1 systemd System and Service Manager
3 CHANGES WITH 243 in spe:
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= settting. 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 system-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 * The special -.service.d dropin directory may be used to add
37 configuration that affects all services. The "-.service" service name
38 is now disallowed (though hopefully no one would use such a service
39 name. -.mount gives us enough grief.)
41 * The RuntimeMaxSec= setting is now supported by scopes, not just
42 .service units. This is particularly useful for PAM sessions which
43 create a scope unit for the user login. systemd.runtime_max_sec=
44 setting may used with the pam_systemd module to limit the duration
45 of the PAM session, for example for time-limited logins.
47 * udev now provides a program (fido_id) that identifies FIDO CTAP1
48 ("U2F")/CTAP2 security tokens based on the usage declared in their
49 report and descriptor and outputs suitable environment variables.
50 This replaces the externally maintained whitelists of all known
51 security tokens that were used previously.
53 * Automatically generated autosuspend udev rules for whitelisted
54 devices have been imported from the Chromium OS project. This should
55 improve power saving with of many more devices.
57 * udev gained a new "CONST{key}=value" setting that allows matching
58 against system-wide constants without forking a helper binary.
59 Currently "arch" and "virt" keys are supported.
61 * udev now opens CDROMs in non-exclusive mode when querying their
62 capabilities. This should fix issues where other programs trying to
63 use the CDROM cannot gain access to it, but carries a risk of
64 interfering with programs writing to the disk, if they did not open
65 the device in exclusive mode as they should.
67 * systemd-networkd does not create a default route for IPv4 link local
68 addressing anymore. The creation of the route was unexpected and was
69 breaking routing in various cases, but people who rely on it being
70 created implicitly will need to adjust. Such a route may be requested
71 with DefaultRouteOnDevice=yes.
73 Similarly, systemd-networkd will not assign a link-local IPv6 address
74 when IPv6 link-local routing is not enabled.
76 * Receive and transmit buffers may now be configured on links with
77 the new RxBufferSize= and TxBufferSize= settings.
79 * systemd-networkd may now advertise additional IPv6 routes. A new
80 [IPv6RoutePrefix] section with Route= and LifetimeSec= options is
83 * systemd-networkd may now configure "next hop" routes using the
84 [NextHop] section and Gateway= and Id= settings.
86 * systemd-networkd will now retain DHCP config on restarts by default
87 (but this may be overridden using the KeepConfiguration= setting).
88 The default for SendRelease= has been changed to true.
90 * The DHCPv4 client now uses the OPTION_INFORMATION_REFRESH_TIME option
91 received from the server.
93 The client will use the received SIP server list if UseSIP=yes is
96 The client may be configured to request specific options from the
97 server using a new RequestOptions= setting.
99 A new IPServiceType= setting has been added to configure the "IP
100 service type" value used by the client.
102 * The DHCPv6 client learnt a new PrefixDelegationHint= option to
103 request prefix hints in the DHCPv6 solicitation.
105 * systemd-networkd and networkctl may now renew DHCP leases on demand.
106 networkctl has a new 'networkctl renew' verb.
108 * systemd-networkd may now reconfigure links on demand. networkctl
109 gained two new verbs: "reload" will reload the configuration, and
110 "reconfigure DEVICE…" will reconfigure one or more devices.
112 * .network files may now match on SSID and BSSID of a wireless network,
113 i.e. the access point name and hardware address using the new SSID=
114 and BSSID= options. networkctl will display the current SSID and
115 BSSID for wireless links.
117 .network files may also match on the wireless network type using the
118 new WLANInterfaceType= option.
120 * systemd-networkd now includes default configuration that enables
121 link-local addressing when connected to an ad-hoc wireless network.
123 * The DHCPv4 server may now be configured to emit SIP server list using
124 the new EmitSIP= and SIP= settings.
126 * systemd-networkd may configure the Traffic Control queueing
127 disciplines in the kernel using the new
128 [TrafficControlQueueingDiscipline] section and Parent=,
129 NetworkEmulatorDelaySec=, NetworkEmulatorDelayJitterSec=,
130 NetworkEmulatorPacketLimit=, NetworkEmulatorLossRate= settings.
132 * systemd-tmpfiles gained a new w+ setting to append to files.
134 * systemd-analyze dump will now report when the memory configuration in
135 the kernel does not match what systemd has configured (usually,
136 because some external program has modified the kernel configuration
139 * journalctl --update-catalog now produces deterministic output (making
140 reproducible image builds easier).
142 * A new devicetree-overlay setting is now documented in the Boot Loader
145 * The default value of the WatchdogSec= setting used in systemd
146 services (the ones bundled with the project itself) may be set at
147 configuration time using the -Dservice-watchdog= setting. If set to
148 empty, the watchdogs will be disabled.
