note that the logic
4355 in timesyncd remains, as it is more comprehensive and ensures
4356 clock monotonicity by maintaining a persistent timestamp file in
4357 /var. Since /var is generally not available in earliest boot or the
4358 initrd, this part of the logic remains in timesyncd, and is not done
4361 * Support for tweaking details in net_cls.class_id through the
4362 NetClass= configuration directive has been removed, as the kernel
4363 people have decided to deprecate that controller in cgroup v2.
4364 Userspace tools such as nftables are moving over to setting rules
4365 that are specific to the full cgroup path of a task, which obsoletes
4366 these controllers anyway. The NetClass= directive is kept around for
4367 legacy compatibility reasons. For a more in-depth description of the
4368 kernel change, please refer to the respective upstream commit:
4370 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
4372 * A new service setting RuntimeMaxSec= has been added that may be used
4373 to specify a maximum runtime for a service. If the timeout is hit, the
4374 service is terminated and put into a failure state.
4376 * A new service setting AmbientCapabilities= has been added. It allows
4377 configuration of additional Linux process capabilities that are
4378 passed to the activated processes. This is only available on very
4381 * The process resource limit settings in service units may now be used
4382 to configure hard and soft limits individually.
4384 * The various libsystemd APIs such as sd-bus or sd-event now publicly
4385 expose support for gcc's __attribute__((cleanup())) C extension.
4386 Specifically, for many object destructor functions alternative
4387 versions have been added that have names suffixed with "p" and take a
4388 pointer to a pointer to the object to destroy, instead of just a
4389 pointer to the object itself. This is useful because these destructor
4390 functions may be used directly as parameters to the cleanup
4391 construct. Internally, systemd has been a heavy user of this GCC
4392 extension for a long time, and with this change similar support is
4393 now available to consumers of the library outside of systemd. Note
4394 that by using this extension in your sources compatibility with old
4395 and strictly ANSI compatible C compilers is lost. However, all gcc or
4396 LLVM versions of recent years support this extension.
4398 * Timer units gained support for a new setting RandomizedDelaySec= that
4399 allows configuring some additional randomized delay to the configured
4400 time. This is useful to spread out timer events to avoid load peaks in
4401 clusters or larger setups.
4403 * Calendar time specifications now support sub-second accuracy.
4405 * Socket units now support listening on SCTP and UDP-lite protocol
4408 * The sd-event API now comes with a full set of man pages.
4410 * Older versions of systemd contained experimental support for
4411 compressing journal files and coredumps with the LZ4 compressor that
4412 was not compatible with the lz4 binary (due to API limitations of the
4413 lz4 library). This support has been removed; only support for files
4414 compatible with the lz4 binary remains. This LZ4 logic is now
4415 officially supported and no longer considered experimental.
4417 * The dkr image import logic has been removed again from importd. dkr's
4418 micro-services focus doesn't fit into the machine image focus of
4419 importd, and quickly got out of date with the upstream dkr API.
4421 * Creation of the /run/lock/lockdev/ directory was dropped from
4422 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
4423 been available for many years. If you still need this, you need to
4424 create your own tmpfiles.d config file with:
4426 d /run/lock/lockdev 0775 root lock -
4428 * The settings StartLimitBurst=, StartLimitInterval=, StartLimitAction=
4429 and RebootArgument= have been moved from the [Service] section of
4430 unit files to [Unit], and they are now supported on all unit types,
4431 not just service units. Of course, systemd will continue to
4432 understand these settings also at the old location, in order to
4433 maintain compatibility.
4435 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
4436 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
4437 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
4438 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
4439 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
4440 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
4441 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
4442 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
4443 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
4444 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
4445 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
4446 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
4447 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
4448 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
4449 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
4450 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
4451 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
4452 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
4453 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
4455 — Berlin, 2016-02-11
4459 * A number of properties previously only settable in unit
4460 files are now also available as properties to set when
4461 creating transient units programmatically via the bus, as it
4462 is exposed with systemd-run's --property=
4463 setting. Specifically, these are: SyslogIdentifier=,
4464 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
4465 EnvironmentFile=, ReadWriteDirectories=,
4466 ReadOnlyDirectories=, InaccessibleDirectories=,
4467 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
4469 * When creating transient services via the bus API it is now
4470 possible to pass in a set of file descriptors to use as
4471 STDIN/STDOUT/STDERR for the invoked process.
4473 * Slice units may now be created transiently via the bus APIs,
4474 similar to the way service and scope units may already be
4475 created transiently.
4477 * Wherever systemd expects a calendar timestamp specification
4478 (like in journalctl's --since= and --until= switches) UTC
4479 timestamps are now supported. Timestamps suffixed with "UTC"
4480 are now considered to be in Universal Time Coordinated
4481 instead of the local timezone. Also, timestamps may now
4482 optionally be specified with sub-second accuracy. Both of
4483 these additions also apply to recurring calendar event
4484 specification, such as OnCalendar= in timer units.
4486 * journalctl gained a new "--sync" switch that asks the
4487 journal daemon to write all so far unwritten log messages to
4488 disk and sync the files, before returning.
4490 * systemd-tmpfiles learned two new line types "q" and "Q" that
4491 operate like "v", but also set up a basic btrfs quota
4492 hierarchy when used on a btrfs file system with quota
4495 * tmpfiles' "v", "q" and "Q" will now create a plain directory
4496 instead of a subvolume (even on a btrfs file system) if the
4497 root directory is a plain directory, and not a
4498 subvolume. This should simplify things with certain chroot()
4499 environments which are not aware of the concept of btrfs
4502 * systemd-detect-virt gained a new --chroot switch to detect
4503 whether execution takes place in a chroot() environment.
4505 * CPUAffinity= now takes CPU index ranges in addition to
4508 * The various memory-related resource limit settings (such as
4509 LimitAS=) now understand the usual K, M, G, ... suffixes to
4510 the base of 1024 (IEC). Similar, the time-related resource
4511 limit settings understand the usual min, h, day, ...
4514 * There's a new system.conf setting DefaultTasksMax= to
4515 control the default TasksMax= setting for services and
4516 scopes running on the system. (TasksMax= is the primary
4517 setting that exposes the "pids" cgroup controller on systemd
4518 and was introduced in the previous systemd release.) The
4519 setting now defaults to 512, which means services that are
4520 not explicitly configured otherwise will only be able to
4521 create 512 processes or threads at maximum, from this
4522 version on. Note that this means that thread- or
4523 process-heavy services might need to be reconfigured to set
4524 TasksMax= to a higher value. It is sufficient to set
4525 TasksMax= in these specific unit files to a higher value, or
4526 even "infinity". Similar, there's now a logind.conf setting
4527 UserTasksMax= that defaults to 4096 and limits the total
4528 number of processes or tasks each user may own
4529 concurrently. nspawn containers also have the TasksMax=
4530 value set by default now, to 8192. Note that all of this
4531 only has an effect if the "pids" cgroup controller is
4532 enabled in the kernel. The general benefit of these changes
4533 should be a more robust and safer system, that provides a
4534 certain amount of per-service fork() bomb protection.
4536 * systemd-nspawn gained the new --network-veth-extra= switch
4537 to define additional and arbitrarily-named virtual Ethernet
4538 links between the host and the container.
4540 * A new service execution setting PassEnvironment= has been
4541 added that allows importing select environment variables
4542 from PID1's environment block into the environment block of
4545 * Timer units gained support for a new RemainAfterElapse=
4546 setting which takes a boolean argument. It defaults to on,
4547 exposing behaviour unchanged to previous releases. If set to
4548 off, timer units are unloaded after they elapsed if they
4549 cannot elapse again. This is particularly useful for
4550 transient timer units, which shall not stay around longer
4551 than until they first elapse.
4553 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
4554 default now (the kernel default is 16). This is beneficial
4555 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
4556 allows substantially larger numbers of queued
4557 datagrams. This should increase the capability of systemd to
4558 parallelize boot-up, as logging and sd_notify() are unlikely
4559 to stall execution anymore. If you need to change the value
4560 from the new defaults, use the usual sysctl.d/ snippets.
4562 * The compression framing format used by the journal or
4563 coredump processing has changed to be in line with what the
4564 official LZ4 tools generate. LZ4 compression support in
4565 systemd was considered unsupported previously, as the format
4566 was not compatible with the normal tools. With this release
4567 this has changed now, and it is hence safe for downstream
4568 distributions to turn it on. While not compressing as well
4569 as the XZ, LZ4 is substantially faster, which makes
4570 it a good default choice for the compression logic in the
4571 journal and in coredump handling.
4573 * Any reference to /etc/mtab has been dropped from
4574 systemd. The file has been obsolete since a while, but
4575 systemd refused to work on systems where it was incorrectly
4576 set up (it should be a symlink or non-existent). Please make
4577 sure to update to util-linux 2.27.1 or newer in conjunction
4578 with this systemd release, which also drops any reference to
4579 /etc/mtab. If you maintain a distribution make sure that no
4580 software you package still references it, as this is a
4581 likely source of bugs. There's also a glibc bug pending,
4582 asking for removal of any reference to this obsolete file:
4584 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
4586 Note that only util-linux versions built with
4587 --enable-libmount-force-mountinfo are supported.
4589 * Support for the ".snapshot" unit type has been removed. This
4590 feature turned out to be little useful and little used, and
4591 has now been removed from the core and from systemctl.
4593 * The dependency types RequiresOverridable= and
4594 RequisiteOverridable= have been removed from systemd. They
4595 have been used only very sparingly to our knowledge and
4596 other options that provide a similar effect (such as
4597 systemctl --mode=ignore-dependencies) are much more useful
4598 and commonly used. Moreover, they were only half-way
4599 implemented as the option to control behaviour regarding
4600 these dependencies was never added to systemctl. By removing
4601 these dependency types the execution engine becomes a bit
4602 simpler. Unit files that use these dependencies should be
4603 changed to use the non-Overridable dependency types
4604 instead. In fact, when parsing unit files with these
4605 options, that's what systemd will automatically convert them
4606 too, but it will also warn, asking users to fix the unit
4607 files accordingly. Removal of these dependency types should
4608 only affect a negligible number of unit files in the wild.
4610 * Behaviour of networkd's IPForward= option changed
4611 (again). It will no longer maintain a per-interface setting,
4612 but propagate one way from interfaces where this is enabled
4613 to the global kernel setting. The global setting will be
4614 enabled when requested by a network that is set up, but
4615 never be disabled again. This change was made to make sure
4616 IPv4 and IPv6 behaviour regarding packet forwarding is
4617 similar (as the Linux IPv6 stack does not support
4618 per-interface control of this setting) and to minimize
4621 * In unit files the behaviour of %u, %U, %h, %s has
4622 changed. These specifiers will now unconditionally resolve
4623 to the various user database fields of the user that the
4624 systemd instance is running as, instead of the user
4625 configured in the specific unit via User=. Note that this
4626 effectively doesn't change much, as resolving of these
4627 specifiers was already turned off in the --system instance
4628 of systemd, as we cannot do NSS lookups from PID 1. In the
4629 --user instance of systemd these specifiers where correctly
4630 resolved, but hardly made any sense, since the user instance
4631 lacks privileges to do user switches anyway, and User= is
4632 hence useless. Moreover, even in the --user instance of
4633 systemd behaviour was awkward as it would only take settings
4634 from User= assignment placed before the specifier into
4635 account. In order to unify and simplify the logic around
4636 this the specifiers will now always resolve to the
4637 credentials of the user invoking the manager (which in case
4638 of PID 1 is the root user).
4640 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
4641 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
4642 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
4643 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
4644 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
4645 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
4646 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
4647 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
4648 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
4649 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
4650 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
4651 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
4652 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
4653 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
4656 — Berlin, 2015-11-18
4660 * systemd now depends on util-linux v2.27. More specifically,
4661 the newly added mount monitor feature in libmount now
4662 replaces systemd's former own implementation.
4664 * libmount mandates /etc/mtab not to be regular file, and
4665 systemd now enforces this condition at early boot.
4666 /etc/mtab has been deprecated and warned about for a very
4667 long time, so systems running systemd should already have
4668 stopped having this file around as anything else than a
4669 symlink to /proc/self/mounts.
4671 * Support for the "pids" cgroup controller has been added. It
4672 allows accounting the number of tasks in a cgroup and
4673 enforcing limits on it. This adds two new setting
4674 TasksAccounting= and TasksMax= to each unit, as well as a
4675 global option DefaultTasksAccounting=.
4677 * Support for the "net_cls" cgroup controller has been added.
4678 It allows assigning a net class ID to each task in the
4679 cgroup, which can then be used in firewall rules and traffic
4680 shaping configurations. Note that the kernel netfilter net
4681 class code does not currently work reliably for ingress
4682 packets on unestablished sockets.
4684 This adds a new config directive called NetClass= to CGroup
4685 enabled units. Allowed values are positive numbers for fixed
4686 assignments and "auto" for picking a free value
4689 * 'systemctl is-system-running' now returns 'offline' if the
4690 system is not booted with systemd. This command can now be
4691 used as a substitute for 'systemd-notify --booted'.
4693 * Watchdog timeouts have been increased to 3 minutes for all
4694 in-tree service files. Apparently, disk IO issues are more
4695 frequent than we hoped, and user reported >1 minute waiting
4698 * 'machine-id-commit' functionality has been merged into
4699 'machine-id-setup --commit'. The separate binary has been
4702 * The WorkingDirectory= directive in unit files may now be set
4703 to the special value '~'. In this case, the working
4704 directory is set to the home directory of the user
4705 configured in User=.
4707 * "machinectl shell" will now open the shell in the home
4708 directory of the selected user by default.
4710 * The CrashChVT= configuration file setting is renamed to
4711 CrashChangeVT=, following our usual logic of not
4712 abbreviating unnecessarily. The old directive is still
4713 supported for compat reasons. Also, this directive now takes
4714 an integer value between 1 and 63, or a boolean value. The
4715 formerly supported '-1' value for disabling stays around for
4718 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
4719 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
4720 RootDirectory= properties can now be set for transient
4723 * The systemd-analyze tool gained a new "set-log-target" verb
4724 to change the logging target the system manager logs to
4725 dynamically during runtime. This is similar to how
4726 "systemd-analyze set-log-level" already changes the log
4729 * In nspawn /sys is now mounted as tmpfs, with only a selected
4730 set of subdirectories mounted in from the real sysfs. This
4731 enhances security slightly, and is useful for ensuring user
4732 namespaces work correctly.
4734 * Support for USB FunctionFS activation has been added. This
4735 allows implementation of USB gadget services that are
4736 activated as soon as they are requested, so that they don't
4737 have to run continuously, similar to classic socket
4740 * The "systemctl exit" command now optionally takes an
4741 additional parameter that sets the exit code to return from
4742 the systemd manager when exiting. This is only relevant when
4743 running the systemd user instance, or when running the
4744 system instance in a container.
4746 * sd-bus gained the new API calls sd_bus_path_encode_many()
4747 and sd_bus_path_decode_many() that allow easy encoding and
4748 decoding of multiple identifier strings inside a D-Bus
4749 object path. Another new call sd_bus_default_flush_close()
4750 has been added to flush and close per-thread default
4753 * systemd-cgtop gained support for a -M/--machine= switch to
4754 show the control groups within a certain container only.
4756 * "systemctl kill" gained support for an optional --fail
4757 switch. If specified the requested operation will fail of no
4758 processes have been killed, because the unit had no
4759 processes attached, or similar.
4761 * A new systemd.crash_reboot=1 kernel command line option has
4762 been added that triggers a reboot after crashing. This can
4763 also be set through CrashReboot= in systemd.conf.
4765 * The RuntimeDirectory= setting now understands unit
4766 specifiers like %i or %f.
4768 * A new (still internal) library API sd-ipv4acd has been added,
4769 that implements address conflict detection for IPv4. It's
4770 based on code from sd-ipv4ll, and will be useful for
4771 detecting DHCP address conflicts.
4773 * File descriptors passed during socket activation may now be
4774 named. A new API sd_listen_fds_with_names() is added to
4775 access the names. The default names may be overridden,
4776 either in the .socket file using the FileDescriptorName=
4777 parameter, or by passing FDNAME= when storing the file
4778 descriptors using sd_notify().
4780 * systemd-networkd gained support for:
4782 - Setting the IPv6 Router Advertisement settings via
4783 IPv6AcceptRouterAdvertisements= in .network files.
4785 - Configuring the HelloTimeSec=, MaxAgeSec= and
4786 ForwardDelaySec= bridge parameters in .netdev files.
4788 - Configuring PreferredSource= for static routes in
4791 * The "ask-password" framework used to query for LUKS harddisk
4792 passwords or SSL passwords during boot gained support for
4793 caching passwords in the kernel keyring, if it is
4794 available. This makes sure that the user only has to type in
4795 a passphrase once if there are multiple objects to unlock
4796 with the same one. Previously, such password caching was
4797 available only when Plymouth was used; this moves the
4798 caching logic into the systemd codebase itself. The
4799 "systemd-ask-password" utility gained a new --keyname=
4800 switch to control which kernel keyring key to use for
4801 caching a password in. This functionality is also useful for
4802 enabling display managers such as gdm to automatically
4803 unlock the user's GNOME keyring if its passphrase, the
4804 user's password and the harddisk password are the same, if
4805 gdm-autologin is used.
4807 * When downloading tar or raw images using "machinectl
4808 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
4809 file is now also downloaded, if it is available and stored
4810 next to the image file.
4812 * Units of type ".socket" gained a new boolean setting
4813 Writable= which is only useful in conjunction with
4814 ListenSpecial=. If true, enables opening the specified
4815 special file in O_RDWR mode rather than O_RDONLY mode.
4817 * systemd-rfkill has been reworked to become a singleton
4818 service that is activated through /dev/rfkill on each rfkill
4819 state change and saves the settings to disk. This way,
4820 systemd-rfkill is now compatible with devices that exist
4821 only intermittendly, and even restores state if the previous
4822 system shutdown was abrupt rather than clean.
4824 * The journal daemon gained support for vacuuming old journal
4825 files controlled by the number of files that shall remain,
4826 in addition to the already existing control by size and by
4827 date. This is useful as journal interleaving performance
4828 degrades with too many separate journal files, and allows
4829 putting an effective limit on them. The new setting defaults
4830 to 100, but this may be changed by setting SystemMaxFiles=
4831 and RuntimeMaxFiles= in journald.conf. Also, the
4832 "journalctl" tool gained the new --vacuum-files= switch to
4833 manually vacuum journal files to leave only the specified
4834 number of files in place.
4836 * udev will now create /dev/disk/by-path links for ATA devices
4837 on kernels where that is supported.
4839 * Galician, Serbian, Turkish and Korean translations were added.
4841 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
4842 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
4843 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
4844 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
4845 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
4846 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
4847 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
4848 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
4849 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
4850 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
4851 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
4852 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
4853 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
4854 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
4855 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
4856 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
4857 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
4858 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
4860 — Berlin, 2015-10-07
4864 * The DHCP implementation of systemd-networkd gained a set of
4867 - The DHCP server now supports emitting DNS and NTP
4868 information. It may be enabled and configured via
4869 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
4870 and NTP information is enabled, but no servers are
4871 configured, the corresponding uplink information (if there
4872 is any) is propagated.
4874 - Server and client now support transmission and reception
4875 of timezone information. It can be configured via the
4876 newly introduced network options UseTimezone=,
4877 EmitTimezone=, and Timezone=. Transmission of timezone
4878 information is enabled between host and containers by
4879 default now: the container will change its local timezone
4880 to what the host has set.
4882 - Lease timeouts can now be configured via
4883 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
4885 - The DHCP server improved on the stability of
4886 leases. Clients are more likely to get the same lease
4887 information back, even if the server loses state.
4889 - The DHCP server supports two new configuration options to
4890 control the lease address pool metrics, PoolOffset= and
4893 * The encapsulation limit of tunnels in systemd-networkd may
4894 now be configured via 'EncapsulationLimit='. It allows
4895 modifying the maximum additional levels of encapsulation
4896 that are permitted to be prepended to a packet.
4898 * systemd now supports the concept of user buses replacing
4899 session buses, if used with dbus-1.10 (and enabled via dbus
4900 --enable-user-session). It previously only supported this on
4901 kdbus-enabled systems, and this release expands this to
4902 'dbus-daemon' systems.
4904 * systemd-networkd now supports predictable interface names
4907 * systemd now optionally supports the new Linux kernel
4908 "unified" control group hierarchy. If enabled via the kernel
4909 command-line option 'systemd.unified_cgroup_hierarchy=1',
4910 systemd will try to mount the unified cgroup hierarchy
4911 directly on /sys/fs/cgroup. If not enabled, or not
4912 available, systemd will fall back to the legacy cgroup
4913 hierarchy setup, as before. Host system and containers can
4914 mix and match legacy and unified hierarchies as they
4915 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
4916 environment variable to individually select the hierarchy to
4917 use for executed containers. By default, nspawn will use the
4918 unified hierarchy for the containers if the host uses the
4919 unified hierarchy, and the legacy hierarchy otherwise.
4920 Please note that at this point the unified hierarchy is an
4921 experimental kernel feature and is likely to change in one
4922 of the next kernel releases. Therefore, it should not be
4923 enabled by default in downstream distributions yet. The
4924 minimum required kernel version for the unified hierarchy to
4925 work is 4.2. Note that when the unified hierarchy is used
4926 for the first time delegated access to controllers is
4927 safe. Because of this systemd-nspawn containers will get
4928 access to controllers now, as will systemd user
4929 sessions. This means containers and user sessions may now
4930 manage their own resources, partitioning up what the system
4933 * A new special scope unit "init.scope" has been introduced
4934 that encapsulates PID 1 of the system. It may be used to
4935 determine resource usage and enforce resource limits on PID
4936 1 itself. PID 1 hence moved out of the root of the control
4939 * The cgtop tool gained support for filtering out kernel
4940 threads when counting tasks in a control group. Also, the
4941 count of processes is now recursively summed up by
4942 default. Two options -k and --recursive= have been added to
4943 revert to old behaviour. The tool has also been updated to
4944 work correctly in containers now.
4946 * systemd-nspawn's --bind= and --bind-ro= options have been
4947 extended to allow creation of non-recursive bind mounts.
4949 * libsystemd gained two new calls sd_pid_get_cgroup() and
4950 sd_peer_get_cgroup() which return the control group path of
4951 a process or peer of a connected AF_UNIX socket. This
4952 function call is particularly useful when implementing
4953 delegated subtrees support in the control group hierarchy.
4955 * The "sd-event" event loop API of libsystemd now supports
4956 correct dequeuing of real-time signals, without losing
4959 * When systemd requests a polkit decision when managing units it
4960 will now add additional fields to the request, including unit
4961 name and desired operation. This enables more powerful polkit
4962 policies, that make decisions depending on these parameters.
4964 * nspawn learnt support for .nspawn settings files, that may
4965 accompany the image files or directories of containers, and
4966 may contain additional settings for the container. This is
4967 an alternative to configuring container parameters via the
4968 nspawn command line.
4970 Contributions from: Cristian Rodríguez, Daniel Mack, David
4971 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
4972 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
4973 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
4974 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
4975 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
4976 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
4977 Andersen, Tom Gundersen, Torstein Husebø
4979 — Berlin, 2015-09-08
4983 * machinectl gained a new verb 'shell' which opens a fresh
4984 shell on the target container or the host. It is similar to
4985 the existing 'login' command of machinectl, but spawns the
4986 shell directly without prompting for username or
4987 password. The pseudo machine '.host' now refers to the local
4988 host and is used by default. Hence, 'machinectl shell' can
4989 be used as replacement for 'su -' which spawns a session as
4990 a fresh systemd unit in a way that is fully isolated from
4991 the originating session.
4993 * systemd-networkd learned to cope with private-zone DHCP
4994 options and allows other programs to query the values.
4996 * SELinux access control when enabling/disabling units is no
4997 longer enforced with this release. The previous implementation
4998 was incorrect, and a new corrected implementation is not yet
4999 available. As unit file operations are still protected via
5000 polkit and D-Bus policy this is not a security problem. Yet,
5001 distributions which care about optimal SELinux support should
5002 probably not stabilize on this release.
5004 * sd-bus gained support for matches of type "arg0has=", that
5005 test for membership of strings in string arrays sent in bus
5008 * systemd-resolved now dumps the contents of its DNS and LLMNR
5009 caches to the logs on reception of the SIGUSR1 signal. This
5010 is useful to debug DNS behaviour.
5012 * The coredumpctl tool gained a new --directory= option to
5013 operate on journal files in a specific directory.
5015 * "systemctl reboot" and related commands gained a new
5016 "--message=" option which may be used to set a free-text
5017 wall message when shutting down or rebooting the
5018 system. This message is also logged, which is useful for
5019 figuring out the reason for a reboot or shutdown a
5022 * The "systemd-resolve-host" tool's -i switch now takes
5023 network interface numbers as alternative to interface names.
5025 * A new unit file setting for services has been introduced:
5026 UtmpMode= allows configuration of how precisely systemd
5027 handles utmp and wtmp entries for the service if this is
5028 enabled. This allows writing services that appear similar to
5029 user sessions in the output of the "w", "who", "last" and
5032 * systemd-resolved will now locally synthesize DNS resource
5033 records for the "localhost" and "gateway" domains as well as
5034 the local hostname. This should ensure that clients querying
5035 RRs via resolved will get similar results as those going via
5036 NSS, if nss-myhostname is enabled.
5038 Contributions from: Alastair Hughes, Alex Crawford, Daniel
5039 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
5040 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
5041 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
5042 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
5043 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
5044 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
5045 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
5046 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
5047 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
5048 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
5049 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
5051 — Berlin, 2015-08-27
5055 * The systemd-efi-boot-generator functionality was merged into
5056 systemd-gpt-auto-generator.
5058 * systemd-networkd now supports Group Policy for vxlan
5059 devices. It can be enabled via the new boolean configuration
5060 option called 'GroupPolicyExtension='.
5062 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
5063 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
5064 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
5066 — Berlin, 2015-07-31
5070 * The python-systemd code has been removed from the systemd repository.
5071 A new repository has been created which accommodates the code from
5072 now on, and we kindly ask distributions to create a separate package
5073 for this: https://github.com/systemd/python-systemd
5075 * The systemd daemon will now reload its main configuration
5076 (/etc/systemd/system.conf) on daemon-reload.
5078 * sd-dhcp now exposes vendor specific extensions via
5079 sd_dhcp_lease_get_vendor_specific().
5081 * systemd-networkd gained a number of new configuration options.
5083 - A new boolean configuration option for TAP devices called
5084 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
5085 device, thus allowing to send and receive GSO packets.
5087 - A new tunnel configuration option called 'CopyDSCP='.
5088 If enabled, the DSCP field of ip6 tunnels is copied into the
5089 decapsulated packet.
5091 - A set of boolean bridge configuration options were added.
5092 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
5093 and 'UnicastFlood=' are now parsed by networkd and applied to the
5094 respective bridge link device via the respective IFLA_BRPORT_*
5097 - A new string configuration option to override the hostname sent
5098 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
5099 is true, networkd will use the configured hostname instead of the
5100 system hostname when sending DHCP requests.
5102 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
5103 networkd will configure the IPv6 flow-label of the tunnel device
5104 according to RFC2460.
5106 - The 'macvtap' virtual network devices are now supported, similar to
5107 the already supported 'macvlan' devices.
5109 * systemd-resolved now implements RFC5452 to improve resilience against
5110 cache poisoning. Additionally, source port randomization is enabled
5111 by default to further protect against DNS spoofing attacks.
5113 * nss-mymachines now supports translating UIDs and GIDs of running
5114 containers with user-namespaces enabled. If a container 'foo'
5115 translates a host uid 'UID' to the container uid 'TUID', then
5116 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
5117 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
5118 mapped as 'vg-foo-TGID'.
5120 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
5121 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
5122 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
5123 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
5124 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
5125 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
5126 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
5127 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
5128 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
5129 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
5131 — Berlin, 2015-07-29
5135 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
5136 There are no known issues with current sysfs, and udev does not need
5137 or should be used to work around such bugs.
5139 * udev does no longer enable USB HID power management. Several reports
5140 indicate, that some devices cannot handle that setting.
5142 * The udev accelerometer helper was removed. The functionality
5143 is now fully included in iio-sensor-proxy. But this means,
5144 older iio-sensor-proxy versions will no longer provide
5145 accelerometer/orientation data with this systemd version.
5146 Please upgrade iio-sensor-proxy to version 1.0.
5148 * networkd gained a new configuration option IPv6PrivacyExtensions=
5149 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
5150 for Stateless Address") on selected networks.
5152 * For the sake of fewer build-time dependencies and less code in the
5153 main repository, the python bindings are about to be removed in the
5154 next release. A new repository has been created which accommodates
5155 the code from now on, and we kindly ask distributions to create a
5156 separate package for this. The removal will take place in v223.
5158 https://github.com/systemd/python-systemd
5160 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
5161 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
5162 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
5163 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
5164 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
5165 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
5166 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
5167 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
5168 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
5169 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
5171 — Berlin, 2015-07-07
5175 * The sd-bus.h and sd-event.h APIs have now been declared
5176 stable and have been added to the official interface of
5177 libsystemd.so. sd-bus implements an alternative D-Bus client
5178 library, that is relatively easy to use, very efficient and
5179 supports both classic D-Bus as well as kdbus as transport
5180 backend. sd-event is a generic event loop abstraction that
5181 is built around Linux epoll, but adds features such as event
5182 prioritization or efficient timer handling. Both APIs are good
5183 choices for C programs looking for a bus and/or event loop
5184 implementation that is minimal and does not have to be
5185 portable to other kernels.
5187 * kdbus support is no longer compile-time optional. It is now
5188 always built-in. However, it can still be disabled at
5189 runtime using the kdbus=0 kernel command line setting, and
5190 that setting may be changed to default to off, by specifying
5191 --disable-kdbus at build-time. Note though that the kernel
5192 command line setting has no effect if the kdbus.ko kernel
5193 module is not installed, in which case kdbus is (obviously)
5194 also disabled. We encourage all downstream distributions to
5195 begin testing kdbus by adding it to the kernel images in the
5196 development distributions, and leaving kdbus support in
5199 * The minimal required util-linux version has been bumped to
5202 * Support for chkconfig (--enable-chkconfig) was removed in
5203 favor of calling an abstraction tool
5204 /lib/systemd/systemd-sysv-install. This needs to be
5205 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
5206 in README for details.
5208 * If there's a systemd unit and a SysV init script for the
5209 same service name, and the user executes "systemctl enable"
5210 for it (or a related call), then this will now enable both
5211 (or execute the related operation on both), not just the
5214 * The libudev API documentation has been converted from gtkdoc
5217 * gudev has been removed from the systemd tree, it is now an
5220 * The systemd-cgtop tool learnt a new --raw switch to generate
5221 "raw" (machine parsable) output.
5223 * networkd's IPForwarding= .network file setting learnt the
5224 new setting "kernel", which ensures that networkd does not
5225 change the IP forwarding sysctl from the default kernel
5228 * The systemd-logind bus API now exposes a new boolean
5229 property "Docked" that reports whether logind considers the
5230 system "docked", i.e. connected to a docking station or not.
5232 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
5233 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
5234 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
5235 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
5236 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
5237 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
5238 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
5239 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
5240 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
5241 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
5242 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
5243 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
5244 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
5245 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
5246 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
5247 Fink, Zbigniew Jędrzejewski-Szmek
5249 — Berlin, 2015-06-19
5253 * The gudev library has been extracted into a separate repository
5254 available at: https://git.gnome.org/browse/libgudev/
5255 It is now managed as part of the Gnome project. Distributions
5256 are recommended to pass --disable-gudev to systemd and use
5257 gudev from the Gnome project instead. gudev is still included
5258 in systemd, for now. It will be removed soon, though. Please
5259 also see the announcement-thread on systemd-devel:
5260 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
5262 * systemd now exposes a CPUUsageNSec= property for each
5263 service unit on the bus, that contains the overall consumed
5264 CPU time of a service (the sum of what each process of the
5265 service consumed). This value is only available if
5266 CPUAccounting= is turned on for a service, and is then shown
5267 in the "systemctl status" output.
5269 * Support for configuring alternative mappings of the old SysV
5270 runlevels to systemd targets has been removed. They are now
5271 hardcoded in a way that runlevels 2, 3, 4 all map to
5272 multi-user.target and 5 to graphical.target (which
5273 previously was already the default behaviour).
5275 * The auto-mounter logic gained support for mount point
5276 expiry, using a new TimeoutIdleSec= setting in .automount
5277 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
5279 * The EFI System Partition (ESP) as mounted to /boot by
5280 systemd-efi-boot-generator will now be unmounted
5281 automatically after 2 minutes of not being used. This should
5282 minimize the risk of ESP corruptions.
5284 * New /etc/fstab options x-systemd.requires= and
5285 x-systemd.requires-mounts-for= are now supported to express
5286 additional dependencies for mounts. This is useful for
5287 journalling file systems that support external journal
5288 devices or overlay file systems that require underlying file
5289 systems to be mounted.
5291 * systemd does not support direct live-upgrades (via systemctl
5292 daemon-reexec) from versions older than v44 anymore. As no
5293 distribution we are aware of shipped such old versions in a
5294 stable release this should not be problematic.
5296 * When systemd forks off a new per-connection service instance
5297 it will now set the $REMOTE_ADDR environment variable to the
5298 remote IP address, and $REMOTE_PORT environment variable to
5299 the remote IP port. This behaviour is similar to the
5300 corresponding environment variables defined by CGI.
5302 * systemd-networkd gained support for uplink failure
5303 detection. The BindCarrier= option allows binding interface
5304 configuration dynamically to the link sense of other
5305 interfaces. This is useful to achieve behaviour like in
5308 * systemd-networkd gained support for configuring the DHCP
5309 client identifier to use when requesting leases.
5311 * systemd-networkd now has a per-network UseNTP= option to
5312 configure whether NTP server information acquired via DHCP
5313 is passed on to services like systemd-timesyncd.
5315 * systemd-networkd gained support for vti6 tunnels.
5317 * Note that systemd-networkd manages the sysctl variable
5318 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
5319 it is configured for since v219. The variable controls IP
5320 forwarding, and is a per-interface alternative to the global
5321 /proc/sys/net/ipv[46]/ip_forward. This setting is
5322 configurable in the IPForward= option, which defaults to
5323 "no". This means if networkd is used for an interface it is
5324 no longer sufficient to set the global sysctl option to turn
5325 on IP forwarding! Instead, the .network file option
5326 IPForward= needs to be turned on! Note that the
5327 implementation of this behaviour was broken in v219 and has
5330 * Many bonding and vxlan options are now configurable in
5333 * systemd-nspawn gained a new --property= setting to set unit
5334 properties for the container scope. This is useful for
5335 setting resource parameters (e.g. "CPUShares=500") on
5336 containers started from the command line.
5338 * systemd-nspawn gained a new --private-users= switch to make
5339 use of user namespacing available on recent Linux kernels.
5341 * systemd-nspawn may now be called as part of a shell pipeline
5342 in which case the pipes used for stdin and stdout are passed
5343 directly to the process invoked in the container, without
5344 indirection via a pseudo tty.
5346 * systemd-nspawn gained a new switch to control the UNIX
5347 signal to use when killing the init process of the container
5350 * systemd-nspawn gained a new --overlay= switch for mounting
5351 overlay file systems into the container using the new kernel
5354 * When a container image is imported via systemd-importd and
5355 the host file system is not btrfs, a loopback block device
5356 file is created in /var/lib/machines.raw with a btrfs file
5357 system inside. It is then mounted to /var/lib/machines to
5358 enable btrfs features for container management. The loopback
5359 file and btrfs file system is grown as needed when container
5360 images are imported via systemd-importd.
5362 * systemd-machined/systemd-importd gained support for btrfs
5363 quota, to enforce container disk space limits on disk. This
5364 is exposed in "machinectl set-limit".
5366 * systemd-importd now can import containers from local .tar,
5367 .raw and .qcow2 images, and export them to .tar and .raw. It
5368 can also import dkr v2 images now from the network (on top
5371 * systemd-importd gained support for verifying downloaded
5372 images with gpg2 (previously only gpg1 was supported).
5374 * systemd-machined, systemd-logind, systemd: most bus calls are
5375 now accessible to unprivileged processes via polkit. Also,
5376 systemd-logind will now allow users to kill their own sessions
5377 without further privileges or authorization.
5379 * systemd-shutdownd has been removed. This service was
5380 previously responsible for implementing scheduled shutdowns
5381 as exposed in /usr/bin/shutdown's time parameter. This
5382 functionality has now been moved into systemd-logind and is
5383 accessible via a bus interface.
5385 * "systemctl reboot" gained a new switch --firmware-setup that
5386 can be used to reboot into the EFI firmware setup, if that
5387 is available. systemd-logind now exposes an API on the bus
5388 to trigger such reboots, in case graphical desktop UIs want
5389 to cover this functionality.
5391 * "systemctl enable", "systemctl disable" and "systemctl mask"
5392 now support a new "--now" switch. If specified the units
5393 that are enabled will also be started, and the ones
5394 disabled/masked also stopped.
5396 * The Gummiboot EFI boot loader tool has been merged into
5397 systemd, and renamed to "systemd-boot". The bootctl tool has been
5398 updated to support systemd-boot.
5400 * An EFI kernel stub has been added that may be used to create
5401 kernel EFI binaries that contain not only the actual kernel,
5402 but also an initrd, boot splash, command line and OS release
5403 information. This combined binary can then be signed as a
5404 single image, so that the firmware can verify it all in one
5405 step. systemd-boot has special support for EFI binaries created
5406 like this and can extract OS release information from them
5407 and show them in the boot menu. This functionality is useful
5408 to implement cryptographically verified boot schemes.
5410 * Optional support has been added to systemd-fsck to pass
5411 fsck's progress report to an AF_UNIX socket in the file
5414 * udev will no longer create device symlinks for all block
5415 devices by default. A blacklist for excluding special block
5416 devices from this logic has been turned into a whitelist
5417 that requires picking block devices explicitly that require
5420 * A new (currently still internal) API sd-device.h has been
5421 added to libsystemd. This modernized API is supposed to
5422 replace libudev eventually. In fact, already much of libudev
5423 is now just a wrapper around sd-device.h.
5425 * A new hwdb database for storing metadata about pointing
5426 stick devices has been added.
5428 * systemd-tmpfiles gained support for setting file attributes
5429 similar to the "chattr" tool with new 'h' and 'H' lines.
5431 * systemd-journald will no longer unconditionally set the
5432 btrfs NOCOW flag on new journal files. This is instead done
5433 with tmpfiles snippet using the new 'h' line type. This
5434 allows easy disabling of this logic, by masking the
5435 journal-nocow.conf tmpfiles file.
5437 * systemd-journald will now translate audit message types to
5438 human readable identifiers when writing them to the
5439 journal. This should improve readability of audit messages.
5441 * The LUKS logic gained support for the offset= and skip=
5442 options in /etc/crypttab, as previously implemented by
5445 * /usr/lib/os-release gained a new optional field VARIANT= for
5446 distributions that support multiple variants (such as a
5447 desktop edition, a server edition, ...)
5449 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
5450 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
5451 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
5452 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
5453 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
5454 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
5455 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
5456 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
5457 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
5458 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
5459 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
5460 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
5461 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
5462 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
5463 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
5464 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
5465 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
5466 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
5467 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
5468 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
5469 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
5470 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
5471 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
5472 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
5473 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
5474 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
5475 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
5477 — Berlin, 2015-05-22
5481 * Introduce a new API "sd-hwdb.h" for querying the hardware
5482 metadata database. With this minimal interface one can query
5483 and enumerate the udev hwdb, decoupled from the old libudev
5484 library. libudev's interface for this is now only a wrapper
5485 around sd-hwdb. A new tool systemd-hwdb has been added to
5486 interface with and update the database.
5488 * When any of systemd's tools copies files (for example due to
5489 tmpfiles' C lines) a btrfs reflink will attempted first,
5490 before bytewise copying is done.
5492 * systemd-nspawn gained a new --ephemeral switch. When
5493 specified a btrfs snapshot is taken of the container's root
5494 directory, and immediately removed when the container
5495 terminates again. Thus, a container can be started whose
5496 changes never alter the container's root directory, and are
5497 lost on container termination. This switch can also be used
5498 for starting a container off the root file system of the
5499 host without affecting the host OS. This switch is only
5500 available on btrfs file systems.
5502 * systemd-nspawn gained a new --template= switch. It takes the
5503 path to a container tree to use as template for the tree
5504 specified via --directory=, should that directory be
5505 missing. This allows instantiating containers dynamically,
5506 on first run. This switch is only available on btrfs file
5509 * When a .mount unit refers to a mount point on which multiple
5510 mounts are stacked, and the .mount unit is stopped all of
5511 the stacked mount points will now be unmounted until no
5512 mount point remains.
5514 * systemd now has an explicit notion of supported and
5515 unsupported unit types. Jobs enqueued for unsupported unit
5516 types will now fail with an "unsupported" error code. More
5517 specifically .swap, .automount and .device units are not
5518 supported in containers, .busname units are not supported on
5519 non-kdbus systems. .swap and .automount are also not
5520 supported if their respective kernel compile time options
5523 * machinectl gained support for two new "copy-from" and
5524 "copy-to" commands for copying files from a running
5525 container to the host or vice versa.
5527 * machinectl gained support for a new "bind" command to bind
5528 mount host directories into local containers. This is
5529 currently only supported for nspawn containers.
5531 * networkd gained support for configuring bridge forwarding
5532 database entries (fdb) from .network files.
5534 * A new tiny daemon "systemd-importd" has been added that can
5535 download container images in tar, raw, qcow2 or dkr formats,
5536 and make them available locally in /var/lib/machines, so
5537 that they can run as nspawn containers. The daemon can GPG
5538 verify the downloads (not supported for dkr, since it has no
5539 provisions for verifying downloads). It will transparently
5540 decompress bz2, xz, gzip compressed downloads if necessary,
5541 and restore sparse files on disk. The daemon uses privilege
5542 separation to ensure the actual download logic runs with
5543 fewer privileges than the daemon itself. machinectl has
5544 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
5545 make the functionality of importd available to the
5546 user. With this in place the Fedora and Ubuntu "Cloud"
5547 images can be downloaded and booted as containers unmodified
5548 (the Fedora images lack the appropriate GPG signature files
5549 currently, so they cannot be verified, but this will change
5550 soon, hopefully). Note that downloading images is currently
5551 only fully supported on btrfs.
5553 * machinectl is now able to list container images found in
5554 /var/lib/machines, along with some metadata about sizes of
5555 disk and similar. If the directory is located on btrfs and
5556 quota is enabled, this includes quota display. A new command
5557 "image-status" has been added that shows additional
5558 information about images.
5560 * machinectl is now able to clone container images
5561 efficiently, if the underlying file system (btrfs) supports
5562 it, with the new "machinectl clone" command. It also
5563 gained commands for renaming and removing images, as well as
5564 marking them read-only or read-write (supported also on
5565 legacy file systems).
5567 * networkd gained support for collecting LLDP network
5568 announcements, from hardware that supports this. This is
5569 shown in networkctl output.
5571 * systemd-run gained support for a new -t (--pty) switch for
5572 invoking a binary on a pty whose input and output is
5573 connected to the invoking terminal. This allows executing
5574 processes as system services while interactively
5575 communicating with them via the terminal. Most interestingly
5576 this is supported across container boundaries. Invoking
5577 "systemd-run -t /bin/bash" is an alternative to running a
5578 full login session, the difference being that the former
5579 will not register a session, nor go through the PAM session
5582 * tmpfiles gained support for a new "v" line type for creating
5583 btrfs subvolumes. If the underlying file system is a legacy
5584 file system, this automatically degrades to creating a
5585 normal directory. Among others /var/lib/machines is now
5586 created like this at boot, should it be missing.
5588 * The directory /var/lib/containers/ has been deprecated and
5589 been replaced by /var/lib/machines. The term "machines" has
5590 been used in the systemd context as generic term for both
5591 VMs and containers, and hence appears more appropriate for
5592 this, as the directory can also contain raw images bootable
5595 * systemd-nspawn when invoked with -M but without --directory=
5596 or --image= is now capable of searching for the container
5597 root directory, subvolume or disk image automatically, in
5598 /var/lib/machines. systemd-nspawn@.service has been updated
5599 to make use of this, thus allowing it to be used for raw
5602 * A new machines.target unit has been introduced that is
5603 supposed to group all containers/VMs invoked as services on
5604 the system. systemd-nspawn@.service has been updated to
5605 integrate with that.
5607 * machinectl gained a new "start" command, for invoking a
5608 container as a service. "machinectl start foo" is mostly
5609 equivalent to "systemctl start systemd-nspawn@foo.service",
5610 but handles escaping in a nicer way.
5612 * systemd-nspawn will now mount most of the cgroupfs tree
5613 read-only into each container, with the exception of the
5614 container's own subtree in the name=systemd hierarchy.
5616 * journald now sets the special FS_NOCOW file flag for its
5617 journal files. This should improve performance on btrfs, by
5618 avoiding heavy fragmentation when journald's write-pattern
5619 is used on COW file systems. It degrades btrfs' data
5620 integrity guarantees for the files to the same levels as for
5621 ext3/ext4 however. This should be OK though as journald does
5622 its own data integrity checks and all its objects are
5623 checksummed on disk. Also, journald should handle btrfs disk
5624 full events a lot more gracefully now, by processing SIGBUS
5625 errors, and not relying on fallocate() anymore.
5627 * When journald detects that journal files it is writing to
5628 have been deleted it will immediately start new journal
5631 * systemd now provides a way to store file descriptors
5632 per-service in PID 1. This is useful for daemons to ensure
5633 that fds they require are not lost during a daemon
5634 restart. The fds are passed to the daemon on the next
5635 invocation in the same way socket activation fds are
5636 passed. This is now used by journald to ensure that the
5637 various sockets connected to all the system's stdout/stderr
5638 are not lost when journald is restarted. File descriptors
5639 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
5640 an extension to sd_notify(). Note that a limit is enforced
5641 on the number of fds a service can store in PID 1, and it
5642 defaults to 0, so that no fds may be stored, unless this is
5643 explicitly turned on.
5645 * The default TERM variable to use for units connected to a
5646 terminal, when no other value is explicitly is set is now
5647 vt220 rather than vt102. This should be fairly safe still,
5648 but allows PgUp/PgDn work.
5650 * The /etc/crypttab option header= as known from Debian is now
5653 * "loginctl user-status" and "loginctl session-status" will
5654 now show the last 10 lines of log messages of the
5655 user/session following the status output. Similar,
5656 "machinectl status" will show the last 10 log lines
5657 associated with a virtual machine or container
5658 service. (Note that this is usually not the log messages
5659 done in the VM/container itself, but simply what the
5660 container manager logs. For nspawn this includes all console
5663 * "loginctl session-status" without further argument will now
5664 show the status of the session of the caller. Similar,
5665 "lock-session", "unlock-session", "activate",
5666 "enable-linger", "disable-linger" may now be called without
5667 session/user parameter in which case they apply to the
5668 caller's session/user.
5670 * An X11 session scriptlet is now shipped that uploads
5671 $DISPLAY and $XAUTHORITY into the environment of the systemd
5672 --user daemon if a session begins. This should improve
5673 compatibility with X11 enabled applications run as systemd
5676 * Generators are now subject to masking via /etc and /run, the
5677 same way as unit files.
5679 * networkd .network files gained support for configuring
5680 per-link IPv4/IPv6 packet forwarding as well as IPv4
5681 masquerading. This is by default turned on for veth links to
5682 containers, as registered by systemd-nspawn. This means that
5683 nspawn containers run with --network-veth will now get
5684 automatic routed access to the host's networks without any
5685 further configuration or setup, as long as networkd runs on
5688 * systemd-nspawn gained the --port= (-p) switch to expose TCP
5689 or UDP posts of a container on the host. With this in place
5690 it is possible to run containers with private veth links
5691 (--network-veth), and have their functionality exposed on
5692 the host as if their services were running directly on the
5695 * systemd-nspawn's --network-veth switch now gained a short
5696 version "-n", since with the changes above it is now truly
5697 useful out-of-the-box. The systemd-nspawn@.service has been
5698 updated to make use of it too by default.
5700 * systemd-nspawn will now maintain a per-image R/W lock, to
5701 ensure that the same image is not started more than once
5702 writable. (It's OK to run an image multiple times
5703 simultaneously in read-only mode.)
5705 * systemd-nspawn's --image= option is now capable of
5706 dissecting and booting MBR and GPT disk images that contain
5707 only a single active Linux partition. Previously it
5708 supported only GPT disk images with proper GPT type
5709 IDs. This allows running cloud images from major
5710 distributions directly with systemd-nspawn, without
5713 * In addition to collecting mouse dpi data in the udev
5714 hardware database, there's now support for collecting angle
5715 information for mouse scroll wheels. The database is
5716 supposed to guarantee similar scrolling behavior on mice
5717 that it knows about. There's also support for collecting
5718 information about Touchpad types.
5720 * udev's input_id built-in will now also collect touch screen
5721 dimension data and attach it to probed devices.
5723 * /etc/os-release gained support for a Distribution Privacy
5726 * networkd gained support for creating "ipvlan", "gretap",
5727 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
5729 * systemd-tmpfiles gained support for "a" lines for setting
5732 * systemd-nspawn will now mount /tmp in the container to
5733 tmpfs, automatically.
5735 * systemd now exposes the memory.usage_in_bytes cgroup
5736 attribute and shows it for each service in the "systemctl
5737 status" output, if available.
5739 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
5740 immediate reboot is triggered. This useful if shutdown is
5741 hung and is unable to complete, to expedite the
5742 operation. Note that this kind of reboot will still unmount
5743 all file systems, and hence should not result in fsck being
5746 * A .device unit for an optical block device will now be
5747 considered active only when a medium is in the drive. Also,
5748 mount units are now bound to their backing devices thus
5749 triggering automatic unmounting when devices become
5750 unavailable. With this in place systemd will now
5751 automatically unmount left-over mounts when a CD-ROM is
5752 ejected or an USB stick is yanked from the system.
5754 * networkd-wait-online now has support for waiting for
5755 specific interfaces only (with globbing), and for giving up
5756 after a configurable timeout.
5758 * networkd now exits when idle. It will be automatically
5759 restarted as soon as interfaces show up, are removed or
5760 change state. networkd will stay around as long as there is
5761 at least one DHCP state machine or similar around, that keep
5764 * networkd may now configure IPv6 link-local addressing in
5765 addition to IPv4 link-local addressing.
5767 * The IPv6 "token" for use in SLAAC may now be configured for
5768 each .network interface in networkd.
5770 * Routes configured with networkd may now be assigned a scope
5773 * networkd's [Match] sections now support globbing and lists
5774 of multiple space-separated matches per item.