150 * libcryptsetup >= 2.0.1 is now required.
154 * This release enables unprivileged programs (i.e. requiring neither
155 setuid nor file capabilities) to send ICMP Echo (i.e. ping) requests
156 by turning on the "net.ipv4.ping_group_range" sysctl of the Linux
157 kernel for the whole UNIX group range, i.e. all processes. This
158 change should be reasonably safe, as the kernel support for it was
159 specifically implemented to allow safe access to ICMP Echo for
160 processes lacking any privileges. If this is not desirable, it can be
161 disabled again by setting the parameter to "1 0".
163 * Previously, filters defined with SystemCallFilter= would have the
164 effect that any calling of an offending system call would terminate
165 the calling thread. This behaviour never made much sense, since
166 killing individual threads of unsuspecting processes is likely to
167 create more problems than it solves. With this release the default
168 action changed from killing the thread to killing the whole
169 process. For this to work correctly both a kernel version (>= 4.14)
170 and a libseccomp version (>= 2.4.0) supporting this new seccomp
171 action is required. If an older kernel or libseccomp is used the old
172 behaviour continues to be used. This change does not affect any
173 services that have no system call filters defined, or that use
174 SystemCallErrorNumber= (and thus see EPERM or another error instead
175 of being killed when calling an offending system call). Note that
176 systemd documentation always claimed that the whole process is
177 killed. With this change behaviour is thus adjusted to match the
180 * On 64 bit systems, the "kernel.pid_max" sysctl is now bumped to
181 4194304 by default, i.e. the full 22bit range the kernel allows, up
182 from the old 16bit range. This should improve security and
183 robustness, as PID collisions are made less likely (though certainly
184 still possible). There are rumours this might create compatibility
185 problems, though at this moment no practical ones are known to
186 us. Downstream distributions are hence advised to undo this change in
187 their builds if they are concerned about maximum compatibility, but
188 for everybody else we recommend leaving the value bumped. Besides
189 improving security and robustness this should also simplify things as
190 the maximum number of allowed concurrent tasks was previously bounded
191 by both "kernel.pid_max" and "kernel.threads-max" and now effectively
192 only a single knob is left ("kernel.threads-max"). There have been
193 concerns that usability is affected by this change because larger PID
194 numbers are harder to type, but we believe the change from 5 digits
195 to 7 digits doesn't hamper usability.
197 * MemoryLow= and MemoryMin= gained hierarchy-aware counterparts,
198 DefaultMemoryLow= and DefaultMemoryMin=, which can be used to
199 hierarchically set default memory protection values for a particular
200 subtree of the unit hierarchy.
202 * Memory protection directives can now take a value of zero, allowing
203 explicit opting out of a default value propagated by an ancestor.
205 * systemd now defaults to the "unified" cgroup hierarchy setup during
206 build-time, i.e. -Ddefault-hierarchy=unified is now the build-time
207 default. Previously, -Ddefault-hierarchy=hybrid was the default. This
208 change reflects the fact that cgroupsv2 support has matured
209 substantially in both systemd and in the kernel, and is clearly the
210 way forward. Downstream production distributions might want to
211 continue to use -Ddefault-hierarchy=hybrid (or even =legacy) for
212 their builds as unfortunately the popular container managers have not
213 caught up with the kernel API changes.
215 * Man pages are not built by default anymore (html pages were already
216 disabled by default), to make development builds quicker. When
217 building systemd for a full installation with documentation, meson
218 should be called with -Dman=true and/or -Dhtml=true as appropriate.
219 The default was changed based on the assumption that quick one-off or
220 repeated development builds are much more common than full optimized
221 builds for installation, and people need to pass various other
222 options to when doing "proper" builds anyway, so the gain from making
223 development builds quicker is bigger than the one time disruption for
226 Two scripts are created in the *build* directory to generate and
227 preview man and html pages on demand, e.g.:
229 build/man/man systemctl
230 build/man/html systemd.index
232 * libidn2 is used by default if both libidn2 and libidn are installed.
233 Please use -Dlibidn=true if libidn is preferred.
235 * The D-Bus "wire format" of the CPUAffinity= attribute is changed on
236 big-endian machines. Before, bytes were written and read in native
237 machine order as exposed by the native libc __cpu_mask interface.
238 Now, little-endian order is always used (CPUs 0–7 are described by
239 bits 0–7 in byte 0, CPUs 8–15 are described by byte 1, and so on).
240 This change fixes D-Bus calls that cross endianness boundary.
242 The presentation format used for CPUAffinity= by "systemctl show" and
243 "systemd-analyze dump" is changed to present CPU indices instead of
244 the raw __cpu_mask bitmask. For example, CPUAffinity=0-1 would be
245 shown as CPUAffinity=03000000000000000000000000000… (on
246 little-endian) or CPUAffinity=00000000000000300000000000000… (on
247 64-bit big-endian), and is now shown as CPUAffinity=0-1, matching the
248 input format. The maximum integer that will be printed in the new
249 format is 8191 (four digits), while the old format always used a very
250 long number (with the length varying by architecture), so they can be
251 unambiguously distinguished.
253 * /usr/sbin/halt.local is no longer supported. Implementation in
254 distributions was inconsistent and it seems this functionality was
257 To replace this functionality, users should:
258 - either define a new unit and make it a dependency of final.target
259 (systemctl add-wants final.target my-halt-local.service)
260 - or move the shutdown script to /usr/lib/systemd/system-shutdown/
261 and ensure that it accepts "halt", "poweroff", "reboot", and
262 "kexec" as an argument, see the description in systemd-shutdown(8).
264 * When a [Match] section in .link or .network file is empty (contains
265 no match patterns), a warning will be emitted. Please add any "match
266 all" pattern instead, e.g. OriginalName=* or Name=* in case all
267 interfaces should really be matched.
269 * A new setting NUMAPolicy= may be used to set process memory
270 allocation policy. This setting can be specified in
271 /etc/systemd/system.conf and hence will set the default policy for
272 PID1. The default policy can be overridden on a per-service
273 basis. The related setting NUMAMask= is used to specify NUMA node
274 mask that should be associated with the selected policy.
276 * PID 1 will now listen to Out-Of-Memory (OOM) events the kernel
277 generates when processes it manages are reaching their memory limits,
278 and will place their units in a special state, and optionally kill or
281 * The service manager will now expose bus properties for the IO
282 resources used by units. This information is also shown in "systemctl
283 status" now (for services that have IOAccounting=yes set). Moreover,
284 the IO accounting data is included in the resource log message
285 generated whenever a unit stops.
287 * Units may now configure an explicit time-out to wait for when killed
288 with SIGABRT, for example when a service watchdog is hit. Previously,
289 the regular TimeoutStopSec= time-out was applied in this case too —
290 now a separate time-out may be set using TimeoutAbortSec=.
292 * Services may now send a special WATCHDOG=trigger message with
293 sd_notify() to trigger an immediate "watchdog missed" event, and thus
294 trigger service termination. This is useful both for testing watchdog
295 handling, but also for defining error paths in services, that shall
296 be handled the same way as watchdog events.
298 * There are two new per-unit settings IPIngressFilterPath= and
299 IPEgressFilterPath= which allow configuration of a BPF program
300 (usually by specifying a path to a program uploaded to /sys/fs/bpf/)
301 to apply to the IP packet ingress/egress path of all processes of a
302 unit. This is useful to allow running systemd services with BPF
303 programs set up externally.
305 * systemctl gained a new "clean" verb for removing the state, cache,
306 runtime or logs directories of a service while it is terminated. The
307 new verb may also be used to remove the state maintained on disk for
308 timer units that have Persistent= configured.
310 * During the last phase of shutdown systemd will now automatically
311 increase the log level configured in the "kernel.printk" sysctl so
312 that any relevant loggable events happening during late shutdown are
313 made visible. Previously, loggable events happening so late during
314 shutdown were generally lost if the "kernel.printk" sysctl was set to
315 high thresholds, as regular logging daemons are terminated at that
316 time and thus nothing is written to disk.
318 * If processes terminated during the last phase of shutdown do not exit
319 quickly systemd will now show their names after a short time, to make
320 debugging easier. After a longer time-out they are forcibly killed,
323 * journalctl (and the other tools that display logs) will now highlight
324 warnings in yellow (previously, both LOG_NOTICE and LOG_WARNING where
325 shown in bright bold, now only LOG_NOTICE is). Moreover, audit logs
326 are now shown in blue color, to separate them visually from regular
327 logs. References to configuration files are now turned into clickable
328 links on terminals that support that.
330 * systemd-journald will now stop logging to /var/log/journal during
331 shutdown when /var/ is on a separate mount, so that it can be
332 unmounted safely during shutdown.
334 * systemd-resolved gained support for a new 'strict' DNS-over-TLS mode.
336 * systemd-resolved "Cache=" configuration option in resolved.conf has
337 been extended to also accept the 'no-negative' value. Previously,
338 only a boolean option was allowed (yes/no), having yes as the
339 default. If this option is set to 'no-negative', negative answers are
340 not cached while the old cache heuristics are used positive answers.
341 The default remains unchanged.
343 * The predictable naming scheme for network devices now supports
344 generating predictable names for "netdevsim" devices.