5776 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
5777 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
5778 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
5779 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
5780 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
5781 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
5782 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
5783 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
5784 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
5785 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
5786 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
5787 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
5788 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
5789 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
5790 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
5791 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
5792 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
5793 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
5794 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
5795 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
5796 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
5797 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
5798 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
5799 Hoffmann, Zbigniew Jędrzejewski-Szmek
5801 — Berlin, 2015-02-16
5805 * When querying unit file enablement status (for example via
5806 "systemctl is-enabled"), a new state "indirect" is now known
5807 which indicates that a unit might not be enabled itself, but
5808 another unit listed in its Also= setting might be.
5810 * Similar to the various existing ConditionXYZ= settings for
5811 units, there are now matching AssertXYZ= settings. While
5812 failing conditions cause a unit to be skipped, but its job
5813 to succeed, failing assertions declared like this will cause
5814 a unit start operation and its job to fail.
5816 * hostnamed now knows a new chassis type "embedded".
5818 * systemctl gained a new "edit" command. When used on a unit
5819 file, this allows extending unit files with .d/ drop-in
5820 configuration snippets or editing the full file (after
5821 copying it from /usr/lib to /etc). This will invoke the
5822 user's editor (as configured with $EDITOR), and reload the
5823 modified configuration after editing.
5825 * "systemctl status" now shows the suggested enablement state
5826 for a unit, as declared in the (usually vendor-supplied)
5827 system preset files.
5829 * nss-myhostname will now resolve the single-label host name
5830 "gateway" to the locally configured default IP routing
5831 gateways, ordered by their metrics. This assigns a stable
5832 name to the used gateways, regardless which ones are
5833 currently configured. Note that the name will only be
5834 resolved after all other name sources (if nss-myhostname is
5835 configured properly) and should hence not negatively impact
5836 systems that use the single-label host name "gateway" in
5839 * systemd-inhibit now allows filtering by mode when listing
5842 * Scope and service units gained a new "Delegate" boolean
5843 property, which, when set, allows processes running inside the
5844 unit to further partition resources. This is primarily
5845 useful for systemd user instances as well as container
5848 * journald will now pick up audit messages directly from
5849 the kernel, and log them like any other log message. The
5850 audit fields are split up and fully indexed. This means that
5851 journalctl in many ways is now a (nicer!) alternative to
5852 ausearch, the traditional audit client. Note that this
5853 implements only a minimal audit client. If you want the
5854 special audit modes like reboot-on-log-overflow, please use
5855 the traditional auditd instead, which can be used in
5856 parallel to journald.
5858 * The ConditionSecurity= unit file option now understands the
5859 special string "audit" to check whether auditing is
5862 * journalctl gained two new commands --vacuum-size= and
5863 --vacuum-time= to delete old journal files until the
5864 remaining ones take up no more than the specified size on disk,
5865 or are not older than the specified time.
5867 * A new, native PPPoE library has been added to sd-network,
5868 systemd's library of light-weight networking protocols. This
5869 library will be used in a future version of networkd to
5870 enable PPPoE communication without an external pppd daemon.
5872 * The busctl tool now understands a new "capture" verb that
5873 works similar to "monitor", but writes a packet capture
5874 trace to STDOUT that can be redirected to a file which is
5875 compatible with libcap's capture file format. This can then
5876 be loaded in Wireshark and similar tools to inspect bus
5879 * The busctl tool now understands a new "tree" verb that shows
5880 the object trees of a specific service on the bus, or of all
5883 * The busctl tool now understands a new "introspect" verb that
5884 shows all interfaces and members of objects on the bus,
5885 including their signature and values. This is particularly
5886 useful to get more information about bus objects shown by
5887 the new "busctl tree" command.
5889 * The busctl tool now understands new verbs "call",
5890 "set-property" and "get-property" for invoking bus method
5891 calls, setting and getting bus object properties in a
5894 * busctl gained a new --augment-creds= argument that controls
5895 whether the tool shall augment credential information it
5896 gets from the bus with data from /proc, in a possibly
5899 * nspawn's --link-journal= switch gained two new values
5900 "try-guest" and "try-host" that work like "guest" and
5901 "host", but do not fail if the host has no persistent
5902 journalling enabled. -j is now equivalent to
5903 --link-journal=try-guest.
5905 * macvlan network devices created by nspawn will now have
5906 stable MAC addresses.
5908 * A new SmackProcessLabel= unit setting has been added, which
5909 controls the SMACK security label processes forked off by
5910 the respective unit shall use.
5912 * If compiled with --enable-xkbcommon, systemd-localed will
5913 verify x11 keymap settings by compiling the given keymap. It
5914 will spew out warnings if the compilation fails. This
5915 requires libxkbcommon to be installed.
5917 * When a coredump is collected, a larger number of metadata
5918 fields is now collected and included in the journal records
5919 created for it. More specifically, control group membership,
5920 environment variables, memory maps, working directory,
5921 chroot directory, /proc/$PID/status, and a list of open file
5922 descriptors is now stored in the log entry.
5924 * The udev hwdb now contains DPI information for mice. For
5927 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
5929 * All systemd programs that read standalone configuration
5930 files in /etc now also support a corresponding series of
5931 .conf.d configuration directories in /etc/, /run/,
5932 /usr/local/lib/, /usr/lib/, and (if configured with
5933 --enable-split-usr) /lib/. In particular, the following
5934 configuration files now have corresponding configuration
5935 directories: system.conf user.conf, logind.conf,
5936 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
5937 resolved.conf, timesyncd.conf, journal-remote.conf, and
5938 journal-upload.conf. Note that distributions should use the
5939 configuration directories in /usr/lib/; the directories in
5940 /etc/ are reserved for the system administrator.
5942 * systemd-rfkill will no longer take the rfkill device name
5943 into account when storing rfkill state on disk, as the name
5944 might be dynamically assigned and not stable. Instead, the
5945 ID_PATH udev variable combined with the rfkill type (wlan,
5946 bluetooth, ...) is used.
5948 * A new service systemd-machine-id-commit.service has been
5949 added. When used on systems where /etc is read-only during
5950 boot, and /etc/machine-id is not initialized (but an empty
5951 file), this service will copy the temporary machine ID
5952 created as replacement into /etc after the system is fully
5953 booted up. This is useful for systems that are freshly
5954 installed with a non-initialized machine ID, but should get
5955 a fixed machine ID for subsequent boots.
5957 * networkd's .netdev files now provide a large set of
5958 configuration parameters for VXLAN devices. Similarly, the
5959 bridge port cost parameter is now configurable in .network
5960 files. There's also new support for configuring IP source
5961 routing. networkd .link files gained support for a new
5962 OriginalName= match that is useful to match against the
5963 original interface name the kernel assigned. .network files
5964 may include MTU= and MACAddress= fields for altering the MTU
5965 and MAC address while being connected to a specific network
5968 * The LUKS logic gained supported for configuring
5969 UUID-specific key files. There's also new support for naming
5970 LUKS device from the kernel command line, using the new
5971 luks.name= argument.
5973 * Timer units may now be transiently created via the bus API
5974 (this was previously already available for scope and service
5975 units). In addition it is now possible to create multiple
5976 transient units at the same time with a single bus call. The
5977 "systemd-run" tool has been updated to make use of this for
5978 running commands on a specified time, in at(1)-style.
5980 * tmpfiles gained support for "t" lines, for assigning
5981 extended attributes to files. Among other uses this may be
5982 used to assign SMACK labels to files.
5984 Contributions from: Alin Rauta, Alison Chaiken, Andrej
5985 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
5986 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
5987 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
5988 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
5989 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
5990 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
5991 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
5992 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
5993 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
5994 Michael Biebl, Michael Chapman, Michael Marineau, Michal
5995 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
5996 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
5997 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
5998 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
5999 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
6000 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
6001 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
6003 — Berlin, 2014-12-10
6007 * journalctl gained the new options -t/--identifier= to match
6008 on the syslog identifier (aka "tag"), as well as --utc to
6009 show log timestamps in the UTC timezone. journalctl now also
6010 accepts -n/--lines=all to disable line capping in a pager.
6012 * journalctl gained a new switch, --flush, that synchronously
6013 flushes logs from /run/log/journal to /var/log/journal if
6014 persistent storage is enabled. systemd-journal-flush.service
6015 now waits until the operation is complete.
6017 * Services can notify the manager before they start a reload
6018 (by sending RELOADING=1) or shutdown (by sending
6019 STOPPING=1). This allows the manager to track and show the
6020 internal state of daemons and closes a race condition when
6021 the process is still running but has closed its D-Bus
6024 * Services with Type=oneshot do not have to have any ExecStart
6027 * User units are now loaded also from
6028 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
6029 /run/systemd/user directory that was already previously
6030 supported, but is under the control of the user.
6032 * Job timeouts (i.e. timeouts on the time a job that is
6033 queued stays in the run queue) can now optionally result in
6034 immediate reboot or power-off actions (JobTimeoutAction= and
6035 JobTimeoutRebootArgument=). This is useful on ".target"
6036 units, to limit the maximum time a target remains
6037 undispatched in the run queue, and to trigger an emergency
6038 operation in such a case. This is now used by default to
6039 turn off the system if boot-up (as defined by everything in
6040 basic.target) hangs and does not complete for at least
6041 15min. Also, if power-off or reboot hang for at least 30min
6042 an immediate power-off/reboot operation is triggered. This
6043 functionality is particularly useful to increase reliability
6044 on embedded devices, but also on laptops which might
6045 accidentally get powered on when carried in a backpack and
6046 whose boot stays stuck in a hard disk encryption passphrase
6049 * systemd-logind can be configured to also handle lid switch
6050 events even when the machine is docked or multiple displays
6051 are attached (HandleLidSwitchDocked= option).
6053 * A helper binary and a service have been added which can be
6054 used to resume from hibernation in the initramfs. A
6055 generator will parse the resume= option on the kernel
6056 command line to trigger resume.
6058 * A user console daemon systemd-consoled has been
6059 added. Currently, it is a preview, and will so far open a
6060 single terminal on each session of the user marked as
6061 Desktop=systemd-console.
6063 * Route metrics can be specified for DHCP routes added by
6066 * The SELinux context of socket-activated services can be set
6067 from the information provided by the networking stack
6068 (SELinuxContextFromNet= option).
6070 * Userspace firmware loading support has been removed and
6071 the minimum supported kernel version is thus bumped to 3.7.
6073 * Timeout for udev workers has been increased from 1 to 3
6074 minutes, but a warning will be printed after 1 minute to
6075 help diagnose kernel modules that take a long time to load.
6077 * Udev rules can now remove tags on devices with TAG-="foobar".
6079 * systemd's readahead implementation has been removed. In many
6080 circumstances it didn't give expected benefits even for
6081 rotational disk drives and was becoming less relevant in the
6082 age of SSDs. As none of the developers has been using
6083 rotating media anymore, and nobody stepped up to actively
6084 maintain this component of systemd it has now been removed.
6086 * Swap units can use Options= to specify discard options.
6087 Discard options specified for swaps in /etc/fstab are now
6090 * Docker containers are now detected as a separate type of
6093 * The Password Agent protocol gained support for queries where
6094 the user input is shown, useful e.g. for user names.
6095 systemd-ask-password gained a new --echo option to turn that
6098 * The default sysctl.d/ snippets will now set:
6100 net.core.default_qdisc = fq_codel
6102 This selects Fair Queuing Controlled Delay as the default
6103 queuing discipline for network interfaces. fq_codel helps
6104 fight the network bufferbloat problem. It is believed to be
6105 a good default with no tuning required for most workloads.
6106 Downstream distributions may override this choice. On 10Gbit
6107 servers that do not do forwarding, "fq" may perform better.
6108 Systems without a good clocksource should use "pfifo_fast".
6110 * If kdbus is enabled during build a new option BusPolicy= is
6111 available for service units, that allows locking all service
6112 processes into a stricter bus policy, in order to limit
6113 access to various bus services, or even hide most of them
6114 from the service's view entirely.
6116 * networkctl will now show the .network and .link file
6117 networkd has applied to a specific interface.
6119 * sd-login gained a new API call sd_session_get_desktop() to
6120 query which desktop environment has been selected for a
6123 * UNIX utmp support is now compile-time optional to support
6124 legacy-free systems.
6126 * systemctl gained two new commands "add-wants" and
6127 "add-requires" for pulling in units from specific targets
6130 * If the word "rescue" is specified on the kernel command line
6131 the system will now boot into rescue mode (aka
6132 rescue.target), which was previously available only by
6133 specifying "1" or "systemd.unit=rescue.target" on the kernel
6134 command line. This new kernel command line option nicely
6135 mirrors the already existing "emergency" kernel command line
6138 * New kernel command line options mount.usr=, mount.usrflags=,
6139 mount.usrfstype= have been added that match root=, rootflags=,
6140 rootfstype= but allow mounting a specific file system to
6143 * The $NOTIFY_SOCKET is now also passed to control processes of
6144 services, not only the main process.
6146 * This version reenables support for fsck's -l switch. This
6147 means at least version v2.25 of util-linux is required for
6148 operation, otherwise dead-locks on device nodes may
6149 occur. Again: you need to update util-linux to at least
6150 v2.25 when updating systemd to v217.
6152 * The "multi-seat-x" tool has been removed from systemd, as
6153 its functionality has been integrated into X servers 1.16,
6154 and the tool is hence redundant. It is recommended to update
6155 display managers invoking this tool to simply invoke X
6156 directly from now on, again.
6158 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
6159 message flag has been added for all of systemd's polkit
6160 authenticated method calls has been added. In particular this
6161 now allows optional interactive authorization via polkit for
6162 many of PID1's privileged operations such as unit file
6163 enabling and disabling.
6165 * "udevadm hwdb --update" learnt a new switch "--usr" for
6166 placing the rebuilt hardware database in /usr instead of
6167 /etc. When used only hardware database entries stored in
6168 /usr will be used, and any user database entries in /etc are
6169 ignored. This functionality is useful for vendors to ship a
6170 pre-built database on systems where local configuration is
6171 unnecessary or unlikely.
6173 * Calendar time specifications in .timer units now also
6174 understand the strings "semi-annually", "quarterly" and
6175 "minutely" as shortcuts (in addition to the preexisting
6176 "anually", "hourly", ...).
6178 * systemd-tmpfiles will now correctly create files in /dev
6179 at boot which are marked for creation only at boot. It is
6180 recommended to always create static device nodes with 'c!'
6181 and 'b!', so that they are created only at boot and not
6182 overwritten at runtime.
6184 * When the watchdog logic is used for a service (WatchdogSec=)
6185 and the watchdog timeout is hit the service will now be
6186 terminated with SIGABRT (instead of just SIGTERM), in order
6187 to make sure a proper coredump and backtrace is
6188 generated. This ensures that hanging services will result in
6189 similar coredump/backtrace behaviour as services that hit a
6192 Contributions from: Andreas Henriksson, Andrei Borzenkov,
6193 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
6194 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
6195 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
6196 Herrmann, David Sommerseth, David Strauss, Emil Renner
6197 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
6198 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
6199 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
6200 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
6201 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
6202 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
6203 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
6204 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
6205 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
6206 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
6207 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
6208 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
6209 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
6210 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
6211 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
6212 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
6215 — Berlin, 2014-10-28
6219 * timedated no longer reads NTP implementation unit names from
6220 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
6221 implementations should add a
6223 Conflicts=systemd-timesyncd.service
6225 to their unit files to take over and replace systemd's NTP
6226 default functionality.
6228 * systemd-sysusers gained a new line type "r" for configuring
6229 which UID/GID ranges to allocate system users/groups
6230 from. Lines of type "u" may now add an additional column
6231 that specifies the home directory for the system user to be
6232 created. Also, systemd-sysusers may now optionally read user
6233 information from STDIN instead of a file. This is useful for
6234 invoking it from RPM preinst scriptlets that need to create
6235 users before the first RPM file is installed since these
6236 files might need to be owned by them. A new
6237 %sysusers_create_inline RPM macro has been introduced to do
6238 just that. systemd-sysusers now updates the shadow files as
6239 well as the user/group databases, which should enhance
6240 compatibility with certain tools like grpck.
6242 * A number of bus APIs of PID 1 now optionally consult polkit to
6243 permit access for otherwise unprivileged clients under certain
6244 conditions. Note that this currently doesn't support
6245 interactive authentication yet, but this is expected to be
6246 added eventually, too.
6248 * /etc/machine-info now has new fields for configuring the
6249 deployment environment of the machine, as well as the
6250 location of the machine. hostnamectl has been updated with
6251 new command to update these fields.
6253 * systemd-timesyncd has been updated to automatically acquire
6254 NTP server information from systemd-networkd, which might
6255 have been discovered via DHCP.
6257 * systemd-resolved now includes a caching DNS stub resolver
6258 and a complete LLMNR name resolution implementation. A new
6259 NSS module "nss-resolve" has been added which can be used
6260 instead of glibc's own "nss-dns" to resolve hostnames via
6261 systemd-resolved. Hostnames, addresses and arbitrary RRs may
6262 be resolved via systemd-resolved D-Bus APIs. In contrast to
6263 the glibc internal resolver systemd-resolved is aware of
6264 multi-homed system, and keeps DNS server and caches separate
6265 and per-interface. Queries are sent simultaneously on all
6266 interfaces that have DNS servers configured, in order to
6267 properly handle VPNs and local LANs which might resolve
6268 separate sets of domain names. systemd-resolved may acquire
6269 DNS server information from systemd-networkd automatically,
6270 which in turn might have discovered them via DHCP. A tool
6271 "systemd-resolve-host" has been added that may be used to
6272 query the DNS logic in resolved. systemd-resolved implements
6273 IDNA and automatically uses IDNA or UTF-8 encoding depending
6274 on whether classic DNS or LLMNR is used as transport. In the
6275 next releases we intend to add a DNSSEC and mDNS/DNS-SD
6276 implementation to systemd-resolved.
6278 * A new NSS module nss-mymachines has been added, that
6279 automatically resolves the names of all local registered
6280 containers to their respective IP addresses.
6282 * A new client tool "networkctl" for systemd-networkd has been
6283 added. It currently is entirely passive and will query
6284 networking configuration from udev, rtnetlink and networkd,
6285 and present it to the user in a very friendly
6286 way. Eventually, we hope to extend it to become a full
6287 control utility for networkd.
6289 * .socket units gained a new DeferAcceptSec= setting that
6290 controls the kernels' TCP_DEFER_ACCEPT sockopt for
6291 TCP. Similarly, support for controlling TCP keep-alive
6292 settings has been added (KeepAliveTimeSec=,
6293 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
6294 turning off Nagle's algorithm on TCP has been added
6297 * logind learned a new session type "web", for use in projects
6298 like Cockpit which register web clients as PAM sessions.
6300 * timer units with at least one OnCalendar= setting will now
6301 be started only after time-sync.target has been
6302 reached. This way they will not elapse before the system
6303 clock has been corrected by a local NTP client or
6304 similar. This is particular useful on RTC-less embedded
6305 machines, that come up with an invalid system clock.
6307 * systemd-nspawn's --network-veth= switch should now result in
6308 stable MAC addresses for both the outer and the inner side
6311 * systemd-nspawn gained a new --volatile= switch for running
6312 container instances with /etc or /var unpopulated.
6314 * The kdbus client code has been updated to use the new Linux
6315 3.17 memfd subsystem instead of the old kdbus-specific one.
6317 * systemd-networkd's DHCP client and server now support
6318 FORCERENEW. There are also new configuration options to
6319 configure the vendor client identifier and broadcast mode
6322 * systemd will no longer inform the kernel about the current
6323 timezone, as this is necessarily incorrect and racy as the
6324 kernel has no understanding of DST and similar
6325 concepts. This hence means FAT timestamps will be always
6326 considered UTC, similar to what Android is already
6327 doing. Also, when the RTC is configured to the local time
6328 (rather than UTC) systemd will never synchronize back to it,
6329 as this might confuse Windows at a later boot.
6331 * systemd-analyze gained a new command "verify" for offline
6332 validation of unit files.
6334 * systemd-networkd gained support for a couple of additional
6335 settings for bonding networking setups. Also, the metric for
6336 statically configured routes may now be configured. For
6337 network interfaces where this is appropriate the peer IP
6338 address may now be configured.
6340 * systemd-networkd's DHCP client will no longer request
6341 broadcasting by default, as this tripped up some networks.
6342 For hardware where broadcast is required the feature should
6343 be switched back on using RequestBroadcast=yes.
6345 * systemd-networkd will now set up IPv4LL addresses (when
6346 enabled) even if DHCP is configured successfully.
6348 * udev will now default to respect network device names given
6349 by the kernel when the kernel indicates that these are
6350 predictable. This behavior can be tweaked by changing
6351 NamePolicy= in the relevant .link file.
6353 * A new library systemd-terminal has been added that
6354 implements full TTY stream parsing and rendering. This
6355 library is supposed to be used later on for implementing a
6356 full userspace VT subsystem, replacing the current kernel
6359 * A new tool systemd-journal-upload has been added to push
6360 journal data to a remote system running
6361 systemd-journal-remote.
6363 * journald will no longer forward all local data to another
6364 running syslog daemon. This change has been made because
6365 rsyslog (which appears to be the most commonly used syslog
6366 implementation these days) no longer makes use of this, and
6367 instead pulls the data out of the journal on its own. Since
6368 forwarding the messages to a non-existent syslog server is
6369 more expensive than we assumed we have now turned this
6370 off. If you run a syslog server that is not a recent rsyslog
6371 version, you have to turn this option on again
6372 (ForwardToSyslog= in journald.conf).
6374 * journald now optionally supports the LZ4 compressor for
6375 larger journal fields. This compressor should perform much
6376 better than XZ which was the previous default.
6378 * machinectl now shows the IP addresses of local containers,
6379 if it knows them, plus the interface name of the container.
6381 * A new tool "systemd-escape" has been added that makes it
6382 easy to escape strings to build unit names and similar.
6384 * sd_notify() messages may now include a new ERRNO= field
6385 which is parsed and collected by systemd and shown among the
6386 "systemctl status" output for a service.
6388 * A new component "systemd-firstboot" has been added that
6389 queries the most basic systemd information (timezone,
6390 hostname, root password) interactively on first
6391 boot. Alternatively it may also be used to provision these
6392 things offline on OS images installed into directories.
6394 * The default sysctl.d/ snippets will now set
6396 net.ipv4.conf.default.promote_secondaries=1
6398 This has the benefit of no flushing secondary IP addresses
6399 when primary addresses are removed.
6401 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
6402 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
6403 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
6404 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
6405 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
6406 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
6407 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
6408 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
6409 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
6410 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
6411 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
6412 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
6413 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
6414 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
6415 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
6417 — Berlin, 2014-08-19
6421 * A new tool systemd-sysusers has been added. This tool
6422 creates system users and groups in /etc/passwd and
6423 /etc/group, based on static declarative system user/group
6424 definitions in /usr/lib/sysusers.d/. This is useful to
6425 enable factory resets and volatile systems that boot up with
6426 an empty /etc directory, and thus need system users and
6427 groups created during early boot. systemd now also ships
6428 with two default sysusers.d/ files for the most basic
6429 users and groups systemd and the core operating system
6432 * A new tmpfiles snippet has been added that rebuilds the
6433 essential files in /etc on boot, should they be missing.
6435 * A directive for ensuring automatic clean-up of
6436 /var/cache/man/ has been removed from the default
6437 configuration. This line should now be shipped by the man
6438 implementation. The necessary change has been made to the
6439 man-db implementation. Note that you need to update your man
6440 implementation to one that ships this line, otherwise no
6441 automatic clean-up of /var/cache/man will take place.
6443 * A new condition ConditionNeedsUpdate= has been added that
6444 may conditionalize services to only run when /etc or /var
6445 are "older" than the vendor operating system resources in
6446 /usr. This is useful for reconstructing or updating /etc
6447 after an offline update of /usr or a factory reset, on the
6448 next reboot. Services that want to run once after such an
6449 update or reset should use this condition and order
6450 themselves before the new systemd-update-done.service, which
6451 will mark the two directories as fully updated. A number of
6452 service files have been added making use of this, to rebuild
6453 the udev hardware database, the journald message catalog and
6454 dynamic loader cache (ldconfig). The systemd-sysusers tool
6455 described above also makes use of this now. With this in
6456 place it is now possible to start up a minimal operating
6457 system with /etc empty cleanly. For more information on the
6458 concepts involved see this recent blog story:
6460 http://0pointer.de/blog/projects/stateless.html
6462 * A new system group "input" has been introduced, and all
6463 input device nodes get this group assigned. This is useful
6464 for system-level software to get access to input devices. It
6465 complements what is already done for "audio" and "video".
6467 * systemd-networkd learnt minimal DHCPv4 server support in
6468 addition to the existing DHCPv4 client support. It also
6469 learnt DHCPv6 client and IPv6 Router Solicitation client
6470 support. The DHCPv4 client gained support for static routes
6471 passed in from the server. Note that the [DHCPv4] section
6472 known in older systemd-networkd versions has been renamed to
6473 [DHCP] and is now also used by the DHCPv6 client. Existing
6474 .network files using settings of this section should be
6475 updated, though compatibility is maintained. Optionally, the
6476 client hostname may now be sent to the DHCP server.
6478 * networkd gained support for vxlan virtual networks as well
6479 as tun/tap and dummy devices.
6481 * networkd gained support for automatic allocation of address
6482 ranges for interfaces from a system-wide pool of
6483 addresses. This is useful for dynamically managing a large
6484 number of interfaces with a single network configuration
6485 file. In particular this is useful to easily assign
6486 appropriate IP addresses to the veth links of a large number
6487 of nspawn instances.
6489 * RPM macros for processing sysusers, sysctl and binfmt
6490 drop-in snippets at package installation time have been
6493 * The /etc/os-release file should now be placed in
6494 /usr/lib/os-release. The old location is automatically
6495 created as symlink. /usr/lib is the more appropriate
6496 location of this file, since it shall actually describe the
6497 vendor operating system shipped in /usr, and not the
6498 configuration stored in /etc.