346 Moreover, the "en" prefix was dropped from the ID_NET_NAME_ONBOARD
349 Those two changes form a new net.naming-policy-scheme= entry.
350 Distributions which want to preserve naming stability may want to set
351 the -Ddefault-net-naming-scheme= configuration option.
353 * systemd-networkd now supports MACsec, nlmon, IPVTAP and Xfrm
356 * systemd-networkd's bridge FDB support now allows configuration of a
357 destination address for each entry (Destination=), as well as the
358 VXLAN VNI (VNI=), as well as an option to declare what an entry is
359 associated with (AssociatedWith=).
361 * systemd-networkd's DHCPv4 support now understands a new MaxAttempts=
362 option for configuring the maximum number of DHCP lease requests. It
363 also learnt a new BlackList= option for blacklisting DHCP servers (a
364 similar setting has also been added to the IPv6 RA client), as well
365 as a SendRelease= option for configuring whether to send a DHCP
366 RELEASE message when terminating.
368 * systemd-networkd's DHCPv4 and DHCPv6 stacks can now be configured
369 separately in the [DHCPv4] and [DHCPv6] sections.
371 * systemd-networkd's DHCP support will now optionally create an
372 implicit host route to the DNS server specified in the DHCP lease, in
373 addition to the routes listed explicitly in the lease. This should
374 ensure that in multi-homed systems DNS traffic leaves the systems on
375 the interface that acquired the DNS server information even if other
376 routes such as default routes exist. This behaviour may be turned on
377 with the new RoutesToDNS= option.
379 * systemd-networkd's VXLAN support gained a new option
380 GenericProtocolExtension= for enabling VXLAN Generic Protocol
381 Extension support, as well as IPDoNotFragment= for setting the IP
382 "Don't fragment" bit on outgoing packets. A similar option has been
383 added to the GENEVE support.
385 * In systemd-networkd's [Route] section you may now configure
386 FastOpenNoCookie= for configuring per-route TCP fast-open support, as
387 well as TTLPropagate= for configuring Label Switched Path (LSP) TTL
388 propagation. The Type= setting now supports local, broadcast,
389 anycast, multicast, any, xresolve routes, too.
391 * systemd-networkd's [Network] section learnt a new option
392 DefaultRouteOnDevice= for automatically configuring a default route
393 onto the network device.
395 * systemd-networkd's bridging support gained two new options ProxyARP=
396 and ProxyARPWifi= for configuring proxy ARP behaviour as well as
397 MulticastRouter= for configuring multicast routing behaviour. A new
398 option MulticastIGMPVersion= may be used to change bridge's multicast
399 Internet Group Management Protocol (IGMP) version.
401 * systemd-networkd's FooOverUDP support gained the ability to configure
402 local and peer IP addresses via Local= and Peer=. A new option
403 PeerPort= may be used to configure the peer's IP port.
405 * systemd-networkd's TUN support gained a new setting VnetHeader= for
406 tweaking Generic Segment Offload support.
408 * networkctl gained a new "delete" command for removing virtual network
409 devices, as well as a new "--stats" switch for showing device
412 * networkd.conf gained a new setting SpeedMeter= and
413 SpeedMeterIntervalSec=, to measure bitrate of network interfaces. The
414 measured speed may be shown by 'networkctl status'.
416 * "networkctl status" now displays MTU and queue lengths, and more
417 detailed information about VXLAN and bridge devices.
419 * systemd-networkd's .network and .link files gained a new Property=
420 setting in the [Match] section, to match against devices with
421 specific udev properties.
423 * systemd-networkd's tunnel support gained a new option
424 AssignToLoopback= for selecting whether to use the loopback device
425 "lo" as underlying device.
427 * systemd-networkd's MACAddress= setting in the [Neighbor] section has
428 been renamed to LinkLayerAddress=, and it now allows configuration of
431 * systemd-networkd's handling of the kernel's disable_ipv6 sysctl is
432 simplified: systemd-networkd will disable the sysctl (enable IPv6) if
433 IPv6 configuration (static or DHCPv6) was found for a given
434 interface. It will not touch the sysctl otherwise.
436 * The order of entries is $PATH used by the user manager instance was
437 changed to put bin/ entries before the corresponding sbin/ entries.
438 It is recommended to not rely on this order, and only ever have one
439 binary with a given name in the system paths under /usr.
441 * A new tool systemd-network-generator has been added that may generate
442 .network, .netdev and .link files from IP configuration specified on
443 the kernel command line in the format used by Dracut.
445 * The CriticalConnection= setting in .network files is now deprecated,
446 and replaced by a new KeepConfiguration= setting which allows more
447 detailed configuration of the IP configuration to keep in place.
449 * systemd-analyze gained a few new verbs:
451 - "systemd-analyze timestamp" parses and converts timestamps. This is
452 similar to the existing "systemd-analyze calendar" command which
453 does the same for recurring calendar events.
455 - "systemd-analyze timespan" parses and converts timespans (i.e.
456 durations as opposed to points in time).
458 - "systemd-analyze condition" will parse and test ConditionXYZ=
461 - "systemd-analyze exit-status" will parse and convert exit status
462 codes to their names and back.
464 - "systemd-analyze unit-files" will print a list of all unit
465 file paths and unit aliases.
467 * SuccessExitStatus=, RestartPreventExitStatus=, and
468 RestartForceExitStatus= now accept exit status names (e.g. "DATAERR"
469 is equivalent to "65"). Those exit status name mappings may be
470 displayed with the sytemd-analyze exit-status verb describe above.
472 * systemd-logind now exposes a per-session SetBrightness() bus call,
473 which may be used to securely change the brightness of a kernel
474 brightness device, if it belongs to the session's seat. By using this
475 call unprivileged clients can make changes to "backlight" and "leds"
476 devices securely with strict requirements on session membership.
477 Desktop environments may use this to generically make brightness
478 changes to such devices without shipping private SUID binaries or
479 udev rules for that purpose.
481 * "udevadm info" gained a --wait-for-initialization switch to wait for
482 a device to be initialized.
484 * systemd-hibernate-resume-generator will now look for resumeflags= on
485 the kernel command line, which is similar to rootflags= and may be
486 used to configure device timeout for the hibernation device.
488 * sd-event learnt a new API call sd_event_source_disable_unref() for
489 disabling and unref'ing an event source in a single function. A
490 related call sd_event_source_disable_unrefp() has been added for use
491 with gcc's cleanup extension.
493 * The sd-id128.h public API gained a new definition
494 SD_ID128_UUID_FORMAT_STR for formatting a 128bit ID in UUID format
497 * "busctl introspect" gained a new switch --xml-interface for dumping
498 XML introspection data unmodified.
500 * PID 1 may now show the unit name instead of the unit description
501 string in its status output during boot. This may be configured in
502 the StatusUnitFormat= setting in /etc/systemd/system.conf or the
503 kernel command line option systemd.status_unit_format=.
505 * PID 1 now understands a new option KExecWatchdogSec= in
506 /etc/systemd/system.conf to set a watchdog timeout for kexec reboots.
507 Previously watchdog functionality was only available for regular
508 reboots. The new setting defaults to off, because we don't know in
509 the general case if the watchdog will be reset after kexec (some
510 drivers do reset it, but not all), and the new userspace might not be
511 configured to handle the watchdog.
513 Moreover, the old ShutdownWatchdogSec= setting has been renamed to
514 RebootWatchdogSec= to more clearly communicate what it is about. The
515 old name is still accepted for compatibility.
517 * The systemd.debug_shell kernel command line option now optionally
518 takes a tty name to spawn the debug shell on, which allows a
519 different tty to be selected than the built-in default.
521 * Service units gained a new ExecCondition= setting which will run
522 before ExecStartPre= and either continue execution of the unit (for
523 clean exit codes), stop execution without marking the unit failed
524 (for exit codes 1 through 254), or stop execution and fail the unit
525 (for exit code 255 or abnormal termination).
527 * A new service systemd-pstore.service has been added that pulls data
528 from /sys/fs/pstore/ and saves it to /var/lib/pstore for later
531 * timedatectl gained new verbs for configuring per-interface NTP
532 service configuration for systemd-timesyncd.
534 * "localectl list-locales" won't list non-UTF-8 locales anymore. It's
535 2019. (You can set non-UTF-8 locales though, if you know their name.)
537 * If variable assignments in sysctl.d/ files are prefixed with "-" any
538 failures to apply them are now ignored.
540 * systemd-random-seed.service now optionally credits entropy when
541 applying the seed to the system. Set $SYSTEMD_RANDOM_SEED_CREDIT to
542 true for the service to enable this behaviour, but please consult the
543 documentation first, since this comes with a couple of caveats.
545 * systemd-random-seed.service is now a synchronization point for full
546 initialization of the kernel's entropy pool. Services that require
547 /dev/urandom to be correctly initialized should be ordered after this
550 * The systemd-boot boot loader has been updated to optionally maintain
551 a random seed file in the EFI System Partition (ESP). During the boot
552 phase, this random seed is read and updated with a new seed
553 cryptographically derived from it. Another derived seed is passed to
554 the OS. The latter seed is then credited to the kernel's entropy pool
555 very early during userspace initialization (from PID 1). This allows
556 systems to boot up with a fully initialized kernel entropy pool from
557 earliest boot on, and thus entirely removes all entropy pool
558 initialization delays from systems using systemd-boot. Special care
559 is taken to ensure different seeds are derived on system images
560 replicated to multiple systems. "bootctl status" will show whether
561 a seed was received from the boot loader.