6500 * .mount units gained a new boolean SloppyOptions= setting
6501 that maps to mount(8)'s -s option which enables permissive
6502 parsing of unknown mount options.
6504 * tmpfiles learnt a new "L+" directive which creates a symlink
6505 but (unlike "L") deletes a pre-existing file first, should
6506 it already exist and not already be the correct
6507 symlink. Similarly, "b+", "c+" and "p+" directives have been
6508 added as well, which create block and character devices, as
6509 well as fifos in the filesystem, possibly removing any
6510 pre-existing files of different types.
6512 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
6513 'argument' field (which so far specified the source to
6514 symlink/copy the files from) is now optional. If omitted the
6515 same file os copied from /usr/share/factory/ suffixed by the
6516 full destination path. This is useful for populating /etc
6517 with essential files, by copying them from vendor defaults
6518 shipped in /usr/share/factory/etc.
6520 * A new command "systemctl preset-all" has been added that
6521 applies the service preset settings to all installed unit
6522 files. A new switch --preset-mode= has been added that
6523 controls whether only enable or only disable operations
6526 * A new command "systemctl is-system-running" has been added
6527 that allows checking the overall state of the system, for
6528 example whether it is fully up and running.
6530 * When the system boots up with an empty /etc, the equivalent
6531 to "systemctl preset-all" is executed during early boot, to
6532 make sure all default services are enabled after a factory
6535 * systemd now contains a minimal preset file that enables the
6536 most basic services systemd ships by default.
6538 * Unit files' [Install] section gained a new DefaultInstance=
6539 field for defining the default instance to create if a
6540 template unit is enabled with no instance specified.
6542 * A new passive target cryptsetup-pre.target has been added
6543 that may be used by services that need to make they run and
6544 finish before the first LUKS cryptographic device is set up.
6546 * The /dev/loop-control and /dev/btrfs-control device nodes
6547 are now owned by the "disk" group by default, opening up
6548 access to this group.
6550 * systemd-coredump will now automatically generate a
6551 stack trace of all core dumps taking place on the system,
6552 based on elfutils' libdw library. This stack trace is logged
6555 * systemd-coredump may now optionally store coredumps directly
6556 on disk (in /var/lib/systemd/coredump, possibly compressed),
6557 instead of storing them unconditionally in the journal. This
6558 mode is the new default. A new configuration file
6559 /etc/systemd/coredump.conf has been added to configure this
6560 and other parameters of systemd-coredump.
6562 * coredumpctl gained a new "info" verb to show details about a
6563 specific coredump. A new switch "-1" has also been added
6564 that makes sure to only show information about the most
6565 recent entry instead of all entries. Also, as the tool is
6566 generally useful now the "systemd-" prefix of the binary
6567 name has been removed. Distributions that want to maintain
6568 compatibility with the old name should add a symlink from
6569 the old name to the new name.
6571 * journald's SplitMode= now defaults to "uid". This makes sure
6572 that unprivileged users can access their own coredumps with
6573 coredumpctl without restrictions.
6575 * New kernel command line options "systemd.wants=" (for
6576 pulling an additional unit during boot), "systemd.mask="
6577 (for masking a specific unit for the boot), and
6578 "systemd.debug-shell" (for enabling the debug shell on tty9)
6579 have been added. This is implemented in the new generator
6580 "systemd-debug-generator".
6582 * systemd-nspawn will now by default filter a couple of
6583 syscalls for containers, among them those required for
6584 kernel module loading, direct x86 IO port access, swap
6585 management, and kexec. Most importantly though
6586 open_by_handle_at() is now prohibited for containers,
6587 closing a hole similar to a recently discussed vulnerability
6588 in docker regarding access to files on file hierarchies the
6589 container should normally not have access to. Note that, for
6590 nspawn, we generally make no security claims anyway (and
6591 this is explicitly documented in the man page), so this is
6592 just a fix for one of the most obvious problems.
6594 * A new man page file-hierarchy(7) has been added that
6595 contains a minimized, modernized version of the file system
6596 layout systemd expects, similar in style to the FHS
6597 specification or hier(5). A new tool systemd-path(1) has
6598 been added to query many of these paths for the local
6601 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
6602 longer done. Since the directory now has a per-user size
6603 limit, and is cleaned on logout this appears unnecessary,
6604 in particular since this now brings the lifecycle of this
6605 directory closer in line with how IPC objects are handled.
6607 * systemd.pc now exports a number of additional directories,
6608 including $libdir (which is useful to identify the library
6609 path for the primary architecture of the system), and a
6610 couple of drop-in directories.
6612 * udev's predictable network interface names now use the dev_port
6613 sysfs attribute, introduced in linux 3.15 instead of dev_id to
6614 distinguish between ports of the same PCI function. dev_id should
6615 only be used for ports using the same HW address, hence the need
6618 * machined has been updated to export the OS version of a
6619 container (read from /etc/os-release and
6620 /usr/lib/os-release) on the bus. This is now shown in
6621 "machinectl status" for a machine.
6623 * A new service setting RestartForceExitStatus= has been
6624 added. If configured to a set of exit signals or process
6625 return values, the service will be restarted when the main
6626 daemon process exits with any of them, regardless of the
6629 * systemctl's -H switch for connecting to remote systemd
6630 machines has been extended so that it may be used to
6631 directly connect to a specific container on the
6632 host. "systemctl -H root@foobar:waldi" will now connect as
6633 user "root" to host "foobar", and then proceed directly to
6634 the container named "waldi". Note that currently you have to
6635 authenticate as user "root" for this to work, as entering
6636 containers is a privileged operation.
6638 Contributions from: Andreas Henriksson, Benjamin Steinwender,
6639 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
6640 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
6641 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
6642 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
6643 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
6644 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
6645 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
6646 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
6647 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
6648 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
6649 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
6651 — Berlin, 2014-07-03
6655 * As an experimental feature, udev now tries to lock the
6656 disk device node (flock(LOCK_SH|LOCK_NB)) while it
6657 executes events for the disk or any of its partitions.
6658 Applications like partitioning programs can lock the
6659 disk device node (flock(LOCK_EX)) and claim temporary
6660 device ownership that way; udev will entirely skip all event
6661 handling for this disk and its partitions. If the disk
6662 was opened for writing, the close will trigger a partition
6663 table rescan in udev's "watch" facility, and if needed
6664 synthesize "change" events for the disk and all its partitions.
6665 This is now unconditionally enabled, and if it turns out to
6666 cause major problems, we might turn it on only for specific
6667 devices, or might need to disable it entirely. Device Mapper
6668 devices are excluded from this logic.
6670 * We temporarily dropped the "-l" switch for fsck invocations,
6671 since they collide with the flock() logic above. util-linux
6672 upstream has been changed already to avoid this conflict,
6673 and we will re-add "-l" as soon as util-linux with this
6674 change has been released.
6676 * The dependency on libattr has been removed. Since a long
6677 time, the extended attribute calls have moved to glibc, and
6678 libattr is thus unnecessary.
6680 * Virtualization detection works without privileges now. This
6681 means the systemd-detect-virt binary no longer requires
6682 CAP_SYS_PTRACE file capabilities, and our daemons can run
6683 with fewer privileges.
6685 * systemd-networkd now runs under its own "systemd-network"
6686 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
6687 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
6688 loses the ability to write to files owned by root this way.
6690 * Similarly, systemd-resolved now runs under its own
6691 "systemd-resolve" user with no capabilities remaining.
6693 * Similarly, systemd-bus-proxyd now runs under its own
6694 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
6696 * systemd-networkd gained support for setting up "veth"
6697 virtual Ethernet devices for container connectivity, as well
6698 as GRE and VTI tunnels.
6700 * systemd-networkd will no longer automatically attempt to
6701 manually load kernel modules necessary for certain tunnel
6702 transports. Instead, it is assumed the kernel loads them
6703 automatically when required. This only works correctly on
6704 very new kernels. On older kernels, please consider adding
6705 the kernel modules to /etc/modules-load.d/ as a work-around.
6707 * The resolv.conf file systemd-resolved generates has been
6708 moved to /run/systemd/resolve/. If you have a symlink from
6709 /etc/resolv.conf, it might be necessary to correct it.
6711 * Two new service settings, ProtectHome= and ProtectSystem=,
6712 have been added. When enabled, they will make the user data
6713 (such as /home) inaccessible or read-only and the system
6714 (such as /usr) read-only, for specific services. This allows
6715 very light-weight per-service sandboxing to avoid
6716 modifications of user data or system files from
6717 services. These two new switches have been enabled for all
6718 of systemd's long-running services, where appropriate.
6720 * Socket units gained new SocketUser= and SocketGroup=
6721 settings to set the owner user and group of AF_UNIX sockets
6722 and FIFOs in the file system.
6724 * Socket units gained a new RemoveOnStop= setting. If enabled,
6725 all FIFOS and sockets in the file system will be removed
6726 when the specific socket unit is stopped.
6728 * Socket units gained a new Symlinks= setting. It takes a list
6729 of symlinks to create to file system sockets or FIFOs
6730 created by the specific Unix sockets. This is useful to
6731 manage symlinks to socket nodes with the same lifecycle as
6734 * The /dev/log socket and /dev/initctl FIFO have been moved to
6735 /run, and have been replaced by symlinks. This allows
6736 connecting to these facilities even if PrivateDevices=yes is
6737 used for a service (which makes /dev/log itself unavailable,
6738 but /run is left). This also has the benefit of ensuring
6739 that /dev only contains device nodes, directories and
6740 symlinks, and nothing else.
6742 * sd-daemon gained two new calls sd_pid_notify() and
6743 sd_pid_notifyf(). They are similar to sd_notify() and
6744 sd_notifyf(), but allow overriding of the source PID of
6745 notification messages if permissions permit this. This is
6746 useful to send notify messages on behalf of a different
6747 process (for example, the parent process). The
6748 systemd-notify tool has been updated to make use of this
6749 when sending messages (so that notification messages now
6750 originate from the shell script invoking systemd-notify and
6751 not the systemd-notify process itself. This should minimize
6752 a race where systemd fails to associate notification
6753 messages to services when the originating process already
6756 * A new "on-abnormal" setting for Restart= has been added. If
6757 set, it will result in automatic restarts on all "abnormal"
6758 reasons for a process to exit, which includes unclean
6759 signals, core dumps, timeouts and watchdog timeouts, but
6760 does not include clean and unclean exit codes or clean
6761 signals. Restart=on-abnormal is an alternative for
6762 Restart=on-failure for services that shall be able to
6763 terminate and avoid restarts on certain errors, by
6764 indicating so with an unclean exit code. Restart=on-failure
6765 or Restart=on-abnormal is now the recommended setting for
6766 all long-running services.
6768 * If the InaccessibleDirectories= service setting points to a
6769 mount point (or if there are any submounts contained within
6770 it), it is now attempted to completely unmount it, to make
6771 the file systems truly unavailable for the respective
6774 * The ReadOnlyDirectories= service setting and
6775 systemd-nspawn's --read-only parameter are now recursively
6776 applied to all submounts, too.
6778 * Mount units may now be created transiently via the bus APIs.
6780 * The support for SysV and LSB init scripts has been removed
6781 from the systemd daemon itself. Instead, it is now
6782 implemented as a generator that creates native systemd units
6783 from these scripts when needed. This enables us to remove a
6784 substantial amount of legacy code from PID 1, following the
6785 fact that many distributions only ship a very small number
6786 of LSB/SysV init scripts nowadays.
6788 * Privileged Xen (dom0) domains are not considered
6789 virtualization anymore by the virtualization detection
6790 logic. After all, they generally have unrestricted access to
6791 the hardware and usually are used to manage the unprivileged
6794 * systemd-tmpfiles gained a new "C" line type, for copying
6795 files or entire directories.
6797 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
6798 lines. So far, they have been non-globbing versions of the
6799 latter, and have thus been redundant. In future, it is
6800 recommended to only use "z". "m" has hence been removed
6801 from the documentation, even though it stays supported.
6803 * A tmpfiles snippet to recreate the most basic structure in
6804 /var has been added. This is enough to create the /var/run →
6805 /run symlink and create a couple of structural
6806 directories. This allows systems to boot up with an empty or
6807 volatile /var. Of course, while with this change, the core OS
6808 now is capable with dealing with a volatile /var, not all
6809 user services are ready for it. However, we hope that sooner
6810 or later, many service daemons will be changed upstream so
6811 that they are able to automatically create their necessary
6812 directories in /var at boot, should they be missing. This is
6813 the first step to allow state-less systems that only require
6814 the vendor image for /usr to boot.
6816 * systemd-nspawn has gained a new --tmpfs= switch to mount an
6817 empty tmpfs instance to a specific directory. This is
6818 particularly useful for making use of the automatic
6819 reconstruction of /var (see above), by passing --tmpfs=/var.
6821 * Access modes specified in tmpfiles snippets may now be
6822 prefixed with "~", which indicates that they shall be masked
6823 by whether the existing file or directory is currently
6824 writable, readable or executable at all. Also, if specified,
6825 the sgid/suid/sticky bits will be masked for all
6828 * A new passive target unit "network-pre.target" has been
6829 added which is useful for services that shall run before any
6830 network is configured, for example firewall scripts.
6832 * The "floppy" group that previously owned the /dev/fd*
6833 devices is no longer used. The "disk" group is now used
6834 instead. Distributions should probably deprecate usage of
6837 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
6838 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
6839 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
6840 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
6841 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
6842 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
6845 — Berlin, 2014-06-11
6849 * A new "systemd-timesyncd" daemon has been added for
6850 synchronizing the system clock across the network. It
6851 implements an SNTP client. In contrast to NTP
6852 implementations such as chrony or the NTP reference server,
6853 this only implements a client side, and does not bother with
6854 the full NTP complexity, focusing only on querying time from
6855 one remote server and synchronizing the local clock to
6856 it. Unless you intend to serve NTP to networked clients or
6857 want to connect to local hardware clocks, this simple NTP
6858 client should be more than appropriate for most
6859 installations. The daemon runs with minimal privileges, and
6860 has been hooked up with networkd to only operate when
6861 network connectivity is available. The daemon saves the
6862 current clock to disk every time a new NTP sync has been
6863 acquired, and uses this to possibly correct the system clock
6864 early at bootup, in order to accommodate for systems that
6865 lack an RTC such as the Raspberry Pi and embedded devices,
6866 and to make sure that time monotonically progresses on these
6867 systems, even if it is not always correct. To make use of
6868 this daemon, a new system user and group "systemd-timesync"
6869 needs to be created on installation of systemd.
6871 * The queue "seqnum" interface of libudev has been disabled, as
6872 it was generally incompatible with device namespacing as
6873 sequence numbers of devices go "missing" if the devices are
6874 part of a different namespace.
6876 * "systemctl list-timers" and "systemctl list-sockets" gained
6877 a --recursive switch for showing units of these types also
6878 for all local containers, similar in style to the already
6879 supported --recursive switch for "systemctl list-units".
6881 * A new RebootArgument= setting has been added for service
6882 units, which may be used to specify a kernel reboot argument
6883 to use when triggering reboots with StartLimitAction=.
6885 * A new FailureAction= setting has been added for service
6886 units which may be used to specify an operation to trigger
6887 when a service fails. This works similarly to
6888 StartLimitAction=, but unlike it, controls what is done
6889 immediately rather than only after several attempts to
6890 restart the service in question.
6892 * hostnamed got updated to also expose the kernel name,
6893 release, and version on the bus. This is useful for
6894 executing commands like hostnamectl with the -H switch.
6895 systemd-analyze makes use of this to properly display
6896 details when running non-locally.
6898 * The bootchart tool can now show cgroup information in the
6899 graphs it generates.
6901 * The CFS CPU quota cgroup attribute is now exposed for
6902 services. The new CPUQuota= switch has been added for this
6903 which takes a percentage value. Setting this will have the
6904 result that a service may never get more CPU time than the
6905 specified percentage, even if the machine is otherwise idle.
6907 * systemd-networkd learned IPIP and SIT tunnel support.
6909 * LSB init scripts exposing a dependency on $network will now
6910 get a dependency on network-online.target rather than simply
6911 network.target. This should bring LSB handling closer to
6912 what it was on SysV systems.
6914 * A new fsck.repair= kernel option has been added to control
6915 how fsck shall deal with unclean file systems at boot.
6917 * The (.ini) configuration file parser will now silently
6918 ignore sections whose name begins with "X-". This may be
6919 used to maintain application-specific extension sections in unit
6922 * machined gained a new API to query the IP addresses of
6923 registered containers. "machinectl status" has been updated
6924 to show these addresses in its output.
6926 * A new call sd_uid_get_display() has been added to the
6927 sd-login APIs for querying the "primary" session of a
6928 user. The "primary" session of the user is elected from the
6929 user's sessions and generally a graphical session is
6930 preferred over a text one.
6932 * A minimal systemd-resolved daemon has been added. It
6933 currently simply acts as a companion to systemd-networkd and
6934 manages resolv.conf based on per-interface DNS
6935 configuration, possibly supplied via DHCP. In the long run
6936 we hope to extend this into a local DNSSEC enabled DNS and
6939 * The systemd-networkd-wait-online tool is now enabled by
6940 default. It will delay network-online.target until a network
6941 connection has been configured. The tool primarily integrates
6942 with networkd, but will also make a best effort to make sense
6943 of network configuration performed in some other way.
6945 * Two new service options StartupCPUShares= and
6946 StartupBlockIOWeight= have been added that work similarly to
6947 CPUShares= and BlockIOWeight= however only apply during
6948 system startup. This is useful to prioritize certain services
6949 differently during bootup than during normal runtime.
6951 * hostnamed has been changed to prefer the statically
6952 configured hostname in /etc/hostname (unless set to
6953 'localhost' or empty) over any dynamic one supplied by
6954 dhcp. With this change, the rules for picking the hostname
6955 match more closely the rules of other configuration settings
6956 where the local administrator's configuration in /etc always
6957 overrides any other settings.
6959 Contributions from: Ali H. Caliskan, Alison Chaiken, Bas van
6960 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
6961 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
6962 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
6963 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
6964 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
6965 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
6966 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
6967 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
6968 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
6969 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
6970 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
6971 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
6972 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
6973 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
6974 Lindskog, WaLyong Cho, Will Woods, Zbigniew
6977 — Beijing, 2014-05-28
6981 * When restoring the screen brightness at boot, stay away from
6982 the darkest setting or from the lowest 5% of the available
6983 range, depending on which is the larger value of both. This
6984 should effectively protect the user from rebooting into a
6985 black screen, should the brightness have been set to minimum
6988 * sd-login gained a new sd_machine_get_class() call to
6989 determine the class ("vm" or "container") of a machine
6990 registered with machined.
6992 * sd-login gained new calls
6993 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
6994 to query the identity of the peer of a local AF_UNIX
6995 connection. They operate similarly to their sd_pid_get_xyz()
6998 * PID 1 will now maintain a system-wide system state engine
6999 with the states "starting", "running", "degraded",
7000 "maintenance", "stopping". These states are bound to system
7001 startup, normal runtime, runtime with at least one failed
7002 service, rescue/emergency mode and system shutdown. This
7003 state is shown in the "systemctl status" output when no unit
7004 name is passed. It is useful to determine system state, in
7005 particularly when doing so for many systems or containers at
7008 * A new command "list-machines" has been added to "systemctl"
7009 that lists all local OS containers and shows their system
7010 state (see above), if systemd runs inside of them.
7012 * systemctl gained a new "-r" switch to recursively enumerate
7013 units on all local containers, when used with the
7014 "list-unit" command (which is the default one that is
7015 executed when no parameters are specified).
7017 * The GPT automatic partition discovery logic will now honour
7018 two GPT partition flags: one may be set on a partition to
7019 cause it to be mounted read-only, and the other may be set
7020 on a partition to ignore it during automatic discovery.
7022 * Two new GPT type UUIDs have been added for automatic root
7023 partition discovery, for 32-bit and 64-bit ARM. This is not
7024 particularly useful for discovering the root directory on
7025 these architectures during bare-metal boots (since UEFI is
7026 not common there), but still very useful to allow booting of
7027 ARM disk images in nspawn with the -i option.
7029 * MAC addresses of interfaces created with nspawn's
7030 --network-interface= switch will now be generated from the
7031 machine name, and thus be stable between multiple invocations
7034 * logind will now automatically remove all IPC objects owned
7035 by a user if she or he fully logs out. This makes sure that
7036 users who are logged out cannot continue to consume IPC
7037 resources. This covers SysV memory, semaphores and message
7038 queues as well as POSIX shared memory and message
7039 queues. Traditionally, SysV and POSIX IPC had no lifecycle
7040 limits. With this functionality, that is corrected. This may
7041 be turned off by using the RemoveIPC= switch of logind.conf.
7043 * The systemd-machine-id-setup and tmpfiles tools gained a
7044 --root= switch to operate on a specific root directory,
7047 * journald can now forward logged messages to the TTYs of all
7048 logged in users ("wall"). This is the default for all
7049 emergency messages now.
7051 * A new tool systemd-journal-remote has been added to stream
7052 journal log messages across the network.
7054 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
7055 controller trees are mounted into it. Note that the
7056 directories mounted beneath it are not read-only. This is a
7057 security measure and is particularly useful because glibc
7058 actually includes a search logic to pick any tmpfs it can
7059 find to implement shm_open() if /dev/shm is not available
7060 (which it might very well be in namespaced setups).
7062 * machinectl gained a new "poweroff" command to cleanly power
7063 down a local OS container.
7065 * The PrivateDevices= unit file setting will now also drop the
7066 CAP_MKNOD capability from the capability bound set, and
7067 imply DevicePolicy=closed.
7069 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
7070 comprehensively on all long-running systemd services where
7071 this is appropriate.
7073 * systemd-udevd will now run in a disassociated mount
7074 namespace. To mount directories from udev rules, make sure to
7075 pull in mount units via SYSTEMD_WANTS properties.
7077 * The kdbus support gained support for uploading policy into
7078 the kernel. sd-bus gained support for creating "monitoring"
7079 connections that can eavesdrop into all bus communication
7080 for debugging purposes.
7082 * Timestamps may now be specified in seconds since the UNIX
7083 epoch Jan 1st, 1970 by specifying "@" followed by the value
7086 * Native tcpwrap support in systemd has been removed. tcpwrap
7087 is old code, not really maintained anymore and has serious
7088 shortcomings, and better options such as firewalls
7089 exist. For setups that require tcpwrap usage, please
7090 consider invoking your socket-activated service via tcpd,
7091 like on traditional inetd.
7093 * A new system.conf configuration option
7094 DefaultTimerAccuracySec= has been added that controls the
7095 default AccuracySec= setting of .timer units.
7097 * Timer units gained a new WakeSystem= switch. If enabled,
7098 timers configured this way will cause the system to resume
7099 from system suspend (if the system supports that, which most
7102 * Timer units gained a new Persistent= switch. If enabled,
7103 timers configured this way will save to disk when they have
7104 been last triggered. This information is then used on next
7105 reboot to possible execute overdue timer events, that
7106 could not take place because the system was powered off.
7107 This enables simple anacron-like behaviour for timer units.
7109 * systemctl's "list-timers" will now also list the time a
7110 timer unit was last triggered in addition to the next time
7111 it will be triggered.
7113 * systemd-networkd will now assign predictable IPv4LL
7114 addresses to its local interfaces.
7116 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
7117 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
7118 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
7119 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
7120 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
7121 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
7122 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
7123 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
7126 — Berlin, 2014-03-25
7130 * A new unit file setting RestrictAddressFamilies= has been
7131 added to restrict which socket address families unit
7132 processes gain access to. This takes address family names
7133 like "AF_INET" or "AF_UNIX", and is useful to minimize the
7134 attack surface of services via exotic protocol stacks. This
7135 is built on seccomp system call filters.
7137 * Two new unit file settings RuntimeDirectory= and
7138 RuntimeDirectoryMode= have been added that may be used to
7139 manage a per-daemon runtime directories below /run. This is
7140 an alternative for setting up directory permissions with
7141 tmpfiles snippets, and has the advantage that the runtime
7142 directory's lifetime is bound to the daemon runtime and that
7143 the daemon starts up with an empty directory each time. This
7144 is particularly useful when writing services that drop
7145 privileges using the User= or Group= setting.
7147 * The DeviceAllow= unit setting now supports globbing for
7148 matching against device group names.
7150 * The systemd configuration file system.conf gained new
7151 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
7152 DefaultMemoryAccounting= to globally turn on/off accounting
7153 for specific resources (cgroups) for all units. These
7154 settings may still be overridden individually in each unit
7157 * systemd-gpt-auto-generator is now able to discover /srv and
7158 root partitions in addition to /home and swap partitions. It
7159 also supports LUKS-encrypted partitions now. With this in
7160 place, automatic discovery of partitions to mount following
7161 the Discoverable Partitions Specification
7162 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
7163 is now a lot more complete. This allows booting without
7164 /etc/fstab and without root= on the kernel command line on
7165 systems prepared appropriately.
7167 * systemd-nspawn gained a new --image= switch which allows
7168 booting up disk images and Linux installations on any block
7169 device that follow the Discoverable Partitions Specification
7170 (see above). This means that installations made with
7171 appropriately updated installers may now be started and
7172 deployed using container managers, completely
7173 unmodified. (We hope that libvirt-lxc will add support for
7174 this feature soon, too.)