563 * bootctl gained two new verbs:
565 - "bootctl random-seed" will generate the file in ESP and an EFI
566 variable to allow a random seed to be passed to the OS as described
569 - "bootctl is-installed" checks whether systemd-boot is currently
572 * bootctl will warn if it detects that boot entries are misconfigured
573 (for example if the kernel image was removed without purging the
576 * A new document has been added describing systemd's use and support
577 for the kernel's entropy pool subsystem:
579 https://systemd.io/RANDOM_SEEDS
581 * When the system is hibernated the swap device to write the
582 hibernation image to is now automatically picked from all available
583 swap devices, preferring the swap device with the highest configured
584 priority over all others, and picking the device with the most free
585 space if there are multiple devices with the highest priority.
587 * /etc/crypttab support has learnt a new keyfile-timeout= per-device
588 option that permits selecting the timout how long to wait for a
589 device with an encryption key before asking for the password.
591 * IOWeight= has learnt to properly set the IO weight when using the
592 BFQ scheduler officially found in kernels 5.0+.
594 * A new mailing list has been created for reporting of security issues:
595 systemd-security@redhat.com. For mode details, see
596 https://systemd.io/CONTRIBUTING#security-vulnerability-reports.
598 Contributions from: Aaron Barany, Adrian Bunk, Alan Jenkins, Albrecht
599 Lohofener, Andrej Valek, Anita Zhang, Arian van Putten, Balint Reczey,
600 Bastien Nocera, Ben Boeckel, Benjamin Robin, camoz, Chen Qi, Chris
601 Chiu, Chris Down, Christian Kellner, Clinton Roy, Connor Reeder, Daniel
602 Black, Daniele Medri, Dan Streetman, Dave Reisner, Dave Ross, David
603 Art, David Tardon, Debarshi Ray, Dimitri John Ledkov, Dominick Grift,
604 Donald Buczek, Douglas Christman, Eric DeVolder, EtherGraf, Evgeny
605 Vereshchagin, Feldwor, Felix Riemann, Florian Dollinger, Francesco
606 Pennica, Franck Bui, Frantisek Sumsal, Franz Pletz, frederik, Hans
607 de Goede, Iago López Galeiras, Insun Pyo, Ivan Shapovalov, Iwan Timmer,
608 Jack, Jakob Unterwurzacher, Jan Chren, Jan Klötzke, Jan Losinski, Jan
609 Pokorný, Jan Synacek, Jan-Michael Brummer, Jeka Pats, Jeremy Soller,
610 Jérémy Rosen, Jiri Pirko, Joe Lin, Joerg Behrmann, Joe Richey, Jóhann
611 B. Guðmundsson, Johannes Christ, Johannes Schmitz, Jonathan Rouleau,
612 Jorge Niedbalski, Kai Krakow, Kai Lüke, Karel Zak, Kashyap Chamarthy,
613 Krayushkin Konstantin, Lennart Poettering, Lubomir Rintel, Luca
614 Boccassi, Luís Ferreira, Marc-André Lureau, Markus Felten, Martin Pitt,
615 Matthew Leeds, Mattias Jernberg, Michael Biebl, Michael Olbrich,
616 Michael Prokop, Michael Stapelberg, Michael Zhivich, Michal Koutný,
617 Michal Sekletar, Mike Gilbert, Milan Broz, Miroslav Lichvar, mpe85,
618 Mr-Foo, Network Silence, Oliver Harley, pan93412, Paul Menzel, pEJipE,
619 Peter A. Bigot, Philip Withnall, Piotr Drąg, Rafael Fontenelle, Roberto
620 Santalla, Ronan Pigott, root, RussianNeuroMancer, Sebastian Jennen,
621 shinygold, Shreyas Behera, Simon Schricker, Susant Sahani, Thadeu Lima
622 de Souza Cascardo, Theo Ouzhinski, Thiebaud Weksteen, Thomas Haller,
623 Thomas Weißschuh, Tomas Mraz, Tommi Rantala, Topi Miettinen, VD-Lycos,
624 ven, Wieland Hoffmann, William A. Kennington III, William Wold, Xi
625 Ruoyao, Yuri Chornoivan, Yu Watanabe, Zach Smith, Zbigniew
626 Jędrzejewski-Szmek, Zhang Xianwei
628 – Camerino, 2019-09-03
632 * In .link files, MACAddressPolicy=persistent (the default) is changed
633 to cover more devices. For devices like bridges, tun, tap, bond, and
634 similar interfaces that do not have other identifying information,
635 the interface name is used as the basis for persistent seed for MAC
636 and IPv4LL addresses. The way that devices that were handled
637 previously is not changed, and this change is about covering more
638 devices then previously by the "persistent" policy.