7176 * systemd-nspawn gained a new --network-macvlan= setting to
7177 set up a private macvlan interface for the
7178 container. Similarly, systemd-networkd gained a new
7179 Kind=macvlan setting in .netdev files.
7181 * systemd-networkd now supports configuring local addresses
7184 * A new tool systemd-network-wait-online has been added to
7185 synchronously wait for network connectivity using
7188 * The sd-bus.h bus API gained a new sd_bus_track object for
7189 tracking the lifecycle of bus peers. Note that sd-bus.h is
7190 still not a public API though (unless you specify
7191 --enable-kdbus on the configure command line, which however
7192 voids your warranty and you get no API stability guarantee).
7194 * The $XDG_RUNTIME_DIR runtime directories for each user are
7195 now individual tmpfs instances, which has the benefit of
7196 introducing separate pools for each user, with individual
7197 size limits, and thus making sure that unprivileged clients
7198 can no longer negatively impact the system or other users by
7199 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
7200 RuntimeDirectorySize= has been introduced that allows
7201 controlling the default size limit for all users. It
7202 defaults to 10% of the available physical memory. This is no
7203 replacement for quotas on tmpfs though (which the kernel
7204 still does not support), as /dev/shm and /tmp are still
7205 shared resources used by both the system and unprivileged
7208 * logind will now automatically turn off automatic suspending
7209 on laptop lid close when more than one display is
7210 connected. This was previously expected to be implemented
7211 individually in desktop environments (such as GNOME),
7212 however has been added to logind now, in order to fix a
7213 boot-time race where a desktop environment might not have
7214 been started yet and thus not been able to take an inhibitor
7215 lock at the time where logind already suspends the system
7216 due to a closed lid.
7218 * logind will now wait at least 30s after each system
7219 suspend/resume cycle, and 3min after system boot before
7220 suspending the system due to a closed laptop lid. This
7221 should give USB docking stations and similar enough time to
7222 be probed and configured after system resume and boot in
7223 order to then act as suspend blocker.
7225 * systemd-run gained a new --property= setting which allows
7226 initialization of resource control properties (and others)
7227 for the created scope or service unit. Example: "systemd-run
7228 --property=BlockIOWeight=10 updatedb" may be used to run
7229 updatedb at a low block IO scheduling weight.
7231 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
7232 now also work in --scope mode.
7234 * When systemd is compiled with kdbus support, basic support
7235 for enforced policies is now in place. (Note that enabling
7236 kdbus still voids your warranty and no API compatibility
7239 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
7240 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
7241 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
7242 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
7243 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
7244 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
7245 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
7246 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
7247 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
7248 Zbigniew Jędrzejewski-Szmek
7250 — Berlin, 2014-03-12
7254 * systemd will now relabel /dev after loading the SMACK policy
7255 according to SMACK rules.
7257 * A new unit file option AppArmorProfile= has been added to
7258 set the AppArmor profile for the processes of a unit.
7260 * A new condition check ConditionArchitecture= has been added
7261 to conditionalize units based on the system architecture, as
7262 reported by uname()'s "machine" field.
7264 * systemd-networkd now supports matching on the system
7265 virtualization, architecture, kernel command line, host name
7268 * logind is now a lot more aggressive when suspending the
7269 machine due to a closed laptop lid. Instead of acting only
7270 on the lid close action, it will continuously watch the lid
7271 status and act on it. This is useful for laptops where the
7272 power button is on the outside of the chassis so that it can
7273 be reached without opening the lid (such as the Lenovo
7274 Yoga). On those machines, logind will now immediately
7275 re-suspend the machine if the power button has been
7276 accidentally pressed while the laptop was suspended and in a
7277 backpack or similar.
7279 * logind will now watch SW_DOCK switches and inhibit reaction
7280 to the lid switch if it is pressed. This means that logind
7281 will not suspend the machine anymore if the lid is closed
7282 and the system is docked, if the laptop supports SW_DOCK
7283 notifications via the input layer. Note that ACPI docking
7284 stations do not generate this currently. Also note that this
7285 logic is usually not fully sufficient and Desktop
7286 Environments should take a lid switch inhibitor lock when an
7287 external display is connected, as systemd will not watch
7290 * nspawn will now make use of the devices cgroup controller by
7291 default, and only permit creation of and access to the usual
7292 API device nodes like /dev/null or /dev/random, as well as
7293 access to (but not creation of) the pty devices.
7295 * We will now ship a default .network file for
7296 systemd-networkd that automatically configures DHCP for
7297 network interfaces created by nspawn's --network-veth or
7298 --network-bridge= switches.
7300 * systemd will now understand the usual M, K, G, T suffixes
7301 according to SI conventions (i.e. to the base 1000) when
7302 referring to throughput and hardware metrics. It will stay
7303 with IEC conventions (i.e. to the base 1024) for software
7304 metrics, according to what is customary according to
7305 Wikipedia. We explicitly document which base applies for
7306 each configuration option.
7308 * The DeviceAllow= setting in unit files now supports a syntax
7309 to whitelist an entire group of devices node majors at once,
7310 based on the /proc/devices listing. For example, with the
7311 string "char-pts", it is now possible to whitelist all
7312 current and future pseudo-TTYs at once.
7314 * sd-event learned a new "post" event source. Event sources of
7315 this type are triggered by the dispatching of any event
7316 source of a type that is not "post". This is useful for
7317 implementing clean-up and check event sources that are
7318 triggered by other work being done in the program.
7320 * systemd-networkd is no longer statically enabled, but uses
7321 the usual [Install] sections so that it can be
7322 enabled/disabled using systemctl. It still is enabled by
7325 * When creating a veth interface pair with systemd-nspawn, the
7326 host side will now be prefixed with "vb-" if
7327 --network-bridge= is used, and with "ve-" if --network-veth
7328 is used. This way, it is easy to distinguish these cases on
7329 the host, for example to apply different configuration to
7330 them with systemd-networkd.
7332 * The compatibility libraries for libsystemd-journal.so,
7333 libsystem-id128.so, libsystemd-login.so and
7334 libsystemd-daemon.so do not make use of IFUNC
7335 anymore. Instead, we now build libsystemd.so multiple times
7336 under these alternative names. This means that the footprint
7337 is drastically increased, but given that these are
7338 transitional compatibility libraries, this should not matter
7339 much. This change has been made necessary to support the ARM
7340 platform for these compatibility libraries, as the ARM
7341 toolchain is not really at the same level as the toolchain
7342 for other architectures like x86 and does not support
7343 IFUNC. Please make sure to use --enable-compat-libs only
7344 during a transitional period!
7346 * The .include syntax has been deprecated and is not documented
7347 anymore. Drop-in files in .d directories should be used instead.
7349 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
7350 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
7351 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
7352 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
7353 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
7354 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
7355 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
7356 Zbigniew Jędrzejewski-Szmek
7358 — Berlin, 2014-02-24
7362 * A new component "systemd-networkd" has been added that can
7363 be used to configure local network interfaces statically or
7364 via DHCP. It is capable of bringing up bridges, VLANs, and
7365 bonding. Currently, no hook-ups for interactive network
7366 configuration are provided. Use this for your initrd,
7367 container, embedded, or server setup if you need a simple,
7368 yet powerful, network configuration solution. This
7369 configuration subsystem is quite nifty, as it allows wildcard
7370 hotplug matching in interfaces. For example, with a single
7371 configuration snippet, you can configure that all Ethernet
7372 interfaces showing up are automatically added to a bridge,
7373 or similar. It supports link-sensing and more.
7375 * A new tool "systemd-socket-proxyd" has been added which can
7376 act as a bidirectional proxy for TCP sockets. This is
7377 useful for adding socket activation support to services that
7378 do not actually support socket activation, including virtual
7379 machines and the like.
7381 * Add a new tool to save/restore rfkill state on
7384 * Save/restore state of keyboard backlights in addition to
7385 display backlights on shutdown/boot.
7387 * udev learned a new SECLABEL{} construct to label device
7388 nodes with a specific security label when they appear. For
7389 now, only SECLABEL{selinux} is supported, but the syntax is
7390 prepared for additional security frameworks.
7392 * udev gained a new scheme to configure link-level attributes
7393 from files in /etc/systemd/network/*.link. These files can
7394 match against MAC address, device path, driver name and type,
7395 and will apply attributes like the naming policy, link speed,
7396 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
7397 address assignment policy (randomized, ...).
7399 * The configuration of network interface naming rules for
7400 "permanent interface names" has changed: a new NamePolicy=
7401 setting in the [Link] section of .link files determines the
7402 priority of possible naming schemes (onboard, slot, MAC,
7403 path). The default value of this setting is determined by
7404 /usr/lib/net/links/99-default.link. Old
7405 80-net-name-slot.rules udev configuration file has been
7406 removed, so local configuration overriding this file should
7407 be adapted to override 99-default.link instead.
7409 * When the User= switch is used in a unit file, also
7410 initialize $SHELL= based on the user database entry.
7412 * systemd no longer depends on libdbus. All communication is
7413 now done with sd-bus, systemd's low-level bus library
7416 * kdbus support has been added to PID 1 itself. When kdbus is
7417 enabled, this causes PID 1 to set up the system bus and
7418 enable support for a new ".busname" unit type that
7419 encapsulates bus name activation on kdbus. It works a little
7420 bit like ".socket" units, except for bus names. A new
7421 generator has been added that converts classic dbus1 service
7422 activation files automatically into native systemd .busname
7425 * sd-bus: add a light-weight vtable implementation that allows
7426 defining objects on the bus with a simple static const
7427 vtable array of its methods, signals and properties.
7429 * systemd will not generate or install static dbus
7430 introspection data anymore to /usr/share/dbus-1/interfaces,
7431 as the precise format of these files is unclear, and
7432 nothing makes use of it.
7434 * A proxy daemon is now provided to proxy clients connecting
7435 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
7436 compatibility with classic D-Bus.
7438 * A bus driver implementation has been added that supports the
7439 classic D-Bus bus driver calls on kdbus, also for
7440 compatibility purposes.
7442 * A new API "sd-event.h" has been added that implements a
7443 minimal event loop API built around epoll. It provides a
7444 couple of features that direct epoll usage is lacking:
7445 prioritization of events, scales to large numbers of timer
7446 events, per-event timer slack (accuracy), system-wide
7447 coalescing of timer events, exit handlers, watchdog
7448 supervision support using systemd's sd_notify() API, child
7451 * A new API "sd-rntl.h" has been added that provides an API
7452 around the route netlink interface of the kernel, similar in
7453 style to "sd-bus.h".
7455 * A new API "sd-dhcp-client.h" has been added that provides a
7456 small DHCPv4 client-side implementation. This is used by
7459 * There is a new kernel command line option
7460 "systemd.restore_state=0|1". When set to "0", none of the
7461 systemd tools will restore saved runtime state to hardware
7462 devices. More specifically, the rfkill and backlight states
7465 * The FsckPassNo= compatibility option in mount/service units
7466 has been removed. The fstab generator will now add the
7467 necessary dependencies automatically, and does not require
7468 PID1's support for that anymore.
7470 * journalctl gained a new switch, --list-boots, that lists
7471 recent boots with their times and boot IDs.
7473 * The various tools like systemctl, loginctl, timedatectl,
7474 busctl, systemd-run, ... have gained a new switch "-M" to
7475 connect to a specific, local OS container (as direct
7476 connection, without requiring SSH). This works on any
7477 container that is registered with machined, such as those
7478 created by libvirt-lxc or nspawn.
7480 * systemd-run and systemd-analyze also gained support for "-H"
7481 to connect to remote hosts via SSH. This is particularly
7482 useful for systemd-run because it enables queuing of jobs
7483 onto remote systems.
7485 * machinectl gained a new command "login" to open a getty
7486 login in any local container. This works with any container
7487 that is registered with machined (such as those created by
7488 libvirt-lxc or nspawn), and which runs systemd inside.
7490 * machinectl gained a new "reboot" command that may be used to
7491 trigger a reboot on a specific container that is registered
7492 with machined. This works on any container that runs an init
7493 system of some kind.
7495 * systemctl gained a new "list-timers" command to print a nice
7496 listing of installed timer units with the times they elapse
7499 * Alternative reboot() parameters may now be specified on the
7500 "systemctl reboot" command line and are passed to the
7501 reboot() system call.
7503 * systemctl gained a new --job-mode= switch to configure the
7504 mode to queue a job with. This is a more generic version of
7505 --fail, --irreversible, and --ignore-dependencies, which are
7506 still available but not advertised anymore.
7508 * /etc/systemd/system.conf gained new settings to configure
7509 various default timeouts of units, as well as the default
7510 start limit interval and burst. These may still be overridden
7513 * PID1 will now export on the bus profile data of the security
7514 policy upload process (such as the SELinux policy upload to
7517 * journald: when forwarding logs to the console, include
7518 timestamps (following the setting in
7519 /sys/module/printk/parameters/time).
7521 * OnCalendar= in timer units now understands the special
7522 strings "yearly" and "annually". (Both are equivalent)
7524 * The accuracy of timer units is now configurable with the new
7525 AccuracySec= setting. It defaults to 1min.
7527 * A new dependency type JoinsNamespaceOf= has been added that
7528 allows running two services within the same /tmp and network
7529 namespace, if PrivateNetwork= or PrivateTmp= are used.
7531 * A new command "cat" has been added to systemctl. It outputs
7532 the original unit file of a unit, and concatenates the
7533 contents of additional "drop-in" unit file snippets, so that
7534 the full configuration is shown.
7536 * systemctl now supports globbing on the various "list-xyz"
7537 commands, like "list-units" or "list-sockets", as well as on
7538 those commands which take multiple unit names.
7540 * journalctl's --unit= switch gained support for globbing.
7542 * All systemd daemons now make use of the watchdog logic so
7543 that systemd automatically notices when they hang.
7545 * If the $container_ttys environment variable is set,
7546 getty-generator will automatically spawn a getty for each
7547 listed tty. This is useful for container managers to request
7548 login gettys to be spawned on as many ttys as needed.
7550 * %h, %s, %U specifier support is not available anymore when
7551 used in unit files for PID 1. This is because NSS calls are
7552 not safe from PID 1. They stay available for --user
7553 instances of systemd, and as special case for the root user.
7555 * loginctl gained a new "--no-legend" switch to turn off output
7558 * The "sd-login.h" API gained three new calls:
7559 sd_session_is_remote(), sd_session_get_remote_user(),
7560 sd_session_get_remote_host() to query information about
7563 * The udev hardware database now also carries vendor/product
7564 information of SDIO devices.
7566 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
7567 determine whether watchdog notifications are requested by
7570 * Socket-activated per-connection services now include a
7571 short description of the connection parameters in the
7574 * tmpfiles gained a new "--boot" option. When this is not used,
7575 only lines where the command character is not suffixed with
7576 "!" are executed. When this option is specified, those
7577 options are executed too. This partitions tmpfiles
7578 directives into those that can be safely executed at any
7579 time, and those which should be run only at boot (for
7580 example, a line that creates /run/nologin).
7582 * A new API "sd-resolve.h" has been added which provides a simple
7583 asynchronous wrapper around glibc NSS host name resolution
7584 calls, such as getaddrinfo(). In contrast to glibc's
7585 getaddrinfo_a(), it does not use signals. In contrast to most
7586 other asynchronous name resolution libraries, this one does
7587 not reimplement DNS, but reuses NSS, so that alternate
7588 host name resolution systems continue to work, such as mDNS,
7589 LDAP, etc. This API is based on libasyncns, but it has been
7590 cleaned up for inclusion in systemd.
7592 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
7593 "sd-daemon.h" are no longer found in individual libraries
7594 libsystemd-journal.so, libsystemd-login.so,
7595 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
7596 merged them into a single library, libsystemd.so, which
7597 provides all symbols. The reason for this is cyclic
7598 dependencies, as these libraries tend to use each other's
7599 symbols. So far, we have managed to workaround that by linking
7600 a copy of a good part of our code into each of these
7601 libraries again and again, which, however, makes certain
7602 things hard to do, like sharing static variables. Also, it
7603 substantially increases footprint. With this change, there
7604 is only one library for the basic APIs systemd
7605 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
7606 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
7607 library as well, however are subject to the --enable-kdbus
7608 switch (see below). Note that "sd-dhcp-client.h" is not part
7609 of this library (this is because it only consumes, never
7610 provides, services of/to other APIs). To make the transition
7611 easy from the separate libraries to the unified one, we
7612 provide the --enable-compat-libs compile-time switch which
7613 will generate stub libraries that are compatible with the
7614 old ones but redirect all calls to the new one.
7616 * All of the kdbus logic and the new APIs "sd-bus.h",
7617 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
7618 and "sd-utf8.h" are compile-time optional via the
7619 "--enable-kdbus" switch, and they are not compiled in by
7620 default. To make use of kdbus, you have to explicitly enable
7621 the switch. Note however, that neither the kernel nor the
7622 userspace API for all of this is considered stable yet. We
7623 want to maintain the freedom to still change the APIs for
7624 now. By specifying this build-time switch, you acknowledge
7625 that you are aware of the instability of the current
7628 * Also, note that while kdbus is pretty much complete,
7629 it lacks one thing: proper policy support. This means you
7630 can build a fully working system with all features; however,
7631 it will be highly insecure. Policy support will be added in
7632 one of the next releases, at the same time that we will
7633 declare the APIs stable.
7635 * When the kernel command line argument "kdbus" is specified,
7636 systemd will automatically load the kdbus.ko kernel module. At
7637 this stage of development, it is only useful for testing kdbus
7638 and should not be used in production. Note: if "--enable-kdbus"
7639 is specified, and the kdbus.ko kernel module is available, and
7640 "kdbus" is added to the kernel command line, the entire system
7641 runs with kdbus instead of dbus-daemon, with the above mentioned
7642 problem of missing the system policy enforcement. Also a future
7643 version of kdbus.ko or a newer systemd will not be compatible with
7644 each other, and will unlikely be able to boot the machine if only
7645 one of them is updated.
7647 * systemctl gained a new "import-environment" command which
7648 uploads the caller's environment (or parts thereof) into the
7649 service manager so that it is inherited by services started
7650 by the manager. This is useful to upload variables like
7651 $DISPLAY into the user service manager.
7653 * A new PrivateDevices= switch has been added to service units
7654 which allows running a service with a namespaced /dev
7655 directory that does not contain any device nodes for
7656 physical devices. More specifically, it only includes devices
7657 such as /dev/null, /dev/urandom, and /dev/zero which are API
7660 * logind has been extended to support behaviour like VT
7661 switching on seats that do not support a VT. This makes
7662 multi-session available on seats that are not the first seat
7663 (seat0), and on systems where kernel support for VTs has
7664 been disabled at compile-time.
7666 * If a process holds a delay lock for system sleep or shutdown
7667 and fails to release it in time, we will now log its
7668 identity. This makes it easier to identify processes that
7669 cause slow suspends or power-offs.
7671 * When parsing /etc/crypttab, support for a new key-slot=
7672 option as supported by Debian is added. It allows indicating
7673 which LUKS slot to use on disk, speeding up key loading.
7675 * The sd_journal_sendv() API call has been checked and
7676 officially declared to be async-signal-safe so that it may
7677 be invoked from signal handlers for logging purposes.
7679 * Boot-time status output is now enabled automatically after a
7680 short timeout if boot does not progress, in order to give
7681 the user an indication what she or he is waiting for.
7683 * The boot-time output has been improved to show how much time
7684 remains until jobs expire.
7686 * The KillMode= switch in service units gained a new possible
7687 value "mixed". If set, and the unit is shut down, then the
7688 initial SIGTERM signal is sent only to the main daemon
7689 process, while the following SIGKILL signal is sent to
7690 all remaining processes of the service.
7692 * When a scope unit is registered, a new property "Controller"
7693 may be set. If set to a valid bus name, systemd will send a
7694 RequestStop() signal to this name when it would like to shut
7695 down the scope. This may be used to hook manager logic into
7696 the shutdown logic of scope units. Also, scope units may now
7697 be put in a special "abandoned" state, in which case the
7698 manager process which created them takes no further
7699 responsibilities for it.
7701 * When reading unit files, systemd will now verify
7702 the access mode of these files, and warn about certain
7703 suspicious combinations. This has been added to make it
7704 easier to track down packaging bugs where unit files are
7705 marked executable or world-writable.
7707 * systemd-nspawn gained a new "--setenv=" switch to set
7708 container-wide environment variables. The similar option in
7709 systemd-activate was renamed from "--environment=" to
7710 "--setenv=" for consistency.
7712 * systemd-nspawn has been updated to create a new kdbus domain
7713 for each container that is invoked, thus allowing each
7714 container to have its own set of system and user buses,
7715 independent of the host.
7717 * systemd-nspawn gained a new --drop-capability= switch to run
7718 the container with less capabilities than the default. Both
7719 --drop-capability= and --capability= now take the special
7720 string "all" for dropping or keeping all capabilities.
7722 * systemd-nspawn gained new switches for executing containers
7723 with specific SELinux labels set.
7725 * systemd-nspawn gained a new --quiet switch to not generate
7726 any additional output but the container's own console
7729 * systemd-nspawn gained a new --share-system switch to run a
7730 container without PID namespacing enabled.
7732 * systemd-nspawn gained a new --register= switch to control
7733 whether the container is registered with systemd-machined or
7734 not. This is useful for containers that do not run full
7735 OS images, but only specific apps.
7737 * systemd-nspawn gained a new --keep-unit which may be used
7738 when invoked as the only program from a service unit, and
7739 results in registration of the unit service itself in
7740 systemd-machined, instead of a newly opened scope unit.
7742 * systemd-nspawn gained a new --network-interface= switch for
7743 moving arbitrary interfaces to the container. The new
7744 --network-veth switch creates a virtual Ethernet connection
7745 between host and container. The new --network-bridge=
7746 switch then allows assigning the host side of this virtual
7747 Ethernet connection to a bridge device.
7749 * systemd-nspawn gained a new --personality= switch for
7750 setting the kernel personality for the container. This is
7751 useful when running a 32-bit container on a 64-bit host. A
7752 similar option Personality= is now also available for service
7755 * logind will now also track a "Desktop" identifier for each
7756 session which encodes the desktop environment of it. This is
7757 useful for desktop environments that want to identify
7758 multiple running sessions of itself easily.
7760 * A new SELinuxContext= setting for service units has been
7761 added that allows setting a specific SELinux execution
7762 context for a service.
7764 * Most systemd client tools will now honour $SYSTEMD_LESS for
7765 settings of the "less" pager. By default, these tools will
7766 override $LESS to allow certain operations to work, such as
7767 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
7768 influence this logic.
7770 * systemd's "seccomp" hook-up has been changed to make use of
7771 the libseccomp library instead of using its own
7772 implementation. This has benefits for portability among
7775 * For usage together with SystemCallFilter=, a new
7776 SystemCallErrorNumber= setting has been introduced that
7777 allows configuration of a system error number to be returned
7778 on filtered system calls, instead of immediately killing the
7779 process. Also, SystemCallArchitectures= has been added to
7780 limit access to system calls of a particular architecture
7781 (in order to turn off support for unused secondary
7782 architectures). There is also a global
7783 SystemCallArchitectures= setting in system.conf now to turn
7784 off support for non-native system calls system-wide.
7786 * systemd requires a kernel with a working name_to_handle_at(),
7787 please see the kernel config requirements in the README file.
7789 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
7790 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
7791 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
7792 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
7793 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
7794 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
7795 Elia Pinto, Florian Weimer, George McCollister, Goffredo
7796 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
7797 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
7798 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
7799 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
7800 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
7801 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
7802 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
7803 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
7804 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
7805 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
7806 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
7807 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
7808 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
7809 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
7810 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
7811 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
7812 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
7814 — Berlin, 2014-02-20
7818 * logind has gained support for facilitating privileged input
7819 and drm device access for unprivileged clients. This work is
7820 useful to allow Wayland display servers (and similar
7821 programs, such as kmscon) to run under the user's ID and
7822 access input and drm devices which are normally
7823 protected. When this is used (and the kernel is new enough)
7824 logind will "mute" IO on the file descriptors passed to
7825 Wayland as long as it is in the background and "unmute" it
7826 if it returns into the foreground. This allows secure
7827 session switching without allowing background sessions to
7828 eavesdrop on input and display data. This also introduces
7829 session switching support if VT support is turned off in the
7830 kernel, and on seats that are not seat0.
7832 * A new kernel command line option luks.options= is understood
7833 now which allows specifying LUKS options for usage for LUKS
7834 encrypted partitions specified with luks.uuid=.
7836 * tmpfiles.d(5) snippets may now use specifier expansion in
7837 path names. More specifically %m, %b, %H, %v, are now
7838 replaced by the local machine id, boot id, hostname, and
7839 kernel version number.
7841 * A new tmpfiles.d(5) command "m" has been introduced which
7842 may be used to change the owner/group/access mode of a file
7843 or directory if it exists, but do nothing if it does not.
7845 * This release removes high-level support for the
7846 MemorySoftLimit= cgroup setting. The underlying kernel
7847 cgroup attribute memory.soft_limit= is currently badly
7848 designed and likely to be removed from the kernel API in its
7849 current form, hence we should not expose it for now.
7851 * The memory.use_hierarchy cgroup attribute is now enabled for
7852 all cgroups systemd creates in the memory cgroup
7853 hierarchy. This option is likely to be come the built-in
7854 default in the kernel anyway, and the non-hierarchical mode
7855 never made much sense in the intrinsically hierarchical
7858 * A new field _SYSTEMD_SLICE= is logged along with all journal
7859 messages containing the slice a message was generated
7860 from. This is useful to allow easy per-customer filtering of
7861 logs among other things.