640 MACAddressPolicy=random may be used to force randomized MACs and
641 IPv4LL addresses for a device if desired.
643 Hint: the log output from udev (at debug level) was enhanced to
644 clarify what policy is followed and which attributes are used.
645 `SYSTEMD_LOG_LEVEL=debug udevadm test-builtin net_setup_link /sys/class/net/<name>`
646 may be used to view this.
648 Hint: if a bridge interface is created without any slaves, and gains
649 a slave later, then now the bridge does not inherit slave's MAC.
650 To inherit slave's MAC, for example, create the following file:
652 # /etc/systemd/network/98-bridge-inherit-mac.link
657 MACAddressPolicy=none
660 * The .device units generated by systemd-fstab-generator and other
661 generators do not automatically pull in the corresponding .mount unit
662 as a Wants= dependency. This means that simply plugging in the device
663 will not cause the mount unit to be started automatically. But please
664 note that the mount unit may be started for other reasons, in
665 particular if it is part of local-fs.target, and any unit which
666 (transitively) depends on local-fs.target is started.
668 * networkctl list/status/lldp now accept globbing wildcards for network
669 interface names to match against all existing interfaces.
671 * The $PIDFILE environment variable is set to point the absolute path
672 configured with PIDFile= for processes of that service.
674 * The fallback DNS server list was augmented with Cloudflare public DNS
675 servers. Use `-Ddns-servers=` to set a different fallback.
677 * A new special target usb-gadget.target will be started automatically
678 when a USB Device Controller is detected (which means that the system
679 is a USB peripheral).
681 * A new unit setting CPUQuotaPeriodSec= assigns the time period
682 relatively to which the CPU time quota specified by CPUQuota= is
685 * A new unit setting ProtectHostname= may be used to prevent services
686 from modifying hostname information (even if they otherwise would
687 have privileges to do so).
689 * A new unit setting NetworkNamespacePath= may be used to specify a
690 namespace for service or socket units through a path referring to a
691 Linux network namespace pseudo-file.
693 * The PrivateNetwork= setting and JoinsNamespaceOf= dependencies now
694 have an effect on .socket units: when used the listening socket is
695 created within the configured network namespace instead of the host
698 * ExecStart= command lines in unit files may now be prefixed with ':'
699 in which case environment variable substitution is
700 disabled. (Supported for the other ExecXYZ= settings, too.)
702 * .timer units gained two new boolean settings OnClockChange= and
703 OnTimezoneChange= which may be used to also trigger a unit when the
704 system clock is changed or the local timezone is
705 modified. systemd-run has been updated to make these options easily
706 accessible from the command line for transient timers.
708 * Two new conditions for units have been added: ConditionMemory= may be
709 used to conditionalize a unit based on installed system
710 RAM. ConditionCPUs= may be used to conditionalize a unit based on
713 * The @default system call filter group understood by SystemCallFilter=
714 has been updated to include the new rseq() system call introduced in
717 * A new time-set.target has been added that indicates that the system
718 time has been set from a local source (possibly imprecise). The
719 existing time-sync.target is stronger and indicates that the time has
720 been synchronized with a precise external source. Services where
721 approximate time is sufficient should use the new target.
723 * "systemctl start" (and related commands) learnt a new
724 --show-transaction option. If specified brief information about all
725 jobs queued because of the requested operation is shown.
727 * systemd-networkd recognizes a new operation state 'enslaved', used
728 (instead of 'degraded' or 'carrier') for interfaces which form a
729 bridge, bond, or similar, and an new 'degraded-carrier' operational
730 state used for the bond or bridge master interface when one of the
731 enslaved devices is not operational.
733 * .network files learnt the new IgnoreCarrierLoss= option for leaving
734 networks configured even if the carrier is lost.
736 * The RequiredForOnline= setting in .network files may now specify a
737 minimum operational state required for the interface to be considered
738 "online" by systemd-networkd-wait-online. Related to this
739 systemd-networkd-wait-online gained a new option --operational-state=
740 to configure the same, and its --interface= option was updated to
741 optionally also take an operational state specific for an interface.
743 * systemd-networkd-wait-online gained a new setting --any for waiting
744 for only one of the requested interfaces instead of all of them.
746 * systemd-networkd now implements L2TP tunnels.
748 * Two new .network settings UseAutonomousPrefix= and UseOnLinkPrefix=
749 may be used to cause autonomous and onlink prefixes received in IPv6
750 Router Advertisements to be ignored.
752 * New MulticastFlood=, NeighborSuppression=, and Learning= .network
753 file settings may be used to tweak bridge behaviour.
755 * The new TripleSampling= option in .network files may be used to
756 configure CAN triple sampling.
758 * A new .netdev settings PrivateKeyFile= and PresharedKeyFile= may be
759 used to point to private or preshared key for a WireGuard interface.
761 * /etc/crypttab now supports the same-cpu-crypt and
762 submit-from-crypt-cpus options to tweak encryption work scheduling
765 * systemd-tmpfiles will now take a BSD file lock before operating on a
766 contents of directory. This may be used to temporarily exclude
767 directories from aging by taking the same lock (useful for example
768 when extracting a tarball into /tmp or /var/tmp as a privileged user,
769 which might create files with really old timestamps, which
770 nevertheless should not be deleted). For further details, see:
772 https://systemd.io/TEMPORARY_DIRECTORIES
774 * systemd-tmpfiles' h line type gained support for the
775 FS_PROJINHERIT_FL ('P') file attribute (introduced in kernel 4.5),
776 controlling project quota inheritance.
778 * sd-boot and bootctl now implement support for an Extended Boot Loader
779 (XBOOTLDR) partition, that is intended to be mounted to /boot, in
780 addition to the ESP partition mounted to /efi or /boot/efi.
781 Configuration file fragments, kernels, initrds and other EFI images
782 to boot will be loaded from both the ESP and XBOOTLDR partitions.
783 The XBOOTLDR partition was previously described by the Boot Loader
784 Specification, but implementation was missing in sd-boot. Support for
785 this concept allows using the sd-boot boot loader in more
786 conservative scenarios where the boot loader itself is placed in the
787 ESP but the kernels to boot (and their metadata) in a separate
790 * A system may now be booted with systemd.volatile=overlay on the
791 kernel command line, which causes the root file system to be set up
792 an overlayfs mount combining the root-only root directory with a
793 writable tmpfs. In this setup, the underlying root device is not
794 modified, and any changes are lost at reboot.
796 * Similar, systemd-nspawn can now boot containers with a volatile
797 overlayfs root with the new --volatile=overlay switch.
799 * systemd-nspawn can now consume OCI runtime bundles using a new
800 --oci-bundle= option. This implementation is fully usable, with most
801 features in the specification implemented, but since this a lot of
802 new code and functionality, this feature should most likely not
803 be used in production yet.
805 * systemd-nspawn now supports various options described by the OCI
806 runtime specification on the command-line and in .nspawn files:
807 --inaccessible=/Inaccessible= may be used to mask parts of the file
808 system tree, --console=/--pipe may be used to configure how standard
809 input, output, and error are set up.
811 * busctl learned the `emit` verb to generate D-Bus signals.
813 * systemd-analyze cat-config may be used to gather and display
814 configuration spread over multiple files, for example system and user
815 presets, tmpfiles.d, sysusers.d, udev rules, etc.
817 * systemd-analyze calendar now takes an optional new parameter
818 --iterations= which may be used to show a maximum number of iterations
819 the specified expression will elapse next.
821 * The sd-bus C API gained support for naming method parameters in the
824 * systemd-logind gained D-Bus APIs to specify the "reboot parameter"
825 the reboot() system call expects.
827 * journalctl learnt a new --cursor-file= option that points to a file
828 from which a cursor should be loaded in the beginning and to which
829 the updated cursor should be stored at the end.
831 * ACRN hypervisor and Windows Subsystem for Linux (WSL) are now
832 detected by systemd-detect-virt (and may also be used in
833 ConditionVirtualization=).
835 * The behaviour of systemd-logind may now be modified with environment
836 variables $SYSTEMD_REBOOT_TO_FIRMWARE_SETUP,
837 $SYSTEMD_REBOOT_TO_BOOT_LOADER_MENU, and
838 $SYSTEMD_REBOOT_TO_BOOT_LOADER_ENTRY. They cause logind to either
839 skip the relevant operation completely (when set to false), or to
840 create a flag file in /run/systemd (when set to true), instead of
841 actually commencing the real operation when requested. The presence
842 of /run/systemd/reboot-to-firmware-setup,
843 /run/systemd/reboot-to-boot-loader-menu, and
844 /run/systemd/reboot-to-boot-loader-entry, may be used by alternative
845 boot loader implementations to replace some steps logind performs
846 during reboot with their own operations.
848 * systemctl can be used to request a reboot into the boot loader menu
849 or a specific boot loader entry with the new --boot-load-menu= and
850 --boot-loader-entry= options to a reboot command. (This requires a
851 boot loader that supports this, for example sd-boot.)
853 * kernel-install will no longer unconditionally create the output
854 directory (e.g. /efi/<machine-id>/<kernel-version>) for boot loader
855 snippets, but will do only if the machine-specific parent directory
856 (i.e. /efi/<machine-id>/) already exists. bootctl has been modified
857 to create this parent directory during sd-boot installation.
859 This makes it easier to use kernel-install with