7863 * systemd-journald will no longer adjust the group of journal
7864 files it creates to the "systemd-journal" group. Instead we
7865 rely on the journal directory to be owned by the
7866 "systemd-journal" group, and its setgid bit set, so that the
7867 kernel file system layer will automatically enforce that
7868 journal files inherit this group assignment. The reason for
7869 this change is that we cannot allow NSS look-ups from
7870 journald which would be necessary to resolve
7871 "systemd-journal" to a numeric GID, because this might
7872 create deadlocks if NSS involves synchronous queries to
7873 other daemons (such as nscd, or sssd) which in turn are
7874 logging clients of journald and might block on it, which
7875 would then dead lock. A tmpfiles.d(5) snippet included in
7876 systemd will make sure the setgid bit and group are
7877 properly set on the journal directory if it exists on every
7878 boot. However, we recommend adjusting it manually after
7879 upgrades too (or from RPM scriptlets), so that the change is
7880 not delayed until next reboot.
7882 * Backlight and random seed files in /var/lib/ have moved into
7883 the /var/lib/systemd/ directory, in order to centralize all
7884 systemd generated files in one directory.
7886 * Boot time performance measurements (as displayed by
7887 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
7888 performance information if that's available to determine how
7889 much time BIOS and boot loader initialization required. With
7890 a sufficiently new BIOS you hence no longer need to boot
7891 with Gummiboot to get access to such information.
7893 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
7894 Cristian Rodríguez, Dave Reisner, David Herrmann, David
7895 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
7896 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
7897 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
7898 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
7899 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
7901 — Berlin, 2013-10-02
7905 * The Restart= option for services now understands a new
7906 on-watchdog setting, which will restart the service
7907 automatically if the service stops sending out watchdog keep
7908 alive messages (as configured with WatchdogSec=).
7910 * The getty generator (which is responsible for bringing up a
7911 getty on configured serial consoles) will no longer only
7912 start a getty on the primary kernel console but on all
7913 others, too. This makes the order in which console= is
7914 specified on the kernel command line less important.
7916 * libsystemd-logind gained a new sd_session_get_vt() call to
7917 retrieve the VT number of a session.
7919 * If the option "tries=0" is set for an entry of /etc/crypttab
7920 its passphrase is queried indefinitely instead of any
7921 maximum number of tries.
7923 * If a service with a configure PID file terminates its PID
7924 file will now be removed automatically if it still exists
7925 afterwards. This should put an end to stale PID files.
7927 * systemd-run will now also take relative binary path names
7928 for execution and no longer insists on absolute paths.
7930 * InaccessibleDirectories= and ReadOnlyDirectories= now take
7931 paths that are optionally prefixed with "-" to indicate that
7932 it should not be considered a failure if they do not exist.
7934 * journalctl -o (and similar commands) now understands a new
7935 output mode "short-precise", it is similar to "short" but
7936 shows timestamps with usec accuracy.
7938 * The option "discard" (as known from Debian) is now
7939 synonymous to "allow-discards" in /etc/crypttab. In fact,
7940 "discard" is preferred now (since it is easier to remember
7943 * Some licensing clean-ups were made, so that more code is now
7944 LGPL-2.1 licensed than before.
7946 * A minimal tool to save/restore the display backlight
7947 brightness across reboots has been added. It will store the
7948 backlight setting as late as possible at shutdown, and
7949 restore it as early as possible during reboot.
7951 * A logic to automatically discover and enable home and swap
7952 partitions on GPT disks has been added. With this in place
7953 /etc/fstab becomes optional for many setups as systemd can
7954 discover certain partitions located on the root disk
7955 automatically. Home partitions are recognized under their
7956 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
7957 partitions are recognized under their GPT type ID
7958 0657fd6da4ab43c484e50933c84b4f4f.
7960 * systemd will no longer pass any environment from the kernel
7961 or initrd to system services. If you want to set an
7962 environment for all services, do so via the kernel command
7963 line systemd.setenv= assignment.
7965 * The systemd-sysctl tool no longer natively reads the file
7966 /etc/sysctl.conf. If desired, the file should be symlinked
7967 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
7968 legacy support by a symlink rather than built-in code, it
7969 also makes the otherwise hidden order of application of the
7970 different files visible. (Note that this partly reverts to a
7971 pre-198 application order of sysctl knobs!)
7973 * The "systemctl set-log-level" and "systemctl dump" commands
7974 have been moved to systemd-analyze.
7976 * systemd-run learned the new --remain-after-exit switch,
7977 which causes the scope unit not to be cleaned up
7978 automatically after the process terminated.
7980 * tmpfiles learned a new --exclude-prefix= switch to exclude
7981 certain paths from operation.
7983 * journald will now automatically flush all messages to disk
7984 as soon as a message at the log level CRIT, ALERT or EMERG
7987 Contributions from: Andrew Cook, Brandon Philips, Christian
7988 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
7989 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
7990 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
7991 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
7992 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
7993 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
7994 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
7995 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
7996 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
7997 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
7998 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
7999 William Giokas, Zbigniew Jędrzejewski-Szmek
8001 — Berlin, 2013-09-13
8005 * The documentation has been updated to cover the various new
8006 concepts introduced with 205.
8008 * Unit files now understand the new %v specifier which
8009 resolves to the kernel version string as returned by "uname
8012 * systemctl now supports filtering the unit list output by
8013 load state, active state and sub state, using the new
8016 * "systemctl status" will now show the results of the
8017 condition checks (like ConditionPathExists= and similar) of
8018 the last start attempts of the unit. They are also logged to
8021 * "journalctl -b" may now be used to look for boot output of a
8022 specific boot. Try "journalctl -b -1" for the previous boot,
8023 but the syntax is substantially more powerful.
8025 * "journalctl --show-cursor" has been added which prints the
8026 cursor string the last shown log line. This may then be used
8027 with the new "journalctl --after-cursor=" switch to continue
8028 browsing logs from that point on.
8030 * "journalctl --force" may now be used to force regeneration
8033 * Creation of "dead" device nodes has been moved from udev
8034 into kmod and tmpfiles. Previously, udev would read the kmod
8035 databases to pre-generate dead device nodes based on meta
8036 information contained in kernel modules, so that these would
8037 be auto-loaded on access rather then at boot. As this
8038 does not really have much to do with the exposing actual
8039 kernel devices to userspace this has always been slightly
8040 alien in the udev codebase. Following the new scheme kmod
8041 will now generate a runtime snippet for tmpfiles from the
8042 module meta information and it now is tmpfiles' job to the
8043 create the nodes. This also allows overriding access and
8044 other parameters for the nodes using the usual tmpfiles
8045 facilities. As side effect this allows us to remove the
8046 CAP_SYS_MKNOD capability bit from udevd entirely.
8048 * logind's device ACLs may now be applied to these "dead"
8049 devices nodes too, thus finally allowing managed access to
8050 devices such as /dev/snd/sequencer without loading the
8051 backing module right-away.
8053 * A new RPM macro has been added that may be used to apply
8054 tmpfiles configuration during package installation.
8056 * systemd-detect-virt and ConditionVirtualization= now can
8057 detect User-Mode-Linux machines (UML).
8059 * journald will now implicitly log the effective capabilities
8060 set of processes in the message metadata.
8062 * systemd-cryptsetup has gained support for TrueCrypt volumes.
8064 * The initrd interface has been simplified (more specifically,
8065 support for passing performance data via environment
8066 variables and fsck results via files in /run has been
8067 removed). These features were non-essential, and are
8068 nowadays available in a much nicer way by having systemd in
8069 the initrd serialize its state and have the hosts systemd
8070 deserialize it again.
8072 * The udev "keymap" data files and tools to apply keyboard
8073 specific mappings of scan to key codes, and force-release
8074 scan code lists have been entirely replaced by a udev
8075 "keyboard" builtin and a hwdb data file.
8077 * systemd will now honour the kernel's "quiet" command line
8078 argument also during late shutdown, resulting in a
8079 completely silent shutdown when used.
8081 * There's now an option to control the SO_REUSEPORT socket
8082 option in .socket units.
8084 * Instance units will now automatically get a per-template
8085 subslice of system.slice unless something else is explicitly
8086 configured. For example, instances of sshd@.service will now
8087 implicitly be placed in system-sshd.slice rather than
8088 system.slice as before.
8090 * Test coverage support may now be enabled at build time.
8092 Contributions from: Dave Reisner, Frederic Crozat, Harald
8093 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
8094 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
8095 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
8096 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
8097 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
8098 Giokas, Zbigniew Jędrzejewski-Szmek
8100 — Berlin, 2013-07-23
8104 * Two new unit types have been introduced:
8106 Scope units are very similar to service units, however, are
8107 created out of pre-existing processes — instead of PID 1
8108 forking off the processes. By using scope units it is
8109 possible for system services and applications to group their
8110 own child processes (worker processes) in a powerful way
8111 which then maybe used to organize them, or kill them
8112 together, or apply resource limits on them.
8114 Slice units may be used to partition system resources in an
8115 hierarchical fashion and then assign other units to them. By
8116 default there are now three slices: system.slice (for all
8117 system services), user.slice (for all user sessions),
8118 machine.slice (for VMs and containers).
8120 Slices and scopes have been introduced primarily in
8121 context of the work to move cgroup handling to a
8122 single-writer scheme, where only PID 1
8123 creates/removes/manages cgroups.
8125 * There's a new concept of "transient" units. In contrast to
8126 normal units these units are created via an API at runtime,
8127 not from configuration from disk. More specifically this
8128 means it is now possible to run arbitrary programs as
8129 independent services, with all execution parameters passed
8130 in via bus APIs rather than read from disk. Transient units
8131 make systemd substantially more dynamic then it ever was,
8132 and useful as a general batch manager.
8134 * logind has been updated to make use of scope and slice units
8135 for managing user sessions. As a user logs in he will get
8136 his own private slice unit, to which all sessions are added
8137 as scope units. We also added support for automatically
8138 adding an instance of user@.service for the user into the
8139 slice. Effectively logind will no longer create cgroup
8140 hierarchies on its own now, it will defer entirely to PID 1
8141 for this by means of scope, service and slice units. Since
8142 user sessions this way become entities managed by PID 1
8143 the output of "systemctl" is now a lot more comprehensive.
8145 * A new mini-daemon "systemd-machined" has been added which
8146 may be used by virtualization managers to register local
8147 VMs/containers. nspawn has been updated accordingly, and
8148 libvirt will be updated shortly. machined will collect a bit
8149 of meta information about the VMs/containers, and assign
8150 them their own scope unit (see above). The collected
8151 meta-data is then made available via the "machinectl" tool,
8152 and exposed in "ps" and similar tools. machined/machinectl
8153 is compile-time optional.
8155 * As discussed earlier, the low-level cgroup configuration
8156 options ControlGroup=, ControlGroupModify=,
8157 ControlGroupPersistent=, ControlGroupAttribute= have been
8158 removed. Please use high-level attribute settings instead as
8159 well as slice units.
8161 * A new bus call SetUnitProperties() has been added to alter
8162 various runtime parameters of a unit. This is primarily
8163 useful to alter cgroup parameters dynamically in a nice way,
8164 but will be extended later on to make more properties
8165 modifiable at runtime. systemctl gained a new set-properties
8166 command that wraps this call.
8168 * A new tool "systemd-run" has been added which can be used to
8169 run arbitrary command lines as transient services or scopes,
8170 while configuring a number of settings via the command
8171 line. This tool is currently very basic, however already
8172 very useful. We plan to extend this tool to even allow
8173 queuing of execution jobs with time triggers from the
8174 command line, similar in fashion to "at".
8176 * nspawn will now inform the user explicitly that kernels with
8177 audit enabled break containers, and suggest the user to turn
8180 * Support for detecting the IMA and AppArmor security
8181 frameworks with ConditionSecurity= has been added.
8183 * journalctl gained a new "-k" switch for showing only kernel
8184 messages, mimicking dmesg output; in addition to "--user"
8185 and "--system" switches for showing only user's own logs
8188 * systemd-delta can now show information about drop-in
8189 snippets extending unit files.
8191 * libsystemd-bus has been substantially updated but is still
8192 not available as public API.
8194 * systemd will now look for the "debug" argument on the kernel
8195 command line and enable debug logging, similar to what
8196 "systemd.log_level=debug" already did before.
8198 * "systemctl set-default", "systemctl get-default" has been
8199 added to configure the default.target symlink, which
8200 controls what to boot into by default.
8202 * "systemctl set-log-level" has been added as a convenient
8203 way to raise and lower systemd logging threshold.
8205 * "systemd-analyze plot" will now show the time the various
8206 generators needed for execution, as well as information
8207 about the unit file loading.
8209 * libsystemd-journal gained a new sd_journal_open_files() call
8210 for opening specific journal files. journactl also gained a
8211 new switch to expose this new functionality. Previously we
8212 only supported opening all files from a directory, or all
8213 files from the system, as opening individual files only is
8214 racy due to journal file rotation.
8216 * systemd gained the new DefaultEnvironment= setting in
8217 /etc/systemd/system.conf to set environment variables for
8220 * If a privileged process logs a journal message with the
8221 OBJECT_PID= field set, then journald will automatically
8222 augment this with additional OBJECT_UID=, OBJECT_GID=,
8223 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
8224 system services want to log events about specific client
8225 processes. journactl/systemctl has been updated to make use
8226 of this information if all log messages regarding a specific
8229 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
8230 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
8231 Reisner, David Coppa, David King, David Strauss, Eelco
8232 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
8233 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
8234 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
8235 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
8236 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
8237 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
8238 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
8239 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
8240 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
8241 Łukasz Stelmach, 장동준
8245 * The Python bindings gained some minimal support for the APIs
8246 exposed by libsystemd-logind.
8248 * ConditionSecurity= gained support for detecting SMACK. Since
8249 this condition already supports SELinux and AppArmor we only
8250 miss IMA for this. Patches welcome!
8252 Contributions from: Karol Lewandowski, Lennart Poettering,
8253 Zbigniew Jędrzejewski-Szmek
8257 * systemd-nspawn will now create /etc/resolv.conf if
8258 necessary, before bind-mounting the host's file onto it.
8260 * systemd-nspawn will now store meta information about a
8261 container on the container's cgroup as extended attribute
8262 fields, including the root directory.
8264 * The cgroup hierarchy has been reworked in many ways. All
8265 objects any of the components systemd creates in the cgroup
8266 tree are now suffixed. More specifically, user sessions are
8267 now placed in cgroups suffixed with ".session", users in
8268 cgroups suffixed with ".user", and nspawn containers in
8269 cgroups suffixed with ".nspawn". Furthermore, all cgroup
8270 names are now escaped in a simple scheme to avoid collision
8271 of userspace object names with kernel filenames. This work
8272 is preparation for making these objects relocatable in the
8273 cgroup tree, in order to allow easy resource partitioning of
8274 these objects without causing naming conflicts.
8276 * systemctl list-dependencies gained the new switches
8277 --plain, --reverse, --after and --before.
8279 * systemd-inhibit now shows the process name of processes that
8280 have taken an inhibitor lock.
8282 * nss-myhostname will now also resolve "localhost"
8283 implicitly. This makes /etc/hosts an optional file and
8284 nicely handles that on IPv6 ::1 maps to both "localhost" and
8287 * libsystemd-logind.so gained a new call
8288 sd_get_machine_names() to enumerate running containers and
8289 VMs (currently only supported by very new libvirt and
8290 nspawn). sd_login_monitor can now be used to watch
8291 VMs/containers coming and going.
8293 * .include is not allowed recursively anymore, and only in
8294 unit files. Usually it is better to use drop-in snippets in
8295 .d/*.conf anyway, as introduced with systemd 198.
8297 * systemd-analyze gained a new "critical-chain" command that
8298 determines the slowest chain of units run during system
8299 boot-up. It is very useful for tracking down where
8300 optimizing boot time is the most beneficial.
8302 * systemd will no longer allow manipulating service paths in
8303 the name=systemd:/system cgroup tree using ControlGroup= in
8304 units. (But is still fine with it in all other dirs.)
8306 * There's a new systemd-nspawn@.service service file that may
8307 be used to easily run nspawn containers as system
8308 services. With the container's root directory in
8309 /var/lib/container/foobar it is now sufficient to run
8310 "systemctl start systemd-nspawn@foobar.service" to boot it.
8312 * systemd-cgls gained a new parameter "--machine" to list only
8313 the processes within a certain container.
8315 * ConditionSecurity= now can check for "apparmor". We still
8316 are lacking checks for SMACK and IMA for this condition
8317 check though. Patches welcome!
8319 * A new configuration file /etc/systemd/sleep.conf has been
8320 added that may be used to configure which kernel operation
8321 systemd is supposed to execute when "suspend", "hibernate"
8322 or "hybrid-sleep" is requested. This makes the new kernel
8323 "freeze" state accessible to the user.
8325 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
8326 the passed argument if applicable.
8328 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
8329 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
8330 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
8331 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
8332 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
8333 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
8334 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
8339 * The output of 'systemctl list-jobs' got some polishing. The
8340 '--type=' argument may now be passed more than once. A new
8341 command 'systemctl list-sockets' has been added which shows
8342 a list of kernel sockets systemd is listening on with the
8343 socket units they belong to, plus the units these socket
8346 * The experimental libsystemd-bus library got substantial
8347 updates to work in conjunction with the (also experimental)
8348 kdbus kernel project. It works well enough to exchange
8349 messages with some sophistication. Note that kdbus is not
8350 ready yet, and the library is mostly an elaborate test case
8351 for now, and not installable.
8353 * systemd gained a new unit 'systemd-static-nodes.service'
8354 that generates static device nodes earlier during boot, and
8355 can run in conjunction with udev.
8357 * libsystemd-login gained a new call sd_pid_get_user_unit()
8358 to retrieve the user systemd unit a process is running
8359 in. This is useful for systems where systemd is used as
8362 * systemd-nspawn now places all containers in the new /machine
8363 top-level cgroup directory in the name=systemd
8364 hierarchy. libvirt will soon do the same, so that we get a
8365 uniform separation of /system, /user and /machine for system
8366 services, user processes and containers/virtual
8367 machines. This new cgroup hierarchy is also useful to stick
8368 stable names to specific container instances, which can be
8369 recognized later this way (this name may be controlled
8370 via systemd-nspawn's new -M switch). libsystemd-login also
8371 gained a new call sd_pid_get_machine_name() to retrieve the
8372 name of the container/VM a specific process belongs to.
8374 * bootchart can now store its data in the journal.
8376 * libsystemd-journal gained a new call
8377 sd_journal_add_conjunction() for AND expressions to the
8378 matching logic. This can be used to express more complex
8379 logical expressions.
8381 * journactl can now take multiple --unit= and --user-unit=
8384 * The cryptsetup logic now understands the "luks.key=" kernel
8385 command line switch for specifying a file to read the
8386 decryption key from. Also, if a configured key file is not
8387 found the tool will now automatically fall back to prompting
8390 * Python systemd.journal module was updated to wrap recently
8391 added functions from libsystemd-journal. The interface was
8392 changed to bring the low level interface in s.j._Reader
8393 closer to the C API, and the high level interface in
8394 s.j.Reader was updated to wrap and convert all data about
8397 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
8398 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
8399 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
8400 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
8401 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
8402 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
8406 * journalctl --update-catalog now understands a new --root=
8407 option to operate on catalogs found in a different root
8410 * During shutdown after systemd has terminated all running
8411 services a final killing loop kills all remaining left-over
8412 processes. We will now print the name of these processes
8413 when we send SIGKILL to them, since this usually indicates a
8416 * If /etc/crypttab refers to password files stored on
8417 configured mount points automatic dependencies will now be
8418 generated to ensure the specific mount is established first
8419 before the key file is attempted to be read.
8421 * 'systemctl status' will now show information about the
8422 network sockets a socket unit is listening on.
8424 * 'systemctl status' will also shown information about any
8425 drop-in configuration file for units. (Drop-In configuration
8426 files in this context are files such as
8427 /etc/systemd/systemd/foobar.service.d/*.conf)
8429 * systemd-cgtop now optionally shows summed up CPU times of
8430 cgroups. Press '%' while running cgtop to switch between
8431 percentage and absolute mode. This is useful to determine
8432 which cgroups use up the most CPU time over the entire
8433 runtime of the system. systemd-cgtop has also been updated
8434 to be 'pipeable' for processing with further shell tools.
8436 * 'hostnamectl set-hostname' will now allow setting of FQDN
8439 * The formatting and parsing of time span values has been
8440 changed. The parser now understands fractional expressions
8441 such as "5.5h". The formatter will now output fractional
8442 expressions for all time spans under 1min, i.e. "5.123456s"
8443 rather than "5s 123ms 456us". For time spans under 1s
8444 millisecond values are shown, for those under 1ms
8445 microsecond values are shown. This should greatly improve
8446 all time-related output of systemd.
8448 * libsystemd-login and libsystemd-journal gained new
8449 functions for querying the poll() events mask and poll()
8450 timeout value for integration into arbitrary event
8453 * localectl gained the ability to list available X11 keymaps
8454 (models, layouts, variants, options).
8456 * 'systemd-analyze dot' gained the ability to filter for
8457 specific units via shell-style globs, to create smaller,
8458 more useful graphs. I.e. it is now possible to create simple
8459 graphs of all the dependencies between only target units, or
8460 of all units that Avahi has dependencies with.
8462 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
8463 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
8464 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
8465 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
8466 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
8467 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
8468 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
8472 * The boot-time readahead implementation for rotating media
8473 will now read the read-ahead data in multiple passes which
8474 consist of all read requests made in equidistant time
8475 intervals. This means instead of strictly reading read-ahead
8476 data in its physical order on disk we now try to find a
8477 middle ground between physical and access time order.
8479 * /etc/os-release files gained a new BUILD_ID= field for usage
8480 on operating systems that provide continuous builds of OS
8483 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
8484 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
8485 William Douglas, Zbigniew Jędrzejewski-Szmek
8489 * systemd-python gained an API exposing libsystemd-daemon.
8491 * The SMACK setup logic gained support for uploading CIPSO
8494 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
8495 ReadOnlyDirectories= and InaccessibleDirectories= has
8496 changed. The private /tmp and /var/tmp directories are now
8497 shared by all processes of a service (which means
8498 ExecStartPre= may now leave data in /tmp that ExecStart= of
8499 the same service can still access). When a service is
8500 stopped its temporary directories are immediately deleted
8501 (normal clean-up with tmpfiles is still done in addition to
8504 * By default, systemd will now set a couple of sysctl
8505 variables in the kernel: the safe sysrq options are turned
8506 on, IP route verification is turned on, and source routing
8507 disabled. The recently added hardlink and softlink
8508 protection of the kernel is turned on. These settings should
8509 be reasonably safe, and good defaults for all new systems.
8511 * The predictable network naming logic may now be turned off
8512 with a new kernel command line switch: net.ifnames=0.
8514 * A new libsystemd-bus module has been added that implements a
8515 pretty complete D-Bus client library. For details see:
8517 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
8519 * journald will now explicitly flush the journal files to disk
8520 at the latest 5min after each write. The file will then also
8521 be marked offline until the next write. This should increase
8522 reliability in case of a crash. The synchronization delay
8523 can be configured via SyncIntervalSec= in journald.conf.
8525 * There's a new remote-fs-setup.target unit that can be used
8526 to pull in specific services when at least one remote file
8527 system is to be mounted.
8529 * There are new targets timers.target and paths.target as
8530 canonical targets to pull user timer and path units in
8531 from. This complements sockets.target with a similar
8532 purpose for socket units.
8534 * libudev gained a new call udev_device_set_attribute_value()
8535 to set sysfs attributes of a device.
8537 * The udev daemon now sets the default number of worker
8538 processes executed in parallel based on the number of available
8539 CPUs instead of the amount of available RAM. This is supposed
8540 to provide a more reliable default and limit a too aggressive
8541 parallelism for setups with 1000s of devices connected.
8543 Contributions from: Auke Kok, Colin Walters, Cristian
8544 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
8545 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
8546 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
8547 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
8548 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
8549 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
8550 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
8551 Zbigniew Jędrzejewski-Szmek
8555 * Configuration of unit files may now be extended via drop-in
8556 files without having to edit/override the unit files
8557 themselves. More specifically, if the administrator wants to
8558 change one value for a service file foobar.service he can
8559 now do so by dropping in a configuration snippet into
8560 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
8561 will load all these snippets and apply them on top of the
8562 main unit configuration file, possibly extending or
8563 overriding its settings. Using these drop-in snippets is
8564 generally nicer than the two earlier options for changing
8565 unit files locally: copying the files from
8566 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
8567 them there; or creating a new file in /etc/systemd/system/
8568 that incorporates the original one via ".include". Drop-in
8569 snippets into these .d/ directories can be placed in any
8570 directory systemd looks for units in, and the usual
8571 overriding semantics between /usr/lib, /etc and /run apply
8574 * Most unit file settings which take lists of items can now be
8575 reset by assigning the empty string to them. For example,
8576 normally, settings such as Environment=FOO=BAR append a new
8577 environment variable assignment to the environment block,
8578 each time they are used. By assigning Environment= the empty
8579 string the environment block can be reset to empty. This is
8580 particularly useful with the .d/*.conf drop-in snippets
8581 mentioned above, since this adds the ability to reset list
8582 settings from vendor unit files via these drop-ins.
8584 * systemctl gained a new "list-dependencies" command for
8585 listing the dependencies of a unit recursively.
8587 * Inhibitors are now honored and listed by "systemctl
8588 suspend", "systemctl poweroff" (and similar) too, not only
8589 GNOME. These commands will also list active sessions by
8592 * Resource limits (as exposed by the various control group
8593 controllers) can now be controlled dynamically at runtime
8594 for all units. More specifically, you can now use a command
8595 like "systemctl set-cgroup-attr foobar.service cpu.shares
8596 2000" to alter the CPU shares a specific service gets. These
8597 settings are stored persistently on disk, and thus allow the
8598 administrator to easily adjust the resource usage of
8599 services with a few simple commands. This dynamic resource
8600 management logic is also available to other programs via the
8601 bus. Almost any kernel cgroup attribute and controller is
8604 * systemd-vconsole-setup will now copy all font settings to
8605 all allocated VTs, where it previously applied them only to
8608 * libsystemd-login gained the new sd_session_get_tty() API
8611 * This release drops support for a few legacy or
8612 distribution-specific LSB facility names when parsing init
8613 scripts: $x-display-manager, $mail-transfer-agent,
8614 $mail-transport-agent, $mail-transfer-agent, $smtp,
8615 $null. Also, the mail-transfer-agent.target unit backing
8616 this has been removed. Distributions which want to retain
8617 compatibility with this should carry the burden for
8618 supporting this themselves and patch support for these back
8619 in, if they really need to. Also, the facilities $syslog and
8620 $local_fs are now ignored, since systemd does not support
8621 early-boot LSB init scripts anymore, and these facilities
8622 are implied anyway for normal services. syslog.target has
8625 * There are new bus calls on PID1's Manager object for
8626 cancelling jobs, and removing snapshot units. Previously,
8627 both calls were only available on the Job and Snapshot
8630 * systemd-journal-gatewayd gained SSL support.
8632 * The various "environment" files, such as /etc/locale.conf
8633 now support continuation lines with a backslash ("\") as
8634 last character in the line, similarly in style (but different)
8635 to how this is supported in shells.
8637 * For normal user processes the _SYSTEMD_USER_UNIT= field is
8638 now implicitly appended to every log entry logged. systemctl
8639 has been updated to filter by this field when operating on a
8640 user systemd instance.
8642 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
8643 CAP_AUDIT_CONTROL capabilities to the capabilities set for
8644 the container. This makes it easier to boot unmodified
8645 Fedora systems in a container, which however still requires
8646 audit=0 to be passed on the kernel command line. Auditing in
8647 kernel and userspace is unfortunately still too broken in
8648 context of containers, hence we recommend compiling it out
8649 of the kernel or using audit=0. Hopefully this will be fixed
8650 one day for good in the kernel.
8652 * nspawn gained the new --bind= and --bind-ro= parameters to
8653 bind mount specific directories from the host into the
8656 * nspawn will now mount its own devpts file system instance
8657 into the container, in order not to leak pty devices from
8658 the host into the container.
8660 * systemd will now read the firmware boot time performance
8661 information from the EFI variables, if the used boot loader
8662 supports this, and takes it into account for boot performance
8663 analysis via "systemd-analyze". This is currently supported
8664 only in conjunction with Gummiboot, but could be supported
8665 by other boot loaders too. For details see:
8667 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
8669 * A new generator has been added that automatically mounts the
8670 EFI System Partition (ESP) to /boot, if that directory
8671 exists, is empty, and no other file system has been
8672 configured to be mounted there.
8674 * logind will now send out PrepareForSleep(false) out
8675 unconditionally, after coming back from suspend. This may be
8676 used by applications as asynchronous notification for
8677 system resume events.
8679 * "systemctl unlock-sessions" has been added, that allows
8680 unlocking the screens of all user sessions at once, similar
8681 to how "systemctl lock-sessions" already locked all users
8682 sessions. This is backed by a new D-Bus call UnlockSessions().
8684 * "loginctl seat-status" will now show the master device of a
8685 seat. (i.e. the device of a seat that needs to be around for
8686 the seat to be considered available, usually the graphics
8689 * tmpfiles gained a new "X" line type, that allows
8690 configuration of files and directories (with wildcards) that
8691 shall be excluded from automatic cleanup ("aging").
8693 * udev default rules set the device node permissions now only
8694 at "add" events, and do not change them any longer with a
8695 later "change" event.
8697 * The log messages for lid events and power/sleep keypresses
8698 now carry a message ID.
8700 * We now have a substantially larger unit test suite, but this
8701 continues to be work in progress.
8703 * udevadm hwdb gained a new --root= parameter to change the
8704 root directory to operate relative to.
8706 * logind will now issue a background sync() request to the kernel
8707 early at shutdown, so that dirty buffers are flushed to disk early
8708 instead of at the last moment, in order to optimize shutdown
8711 * A new bootctl tool has been added that is an interface for
8712 certain boot loader operations. This is currently a preview
8713 and is likely to be extended into a small mechanism daemon
8714 like timedated, localed, hostnamed, and can be used by
8715 graphical UIs to enumerate available boot options, and
8716 request boot into firmware operations.
8718 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
8719 the rest of the package. It also has been updated to work
8720 correctly in initrds.
8722 * polkit previously has been runtime optional, and is now also
8723 compile time optional via a configure switch.
8725 * systemd-analyze has been reimplemented in C. Also "systemctl
8726 dot" has moved into systemd-analyze.
8728 * "systemctl status" with no further parameters will now print
8729 the status of all active or failed units.
8731 * Operations such as "systemctl start" can now be executed
8732 with a new mode "--irreversible" which may be used to queue
8733 operations that cannot accidentally be reversed by a later
8734 job queuing. This is by default used to make shutdown
8735 requests more robust.
8737 * The Python API of systemd now gained a new module for
8738 reading journal files.
8740 * A new tool kernel-install has been added that can install
8741 kernel images according to the Boot Loader Specification:
8743 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
8745 * Boot time console output has been improved to provide
8746 animated boot time output for hanging jobs.
8748 * A new tool systemd-activate has been added which can be used
8749 to test socket activation with, directly from the command
8750 line. This should make it much easier to test and debug
8751 socket activation in daemons.
8753 * journalctl gained a new "--reverse" (or -r) option to show
8754 journal output in reverse order (i.e. newest line first).
8756 * journalctl gained a new "--pager-end" (or -e) option to jump
8757 to immediately jump to the end of the journal in the
8758 pager. This is only supported in conjunction with "less".
8760 * journalctl gained a new "--user-unit=" option, that works
8761 similarly to "--unit=" but filters for user units rather than
8764 * A number of unit files to ease adoption of systemd in
8765 initrds has been added. This moves some minimal logic from
8766 the various initrd implementations into systemd proper.
8768 * The journal files are now owned by a new group
8769 "systemd-journal", which exists specifically to allow access
8770 to the journal, and nothing else. Previously, we used the
8771 "adm" group for that, which however possibly covers more
8772 than just journal/log file access. This new group is now
8773 already used by systemd-journal-gatewayd to ensure this
8774 daemon gets access to the journal files and as little else
8775 as possible. Note that "make install" will also set FS ACLs
8776 up for /var/log/journal to give "adm" and "wheel" read
8777 access to it, in addition to "systemd-journal" which owns
8778 the journal files. We recommend that packaging scripts also
8779 add read access to "adm" + "wheel" to /var/log/journal, and
8780 all existing/future journal files. To normal users and
8781 administrators little changes, however packagers need to
8782 ensure to create the "systemd-journal" system group at
8783 package installation time.
8785 * The systemd-journal-gatewayd now runs as unprivileged user
8786 systemd-journal-gateway:systemd-journal-gateway. Packaging
8787 scripts need to create these system user/group at
8790 * timedated now exposes a new boolean property CanNTP that
8791 indicates whether a local NTP service is available or not.
8793 * systemd-detect-virt will now also detect xen PVs
8795 * The pstore file system is now mounted by default, if it is
8798 * In addition to the SELinux and IMA policies we will now also
8799 load SMACK policies at early boot.
8801 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
8802 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
8803 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
8804 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
8805 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
8806 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
8807 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
8808 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
8809 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
8810 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
8811 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
8812 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
8813 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
8814 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
8818 * Timer units now support calendar time events in addition to
8819 monotonic time events. That means you can now trigger a unit
8820 based on a calendar time specification such as "Thu,Fri
8821 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
8822 or fifth day of any month of the year 2013, given that it is
8823 a thursday or friday. This brings timer event support
8824 considerably closer to cron's capabilities. For details on
8825 the supported calendar time specification language see
8828 * udev now supports a number of different naming policies for
8829 network interfaces for predictable names, and a combination
8830 of these policies is now the default. Please see this wiki
8831 document for details:
8833 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
8835 * Auke Kok's bootchart implementation has been added to the
8836 systemd tree. It is an optional component that can graph the
8837 boot in quite some detail. It is one of the best bootchart
8838 implementations around and minimal in its code and
8841 * nss-myhostname has been integrated into the systemd source
8842 tree. nss-myhostname guarantees that the local hostname
8843 always stays resolvable via NSS. It has been a weak
8844 requirement of systemd-hostnamed since a long time, and
8845 since its code is actually trivial we decided to just
8846 include it in systemd's source tree. It can be turned off
8847 with a configure switch.
8849 * The read-ahead logic is now capable of properly detecting
8850 whether a btrfs file system is on SSD or rotating media, in
8851 order to optimize the read-ahead scheme. Previously, it was
8852 only capable of detecting this on traditional file systems
8855 * In udev, additional device properties are now read from the
8856 IAB in addition to the OUI database. Also, Bluetooth company
8857 identities are attached to the devices as well.
8859 * In service files %U may be used as specifier that is
8860 replaced by the configured user name of the service.
8862 * nspawn may now be invoked without a controlling TTY. This
8863 makes it suitable for invocation as its own service. This
8864 may be used to set up a simple containerized server system
8865 using only core OS tools.
8867 * systemd and nspawn can now accept socket file descriptors
8868 when they are started for socket activation. This enables
8869 implementation of socket activated nspawn
8870 containers. i.e. think about autospawning an entire OS image
8871 when the first SSH or HTTP connection is received. We expect
8872 that similar functionality will also be added to libvirt-lxc
8875 * journalctl will now suppress ANSI color codes when
8876 presenting log data.
8878 * systemctl will no longer show control group information for
8879 a unit if the control group is empty anyway.
8881 * logind can now automatically suspend/hibernate/shutdown the
8884 * /etc/machine-info and hostnamed now also expose the chassis
8885 type of the system. This can be used to determine whether
8886 the local system is a laptop, desktop, handset or
8887 tablet. This information may either be configured by the
8888 user/vendor or is automatically determined from ACPI and DMI
8889 information if possible.
8891 * A number of polkit actions are now bound together with "imply"
8892 rules. This should simplify creating UIs because many actions
8893 will now authenticate similar ones as well.
8895 * Unit files learnt a new condition ConditionACPower= which
8896 may be used to conditionalize a unit depending on whether an
8897 AC power source is connected or not, of whether the system
8898 is running on battery power.
8900 * systemctl gained a new "is-failed" verb that may be used in
8901 shell scripts and suchlike to check whether a specific unit
8902 is in the "failed" state.
8904 * The EnvironmentFile= setting in unit files now supports file
8905 globbing, and can hence be used to easily read a number of
8906 environment files at once.
8908 * systemd will no longer detect and recognize specific
8909 distributions. All distribution-specific #ifdeffery has been
8910 removed, systemd is now fully generic and
8911 distribution-agnostic. Effectively, not too much is lost as
8912 a lot of the code is still accessible via explicit configure
8913 switches. However, support for some distribution specific
8914 legacy configuration file formats has been dropped. We
8915 recommend distributions to simply adopt the configuration
8916 files everybody else uses now and convert the old
8917 configuration from packaging scripts. Most distributions
8918 already did that. If that's not possible or desirable,
8919 distributions are welcome to forward port the specific
8920 pieces of code locally from the git history.
8922 * When logging a message about a unit systemd will now always
8923 log the unit name in the message meta data.
8925 * localectl will now also discover system locale data that is
8926 not stored in locale archives, but directly unpacked.
8928 * logind will no longer unconditionally use framebuffer
8929 devices as seat masters, i.e. as devices that are required
8930 to be existing before a seat is considered preset. Instead,
8931 it will now look for all devices that are tagged as
8932 "seat-master" in udev. By default, framebuffer devices will
8933 be marked as such, but depending on local systems, other
8934 devices might be marked as well. This may be used to
8935 integrate graphics cards using closed source drivers (such
8936 as NVidia ones) more nicely into logind. Note however, that
8937 we recommend using the open source NVidia drivers instead,
8938 and no udev rules for the closed-source drivers will be
8939 shipped from us upstream.
8941 Contributions from: Adam Williamson, Alessandro Crismani, Auke
8942 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
8943 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
8944 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
8945 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
8946 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
8947 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
8948 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
8949 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
8950 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
8951 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
8952 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
8957 * udev gained support for loading additional device properties
8958 from an indexed database that is keyed by vendor/product IDs
8959 and similar device identifiers. For the beginning this
8960 "hwdb" is populated with data from the well-known PCI and
8961 USB database, but also includes PNP, ACPI and OID data. In
8962 the longer run this indexed database shall grow into
8963 becoming the one central database for non-essential
8964 userspace device metadata. Previously, data from the PCI/USB
8965 database was only attached to select devices, since the
8966 lookup was a relatively expensive operation due to O(n) time
8967 complexity (with n being the number of entries in the
8968 database). Since this is now O(1), we decided to add in this
8969 data for all devices where this is available, by
8970 default. Note that the indexed database needs to be rebuilt
8971 when new data files are installed. To achieve this you need
8972 to update your packaging scripts to invoke "udevadm hwdb
8973 --update" after installation of hwdb data files. For
8974 RPM-based distributions we introduced the new
8975 %udev_hwdb_update macro for this purpose.
8977 * The Journal gained support for the "Message Catalog", an
8978 indexed database to link up additional information with
8979 journal entries. For further details please check:
8981 https://www.freedesktop.org/wiki/Software/systemd/catalog
8983 The indexed message catalog database also needs to be
8984 rebuilt after installation of message catalog files. Use
8985 "journalctl --update-catalog" for this. For RPM-based
8986 distributions we introduced the %journal_catalog_update
8987 macro for this purpose.
8989 * The Python Journal bindings gained support for the standard
8990 Python logging framework.
8992 * The Journal API gained new functions for checking whether
8993 the underlying file system of a journal file is capable of
8994 properly reporting file change notifications, or whether
8995 applications that want to reflect journal changes "live"
8996 need to recheck journal files continuously in appropriate
8999 * It is now possible to set the "age" field for tmpfiles
9000 entries to 0, indicating that files matching this entry
9001 shall always be removed when the directories are cleaned up.
9003 * coredumpctl gained a new "gdb" verb which invokes gdb
9004 right-away on the selected coredump.
9006 * There's now support for "hybrid sleep" on kernels that
9007 support this, in addition to "suspend" and "hibernate". Use
9008 "systemctl hybrid-sleep" to make use of this.
9010 * logind's HandleSuspendKey= setting (and related settings)
9011 now gained support for a new "lock" setting to simply
9012 request the screen lock on all local sessions, instead of
9013 actually executing a suspend or hibernation.
9015 * systemd will now mount the EFI variables file system by
9018 * Socket units now gained support for configuration of the
9019 SMACK security label.
9021 * timedatectl will now output the time of the last and next
9022 daylight saving change.
9024 * We dropped support for various legacy and distro-specific
9025 concepts, such as insserv, early-boot SysV services
9026 (i.e. those for non-standard runlevels such as 'b' or 'S')
9027 or ArchLinux /etc/rc.conf support. We recommend the
9028 distributions who still need support this to either continue
9029 to maintain the necessary patches downstream, or find a
9030 different solution. (Talk to us if you have questions!)
9032 * Various systemd components will now bypass polkit checks for
9033 root and otherwise handle properly if polkit is not found to
9034 be around. This should fix most issues for polkit-less
9035 systems. Quite frankly this should have been this way since
9036 day one. It is absolutely our intention to make systemd work
9037 fine on polkit-less systems, and we consider it a bug if
9038 something does not work as it should if polkit is not around.
9040 * For embedded systems it is now possible to build udev and
9041 systemd without blkid and/or kmod support.
9043 * "systemctl switch-root" is now capable of switching root
9044 more than once. I.e. in addition to transitions from the
9045 initrd to the host OS it is now possible to transition to
9046 further OS images from the host. This is useful to implement
9047 offline updating tools.
9049 * Various other additions have been made to the RPM macros
9050 shipped with systemd. Use %udev_rules_update() after
9051 installing new udev rules files. %_udevhwdbdir,
9052 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
9053 %_sysctldir are now available which resolve to the right
9054 directories for packages to place various data files in.
9056 * journalctl gained the new --full switch (in addition to
9057 --all, to disable ellipsation for long messages.
9059 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
9060 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
9061 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
9062 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
9063 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
9064 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
9065 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
9066 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
9067 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
9071 * journalctl gained new --since= and --until= switches to
9072 filter by time. It also now supports nice filtering for
9073 units via --unit=/-u.
9075 * Type=oneshot services may use ExecReload= and do the
9078 * The journal daemon now supports time-based rotation and
9079 vacuuming, in addition to the usual disk-space based
9082 * The journal will now index the available field values for
9083 each field name. This enables clients to show pretty drop
9084 downs of available match values when filtering. The bash
9085 completion of journalctl has been updated
9086 accordingly. journalctl gained a new switch -F to list all
9087 values a certain field takes in the journal database.
9089 * More service events are now written as structured messages
9090 to the journal, and made recognizable via message IDs.
9092 * The timedated, localed and hostnamed mini-services which
9093 previously only provided support for changing time, locale
9094 and hostname settings from graphical DEs such as GNOME now
9095 also have a minimal (but very useful) text-based client
9096 utility each. This is probably the nicest way to changing
9097 these settings from the command line now, especially since
9098 it lists available options and is fully integrated with bash
9101 * There's now a new tool "systemd-coredumpctl" to list and
9102 extract coredumps from the journal.
9104 * We now install a README each in /var/log/ and
9105 /etc/rc.d/init.d explaining where the system logs and init
9106 scripts went. This hopefully should help folks who go to
9107 that dirs and look into the otherwise now empty void and
9108 scratch their heads.
9110 * When user-services are invoked (by systemd --user) the
9111 $MANAGERPID env var is set to the PID of systemd.
9113 * SIGRTMIN+24 when sent to a --user instance will now result
9114 in immediate termination of systemd.
9116 * gatewayd received numerous feature additions such as a
9117 "follow" mode, for live syncing and filtering.
9119 * browse.html now allows filtering and showing detailed
9120 information on specific entries. Keyboard navigation and
9121 mouse screen support has been added.
9123 * gatewayd/journalctl now supports HTML5/JSON
9124 Server-Sent-Events as output.
9126 * The SysV init script compatibility logic will now
9127 heuristically determine whether a script supports the
9128 "reload" verb, and only then make this available as
9131 * "systemctl status --follow" has been removed, use "journalctl
9134 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
9135 have been removed since they are hardly useful to be
9138 * And I'd like to take the opportunity to specifically mention
9139 Zbigniew for his great contributions. Zbigniew, you rock!
9141 Contributions from: Andrew Eikum, Christian Hesse, Colin
9142 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
9143 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
9144 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
9145 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
9146 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
9147 Jędrzejewski-Szmek, Сковорода Никита Андреевич
9151 * If /etc/vconsole.conf is non-existent or empty we will no
9152 longer load any console font or key map at boot by
9153 default. Instead the kernel defaults will be left
9154 intact. This is definitely the right thing to do, as no
9155 configuration should mean no configuration, and hard-coding
9156 font names that are different on all archs is probably a bad
9157 idea. Also, the kernel default key map and font should be
9158 good enough for most cases anyway, and mostly identical to
9159 the userspace fonts/key maps we previously overloaded them
9160 with. If distributions want to continue to default to a
9161 non-kernel font or key map they should ship a default
9162 /etc/vconsole.conf with the appropriate contents.
9164 Contributions from: Colin Walters, Daniel J Walsh, Dave
9165 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
9166 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
9170 * journalctl gained a new --cursor= switch to show entries
9171 starting from the specified location in the journal.
9173 * We now enforce a size limit on journal entry fields exported
9174 with "-o json" in journalctl. Fields larger than 4K will be
9175 assigned null. This can be turned off with --all.
9177 * An (optional) journal gateway daemon is now available as
9178 "systemd-journal-gatewayd.service". This service provides
9179 access to the journal via HTTP and JSON. This functionality
9180 will be used to implement live log synchronization in both
9181 pull and push modes, but has various other users too, such
9182 as easy log access for debugging of embedded devices. Right
9183 now it is already useful to retrieve the journal via HTTP:
9185 # systemctl start systemd-journal-gatewayd.service
9186 # wget http://localhost:19531/entries
9188 This will download the journal contents in a
9189 /var/log/messages compatible format. The same as JSON:
9191 # curl -H"Accept: application/json" http://localhost:19531/entries
9193 This service is also accessible via a web browser where a
9194 single static HTML5 app is served that uses the JSON logic
9195 to enable the user to do some basic browsing of the
9196 journal. This will be extended later on. Here's an example
9197 screenshot of this app in its current state:
9199 http://0pointer.de/public/journal-gatewayd
9201 Contributions from: Kay Sievers, Lennart Poettering, Robert
9202 Milasan, Tom Gundersen
9206 * The bash completion logic is now available for journalctl
9209 * We do not mount the "cpuset" controller anymore together with
9210 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
9211 started if no parameters are assigned to it. "cpuset" hence
9212 broke code that assumed it could create "cpu" groups and
9215 * journalctl -f will now subscribe to terminal size changes,
9216 and line break accordingly.
9218 Contributions from: Dave Reisner, Kay Sievers, Lennart
9219 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
9223 * nspawn will now create a symlink /etc/localtime in the
9224 container environment, copying the host's timezone
9225 setting. Previously this has been done via a bind mount, but
9226 since symlinks cannot be bind mounted this has now been
9227 changed to create/update the appropriate symlink.
9229 * journalctl -n's line number argument is now optional, and
9230 will default to 10 if omitted.
9232 * journald will now log the maximum size the journal files may
9233 take up on disk. This is particularly useful if the default
9234 built-in logic of determining this parameter from the file
9235 system size is used. Use "systemctl status
9236 systemd-journald.service" to see this information.
9238 * The multi-seat X wrapper tool has been stripped down. As X
9239 is now capable of enumerating graphics devices via udev in a
9240 seat-aware way the wrapper is not strictly necessary
9241 anymore. A stripped down temporary stop-gap is still shipped
9242 until the upstream display managers have been updated to
9243 fully support the new X logic. Expect this wrapper to be
9244 removed entirely in one of the next releases.
9246 * HandleSleepKey= in logind.conf has been split up into
9247 HandleSuspendKey= and HandleHibernateKey=. The old setting
9248 is not available anymore. X11 and the kernel are
9249 distinguishing between these keys and we should too. This
9250 also means the inhibition lock for these keys has been split
9253 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
9254 Poettering, Lukas Nykryn, Václav Pavlín
9258 * Whenever a unit changes state we will now log this to the
9259 journal and show along the unit's own log output in
9262 * ConditionPathIsMountPoint= can now properly detect bind
9263 mount points too. (Previously, a bind mount of one file
9264 system to another place in the same file system could not be
9265 detected as mount, since they shared struct stat's st_dev
9268 * We will now mount the cgroup controllers cpu, cpuacct,
9269 cpuset and the controllers net_cls, net_prio together by
9272 * nspawn containers will now have a virtualized boot
9273 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
9274 over with a randomized ID at container initialization). This
9275 has the effect of making "journalctl -b" do the right thing
9278 * The JSON output journal serialization has been updated not
9279 to generate "endless" list objects anymore, but rather one
9280 JSON object per line. This is more in line how most JSON
9281 parsers expect JSON objects. The new output mode
9282 "json-pretty" has been added to provide similar output, but
9283 neatly aligned for readability by humans.
9285 * We dropped all explicit sync() invocations in the shutdown
9286 code. The kernel does this implicitly anyway in the kernel
9287 reboot() syscall. halt(8)'s -n option is now a compatibility
9290 * We now support virtualized reboot() in containers, as
9291 supported by newer kernels. We will fall back to exit() if
9292 CAP_SYS_REBOOT is not available to the container. Also,
9293 nspawn makes use of this now and will actually reboot the
9294 container if the containerized OS asks for that.
9296 * journalctl will only show local log output by default
9297 now. Use --merge (-m) to show remote log output, too.
9299 * libsystemd-journal gained the new sd_journal_get_usage()
9300 call to determine the current disk usage of all journal
9301 files. This is exposed in the new "journalctl --disk-usage"
9304 * journald gained a new configuration setting SplitMode= in
9305 journald.conf which may be used to control how user journals
9306 are split off. See journald.conf(5) for details.
9308 * A new condition type ConditionFileNotEmpty= has been added.
9310 * tmpfiles' "w" lines now support file globbing, to write
9311 multiple files at once.
9313 * We added Python bindings for the journal submission
9314 APIs. More Python APIs for a number of selected APIs will
9315 likely follow. Note that we intend to add native bindings
9316 only for the Python language, as we consider it common
9317 enough to deserve bindings shipped within systemd. There are
9318 various projects outside of systemd that provide bindings
9319 for languages such as PHP or Lua.
9321 * Many conditions will now resolve specifiers such as %i. In
9322 addition, PathChanged= and related directives of .path units
9323 now support specifiers as well.
9325 * There's now a new RPM macro definition for the system preset
9328 * journald will now warn if it ca not forward a message to the
9329 syslog daemon because its socket is full.
9331 * timedated will no longer write or process /etc/timezone,
9332 except on Debian. As we do not support late mounted /usr
9333 anymore /etc/localtime always being a symlink is now safe,
9334 and hence the information in /etc/timezone is not necessary
9337 * logind will now always reserve one VT for a text getty (VT6
9338 by default). Previously if more than 6 X sessions where
9339 started they took up all the VTs with auto-spawned gettys,
9340 so that no text gettys were available anymore.
9342 * udev will now automatically inform the btrfs kernel logic
9343 about btrfs RAID components showing up. This should make
9344 simple hotplug based btrfs RAID assembly work.
9346 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
9347 (but not for its children which will stay at the kernel
9348 default). This should allow setups with a lot more listening
9351 * systemd will now always pass the configured timezone to the
9352 kernel at boot. timedated will do the same when the timezone
9355 * logind's inhibition logic has been updated. By default,
9356 logind will now handle the lid switch, the power and sleep
9357 keys all the time, even in graphical sessions. If DEs want
9358 to handle these events on their own they should take the new
9359 handle-power-key, handle-sleep-key and handle-lid-switch
9360 inhibitors during their runtime. A simple way to achieve
9361 that is to invoke the DE wrapped in an invocation of:
9363 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
9365 * Access to unit operations is now checked via SELinux taking
9366 the unit file label and client process label into account.
9368 * systemd will now notify the administrator in the journal
9369 when he over-mounts a non-empty directory.
9371 * There are new specifiers that are resolved in unit files,
9372 for the host name (%H), the machine ID (%m) and the boot ID
9375 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
9376 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
9377 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
9378 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
9379 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
9380 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
9381 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
9385 * Support for reading structured kernel messages from
9386 /dev/kmsg has now been added and is enabled by default.
9388 * Support for reading kernel messages from /proc/kmsg has now
9389 been removed. If you want kernel messages in the journal
9390 make sure to run a recent kernel (>= 3.5) that supports
9391 reading structured messages from /dev/kmsg (see
9392 above). /proc/kmsg is now exclusive property of classic
9393 syslog daemons again.
9395 * The libudev API gained the new
9396 udev_device_new_from_device_id() call.
9398 * The logic for file system namespace (ReadOnlyDirectory=,
9399 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
9400 require pivot_root() anymore. This means fewer temporary
9401 directories are created below /tmp for this feature.
9403 * nspawn containers will now see and receive all submounts
9404 made on the host OS below the root file system of the
9407 * Forward Secure Sealing is now supported for Journal files,
9408 which provide cryptographical sealing of journal files so
9409 that attackers cannot alter log history anymore without this
9410 being detectable. Lennart will soon post a blog story about
9411 this explaining it in more detail.
9413 * There are two new service settings RestartPreventExitStatus=
9414 and SuccessExitStatus= which allow configuration of exit
9415 status (exit code or signal) which will be excepted from the
9416 restart logic, resp. consider successful.
9418 * journalctl gained the new --verify switch that can be used
9419 to check the integrity of the structure of journal files and
9420 (if Forward Secure Sealing is enabled) the contents of
9423 * nspawn containers will now be run with /dev/stdin, /dev/fd/
9424 and similar symlinks pre-created. This makes running shells
9425 as container init process a lot more fun.
9427 * The fstab support can now handle PARTUUID= and PARTLABEL=
9430 * A new ConditionHost= condition has been added to match
9431 against the hostname (with globs) and machine ID. This is
9432 useful for clusters where a single OS image is used to
9433 provision a large number of hosts which shall run slightly
9434 different sets of services.
9436 * Services which hit the restart limit will now be placed in a
9439 Contributions from: Bertram Poettering, Dave Reisner, Huang
9440 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
9441 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
9445 * When running in --user mode systemd will now become a
9446 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
9447 tree a lot more organized.
9449 * A new PartOf= unit dependency type has been introduced that
9450 may be used to group services in a natural way.
9452 * "systemctl enable" may now be used to enable instances of
9455 * journalctl now prints error log levels in red, and
9456 warning/notice log levels in bright white. It also supports
9457 filtering by log level now.
9459 * cgtop gained a new -n switch (similar to top), to configure
9460 the maximum number of iterations to run for. It also gained
9461 -b, to run in batch mode (accepting no input).
9463 * The suffix ".service" may now be omitted on most systemctl
9464 command lines involving service unit names.
9466 * There's a new bus call in logind to lock all sessions, as
9467 well as a loginctl verb for it "lock-sessions".
9469 * libsystemd-logind.so gained a new call sd_journal_perror()
9470 that works similar to libc perror() but logs to the journal
9471 and encodes structured information about the error number.
9473 * /etc/crypttab entries now understand the new keyfile-size=
9476 * shutdown(8) now can send a (configurable) wall message when
9477 a shutdown is cancelled.
9479 * The mount propagation mode for the root file system will now
9480 default to "shared", which is useful to make containers work
9481 nicely out-of-the-box so that they receive new mounts from
9482 the host. This can be undone locally by running "mount
9483 --make-rprivate /" if needed.
9485 * The prefdm.service file has been removed. Distributions
9486 should maintain this unit downstream if they intend to keep
9487 it around. However, we recommend writing normal unit files
9488 for display managers instead.
9490 * Since systemd is a crucial part of the OS we will now
9491 default to a number of compiler switches that improve
9492 security (hardening) such as read-only relocations, stack
9493 protection, and suchlike.
9495 * The TimeoutSec= setting for services is now split into
9496 TimeoutStartSec= and TimeoutStopSec= to allow configuration
9497 of individual time outs for the start and the stop phase of
9500 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
9501 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
9502 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
9503 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
9504 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
9505 Gundersen, Zbigniew Jędrzejewski-Szmek
9509 * The journal and id128 C APIs are now fully documented as man
9512 * Extra safety checks have been added when transitioning from
9513 the initial RAM disk to the main system to avoid accidental
9516 * /etc/crypttab entries now understand the new keyfile-offset=
9519 * systemctl -t can now be used to filter by unit load state.
9521 * The journal C API gained the new sd_journal_wait() call to
9522 make writing synchronous journal clients easier.
9524 * journalctl gained the new -D switch to show journals from a
9527 * journalctl now displays a special marker between log
9528 messages of two different boots.
9530 * The journal is now explicitly flushed to /var via a service
9531 systemd-journal-flush.service, rather than implicitly simply
9532 by seeing /var/log/journal to be writable.
9534 * journalctl (and the journal C APIs) can now match for much
9535 more complex expressions, with alternatives and
9538 * When transitioning from the initial RAM disk to the main
9539 system we will now kill all processes in a killing spree to
9540 ensure no processes stay around by accident.
9542 * Three new specifiers may be used in unit files: %u, %h, %s
9543 resolve to the user name, user home directory resp. user
9544 shell. This is useful for running systemd user instances.
9546 * We now automatically rotate journal files if their data
9547 object hash table gets a fill level > 75%. We also size the
9548 hash table based on the configured maximum file size. This
9549 together should lower hash collisions drastically and thus
9550 speed things up a bit.
9552 * journalctl gained the new "--header" switch to introspect
9553 header data of journal files.
9555 * A new setting SystemCallFilters= has been added to services
9556 which may be used to apply blacklists or whitelists to
9557 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
9559 * nspawn gained a new --link-journal= switch (and quicker: -j)
9560 to link the container journal with the host. This makes it
9561 very easy to centralize log viewing on the host for all
9562 guests while still keeping the journal files separated.
9564 * Many bugfixes and optimizations
9566 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
9567 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
9568 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
9573 * Several tools now understand kernel command line arguments,
9574 which are only read when run in an initial RAM disk. They
9575 usually follow closely their normal counterparts, but are
9578 * There's a new tool to analyze the readahead files that are
9579 automatically generated at boot. Use:
9581 /usr/lib/systemd/systemd-readahead analyze /.readahead
9583 * We now provide an early debug shell on tty9 if this enabled. Use:
9585 systemctl enable debug-shell.service
9587 * All plymouth related units have been moved into the Plymouth
9588 package. Please make sure to upgrade your Plymouth version
9591 * systemd-tmpfiles now supports getting passed the basename of
9592 a configuration file only, in which case it will look for it
9593 in all appropriate directories automatically.
9595 * udevadm info now takes a /dev or /sys path as argument, and
9596 does the right thing. Example:
9598 udevadm info /dev/sda
9599 udevadm info /sys/class/block/sda
9601 * systemctl now prints a warning if a unit is stopped but a
9602 unit that might trigger it continues to run. Example: a
9603 service is stopped but the socket that activates it is left
9606 * "systemctl status" will now mention if the log output was
9607 shortened due to rotation since a service has been started.
9609 * The journal API now exposes functions to determine the
9610 "cutoff" times due to rotation.
9612 * journald now understands SIGUSR1 and SIGUSR2 for triggering
9613 immediately flushing of runtime logs to /var if possible,
9614 resp. for triggering immediate rotation of the journal
9617 * It is now considered an error if a service is attempted to
9618 be stopped that is not loaded.
9620 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
9622 * systemd-analyze now supports Python 3
9624 * tmpfiles now supports cleaning up directories via aging
9625 where the first level dirs are always kept around but
9626 directories beneath it automatically aged. This is enabled
9627 by prefixing the age field with '~'.
9629 * Seat objects now expose CanGraphical, CanTTY properties
9630 which is required to deal with very fast bootups where the
9631 display manager might be running before the graphics drivers
9632 completed initialization.
9634 * Seat objects now expose a State property.
9636 * We now include RPM macros for service enabling/disabling
9637 based on the preset logic. We recommend RPM based
9638 distributions to make use of these macros if possible. This
9639 makes it simpler to reuse RPM spec files across
9642 * We now make sure that the collected systemd unit name is
9643 always valid when services log to the journal via
9646 * There's a new man page kernel-command-line(7) detailing all
9647 command line options we understand.
9649 * The fstab generator may now be disabled at boot by passing
9650 fstab=0 on the kernel command line.
9652 * A new kernel command line option modules-load= is now understood
9653 to load a specific kernel module statically, early at boot.
9655 * Unit names specified on the systemctl command line are now
9656 automatically escaped as needed. Also, if file system or
9657 device paths are specified they are automatically turned
9658 into the appropriate mount or device unit names. Example:
9660 systemctl status /home
9661 systemctl status /dev/sda
9663 * The SysVConsole= configuration option has been removed from
9664 system.conf parsing.
9666 * The SysV search path is no longer exported on the D-Bus
9669 * The Names= option has been removed from unit file parsing.
9671 * There's a new man page bootup(7) detailing the boot process.
9673 * Every unit and every generator we ship with systemd now
9674 comes with full documentation. The self-explanatory boot is
9677 * A couple of services gained "systemd-" prefixes in their
9678 name if they wrap systemd code, rather than only external
9679 code. Among them fsck@.service which is now
9680 systemd-fsck@.service.
9682 * The HaveWatchdog property has been removed from the D-Bus
9685 * systemd.confirm_spawn= on the kernel command line should now
9688 * There's a new man page crypttab(5) which details all options
9689 we actually understand.
9691 * systemd-nspawn gained a new --capability= switch to pass
9692 additional capabilities to the container.
9694 * timedated will now read known NTP implementation unit names
9695 from /usr/lib/systemd/ntp-units.d/*.list,
9696 systemd-timedated-ntp.target has been removed.
9698 * journalctl gained a new switch "-b" that lists log data of
9699 the current boot only.
9701 * The notify socket is in the abstract namespace again, in
9702 order to support daemons which chroot() at start-up.
9704 * There is a new Storage= configuration option for journald
9705 which allows configuration of where log data should go. This
9706 also provides a way to disable journal logging entirely, so
9707 that data collected is only forwarded to the console, the
9708 kernel log buffer or another syslog implementation.
9710 * Many bugfixes and optimizations
9712 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
9713 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
9714 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
9715 Shawn Landden, Tom Gundersen
9719 * "systemctl help <unit>" now shows the man page if one is
9722 * Several new man pages have been added.
9724 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
9725 MaxLevelConsole= can now be specified in
9726 journald.conf. These options allow reducing the amount of
9727 data stored on disk or forwarded by the log level.
9729 * TimerSlackNSec= can now be specified in system.conf for
9730 PID1. This allows system-wide power savings.
9732 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
9733 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
9738 * logind is now capable of (optionally) handling power and
9739 sleep keys as well as the lid switch.
9741 * journalctl now understands the syntax "journalctl
9742 /usr/bin/avahi-daemon" to get all log output of a specific
9745 * CapabilityBoundingSet= in system.conf now also influences
9746 the capability bound set of usermode helpers of the kernel.
9748 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
9749 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
9750 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
9751 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
9755 * Note that we skipped 139 releases here in order to set the
9756 new version to something that is greater than both udev's
9757 and systemd's most recent version number.
9759 * udev: all udev sources are merged into the systemd source tree now.
9760 All future udev development will happen in the systemd tree. It
9761 is still fully supported to use the udev daemon and tools without
9762 systemd running, like in initramfs or other init systems. Building
9763 udev though, will require the *build* of the systemd tree, but
9764 udev can be properly *run* without systemd.
9766 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
9767 should be used to create dead device nodes as workarounds for broken
9770 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
9771 no longer supported. udev_monitor_new_from_netlink() needs to be
9772 used to subscribe to events.
9774 * udev: when udevd is started by systemd, processes which are left
9775 behind by forking them off of udev rules, are unconditionally cleaned
9776 up and killed now after the event handling has finished. Services or
9777 daemons must be started as systemd services. Services can be
9778 pulled-in by udev to get started, but they can no longer be directly
9779 forked by udev rules.
9781 * udev: the daemon binary is called systemd-udevd now and installed
9782 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
9783 to adapt to that, create symlink, or rename the binary after building
9786 * libudev no longer provides these symbols:
9787 udev_monitor_from_socket()
9788 udev_queue_get_failed_list_entry()
9789 udev_get_{dev,sys,run}_path()
9790 The versions number was bumped and symbol versioning introduced.
9792 * systemd-loginctl and systemd-journalctl have been renamed
9793 to loginctl and journalctl to match systemctl.
9795 * The config files: /etc/systemd/systemd-logind.conf and
9796 /etc/systemd/systemd-journald.conf have been renamed to
9797 logind.conf and journald.conf. Package updates should rename
9798 the files to the new names on upgrade.
9800 * For almost all files the license is now LGPL2.1+, changed
9801 from the previous GPL2.0+. Exceptions are some minor stuff
9802 of udev (which will be changed to LGPL2.1 eventually, too),
9803 and the MIT licensed sd-daemon.[ch] library that is suitable
9804 to be used as drop-in files.
9806 * systemd and logind now handle system sleep states, in
9807 particular suspending and hibernating.
9809 * logind now implements a sleep/shutdown/idle inhibiting logic
9810 suitable for a variety of uses. Soonishly Lennart will blog
9811 about this in more detail.
9813 * var-run.mount and var-lock.mount are no longer provided
9814 (which previously bind mounted these directories to their new
9815 places). Distributions which have not converted these
9816 directories to symlinks should consider stealing these files
9817 from git history and add them downstream.
9819 * We introduced the Documentation= field for units and added
9820 this to all our shipped units. This is useful to make it
9821 easier to explore the boot and the purpose of the various
9824 * All smaller setup units (such as
9825 systemd-vconsole-setup.service) now detect properly if they
9826 are run in a container and are skipped when
9827 appropriate. This guarantees an entirely noise-free boot in
9828 Linux container environments such as systemd-nspawn.
9830 * A framework for implementing offline system updates is now
9831 integrated, for details see:
9832 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
9834 * A new service type Type=idle is available now which helps us
9835 avoiding ugly interleaving of getty output and boot status
9838 * There's now a system-wide CapabilityBoundingSet= option to
9839 globally reduce the set of capabilities for the
9840 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
9841 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
9842 even CAP_NET_ADMIN system-wide for secure systems.
9844 * There are now system-wide DefaultLimitXXX= options to
9845 globally change the defaults of the various resource limits
9846 for all units started by PID 1.
9848 * Harald Hoyer's systemd test suite has been integrated into
9849 systemd which allows easy testing of systemd builds in qemu
9850 and nspawn. (This is really awesome! Ask us for details!)
9852 * The fstab parser is now implemented as generator, not inside
9855 * systemctl will now warn you if .mount units generated from
9856 /etc/fstab are out of date due to changes in fstab that
9857 have not been read by systemd yet.
9859 * systemd is now suitable for usage in initrds. Dracut has
9860 already been updated to make use of this. With this in place
9861 initrds get a slight bit faster but primarily are much
9862 easier to introspect and debug since "systemctl status" in
9863 the host system can be used to introspect initrd services,
9864 and the journal from the initrd is kept around too.
9866 * systemd-delta has been added, a tool to explore differences
9867 between user/admin configuration and vendor defaults.
9869 * PrivateTmp= now affects both /tmp and /var/tmp.
9871 * Boot time status messages are now much prettier and feature
9872 proper english language. Booting up systemd has never been
9875 * Read-ahead pack files now include the inode number of all
9876 files to pre-cache. When the inode changes the pre-caching
9877 is not attempted. This should be nicer to deal with updated
9878 packages which might result in changes of read-ahead
9881 * We now temporaritly lower the kernel's read_ahead_kb variable
9882 when collecting read-ahead data to ensure the kernel's
9883 built-in read-ahead does not add noise to our measurements
9884 of necessary blocks to pre-cache.
9886 * There's now RequiresMountsFor= to add automatic dependencies
9887 for all mounts necessary for a specific file system path.
9889 * MountAuto= and SwapAuto= have been removed from
9890 system.conf. Mounting file systems at boot has to take place
9893 * nspawn now learned a new switch --uuid= to set the machine
9894 ID on the command line.
9896 * nspawn now learned the -b switch to automatically search
9899 * vt102 is now the default TERM for serial TTYs, upgraded from
9902 * systemd-logind now works on VT-less systems.
9904 * The build tree has been reorganized. The individual
9905 components now have directories of their own.
9907 * A new condition type ConditionPathIsReadWrite= is now available.
9909 * nspawn learned the new -C switch to create cgroups for the
9910 container in other hierarchies.
9912 * We now have support for hardware watchdogs, configurable in
9915 * The scheduled shutdown logic now has a public API.
9917 * We now mount /tmp as tmpfs by default, but this can be
9918 masked and /etc/fstab can override it.
9920 * Since udisks does not make use of /media anymore we are not
9921 mounting a tmpfs on it anymore.
9923 * journalctl gained a new --local switch to only interleave
9924 locally generated journal files.
9926 * We can now load the IMA policy at boot automatically.
9928 * The GTK tools have been split off into a systemd-ui.
9930 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
9931 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
9932 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
9933 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
9934 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
9935 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
9936 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
9937 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
9938 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
9943 * This is mostly a bugfix release
9945 * Support optional initialization of the machine ID from the
9946 KVM or container configured UUID.
9948 * Support immediate reboots with "systemctl reboot -ff"
9950 * Show /etc/os-release data in systemd-analyze output
9952 * Many bugfixes for the journal, including endianness fixes and
9953 ensuring that disk space enforcement works
9955 * sd-login.h is C++ compatible again
9957 * Extend the /etc/os-release format on request of the Debian
9960 * We now refuse non-UTF8 strings used in various configuration
9961 and unit files. This is done to ensure we do not pass invalid
9962 data over D-Bus or expose it elsewhere.
9964 * Register Mimo USB Screens as suitable for automatic seat
9967 * Read SELinux client context from journal clients in a race
9970 * Reorder configuration file lookup order. /etc now always
9971 overrides /run in order to allow the administrator to always
9972 and unconditionally override vendor-supplied or
9973 automatically generated data.
9975 * The various user visible bits of the journal now have man
9976 pages. We still lack man pages for the journal API calls
9979 * We now ship all man pages in HTML format again in the
9982 Contributions from: Dave Reisner, Dirk Eibach, Frederic
9983 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
9984 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
9989 * This is mostly a bugfix release
9991 * systems lacking /etc/os-release are no longer supported.
9993 * Various functionality updates to libsystemd-login.so
9995 * Track class of PAM logins to distinguish greeters from
9998 Contributions from: Kay Sievers, Lennart Poettering, Michael
10003 * This is an important bugfix release for v41.
10005 * Building man pages is now optional which should be useful
10006 for those building systemd from git but unwilling to install
10009 * Watchdog support for supervising services is now usable. In
10010 a future release support for hardware watchdogs
10011 (i.e. /dev/watchdog) will be added building on this.
10013 * Service start rate limiting is now configurable and can be
10014 turned off per service. When a start rate limit is hit a
10015 reboot can automatically be triggered.
10017 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
10019 Contributions from: Benjamin Franzke, Bill Nottingham,
10020 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
10021 Schmidt, Michał Górny, Piotr Drąg
10025 * The systemd binary is installed /usr/lib/systemd/systemd now;
10026 An existing /sbin/init symlink needs to be adapted with the
10029 * The code that loads kernel modules has been ported to invoke
10030 libkmod directly, instead of modprobe. This means we do not
10031 support systems with module-init-tools anymore.
10033 * Watchdog support is now already useful, but still not
10036 * A new kernel command line option systemd.setenv= is
10037 understood to set system wide environment variables
10038 dynamically at boot.
10040 * We now limit the set of capabilities of systemd-journald.
10042 * We now set SIGPIPE to ignore by default, since it only is
10043 useful in shell pipelines, and has little use in general
10044 code. This can be disabled with IgnoreSIPIPE=no in unit
10047 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
10048 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
10053 * This is mostly a bugfix release
10055 * We now expose the reason why a service failed in the
10056 "Result" D-Bus property.
10058 * Rudimentary service watchdog support (will be completed over
10059 the next few releases.)
10061 * When systemd forks off in order execute some service we will
10062 now immediately changes its argv[0] to reflect which process
10063 it will execute. This is useful to minimize the time window
10064 with a generic argv[0], which makes bootcharts more useful
10066 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
10067 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
10068 Mike Kazantsev, Ray Strode
10072 * This is mostly a test release, but incorporates many
10075 * New systemd-cgtop tool to show control groups by their
10078 * Linking against libacl for ACLs is optional again. If
10079 disabled, support tracking device access for active logins
10080 goes becomes unavailable, and so does access to the user
10081 journals by the respective users.
10083 * If a group "adm" exists, journal files are automatically
10084 owned by them, thus allow members of this group full access
10085 to the system journal as well as all user journals.
10087 * The journal now stores the SELinux context of the logging
10088 client for all entries.
10090 * Add C++ inclusion guards to all public headers
10092 * New output mode "cat" in the journal to print only text
10093 messages, without any meta data like date or time.
10095 * Include tiny X server wrapper as a temporary stop-gap to
10096 teach XOrg udev display enumeration. This is used by display
10097 managers such as gdm, and will go away as soon as XOrg
10098 learned native udev hotplugging for display devices.
10100 * Add new systemd-cat tool for executing arbitrary programs
10101 with STDERR/STDOUT connected to the journal. Can also act as
10102 BSD logger replacement, and does so by default.
10104 * Optionally store all locally generated coredumps in the
10105 journal along with meta data.
10107 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
10108 writing short strings to files (for usage for /sys), and for
10109 creating symlinks, character and block device nodes.
10111 * New unit file option ControlGroupPersistent= to make cgroups
10112 persistent, following the mechanisms outlined in
10113 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
10115 * Support multiple local RTCs in a sane way
10117 * No longer monopolize IO when replaying readahead data on
10118 rotating disks, since we might starve non-file-system IO to
10119 death, since fanotify() will not see accesses done by blkid,
10122 * Do not show kernel threads in systemd-cgls anymore, unless
10123 requested with new -k switch.
10125 Contributions from: Dan Horák, Kay Sievers, Lennart
10126 Poettering, Michal Schmidt
10130 * This is mostly a test release, but incorporates many
10133 * The git repository moved to:
10134 git://anongit.freedesktop.org/systemd/systemd
10135 ssh://git.freedesktop.org/git/systemd/systemd
10137 * First release with the journal
10138 http://0pointer.de/blog/projects/the-journal.html
10140 * The journal replaces both systemd-kmsg-syslogd and
10141 systemd-stdout-bridge.
10143 * New sd_pid_get_unit() API call in libsystemd-logind
10145 * Many systemadm clean-ups
10147 * Introduce remote-fs-pre.target which is ordered before all
10148 remote mounts and may be used to start services before all
10151 * Added Mageia support
10153 * Add bash completion for systemd-loginctl
10155 * Actively monitor PID file creation for daemons which exit in
10156 the parent process before having finished writing the PID
10157 file in the daemon process. Daemons which do this need to be
10158 fixed (i.e. PID file creation must have finished before the
10159 parent exits), but we now react a bit more gracefully to them.
10161 * Add colourful boot output, mimicking the well-known output
10162 of existing distributions.
10164 * New option PassCredentials= for socket units, for
10165 compatibility with a recent kernel ABI breakage.
10167 * /etc/rc.local is now hooked in via a generator binary, and
10168 thus will no longer act as synchronization point during
10171 * systemctl list-unit-files now supports --root=.
10173 * systemd-tmpfiles now understands two new commands: z, Z for
10174 relabelling files according to the SELinux database. This is
10175 useful to apply SELinux labels to specific files in /sys,
10176 among other things.
10178 * Output of SysV services is now forwarded to both the console
10179 and the journal by default, not only just the console.
10181 * New man pages for all APIs from libsystemd-login.
10183 * The build tree got reorganized and the build system is a
10184 lot more modular allowing embedded setups to specifically
10185 select the components of systemd they are interested in.
10187 * Support for Linux systems lacking the kernel VT subsystem is
10190 * configure's --with-rootdir= got renamed to
10191 --with-rootprefix= to follow the naming used by udev and
10194 * Unless specified otherwise we will now install to /usr instead
10195 of /usr/local by default.
10197 * Processes with '@' in argv[0][0] are now excluded from the
10198 final shut-down killing spree, following the logic explained
10200 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
10202 * All processes remaining in a service cgroup when we enter
10203 the START or START_PRE states are now killed with
10204 SIGKILL. That means it is no longer possible to spawn
10205 background processes from ExecStart= lines (which was never
10206 supported anyway, and bad style).
10208 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
10209 reloading of units together.
10211 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
10212 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
10213 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
10214 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
10215 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek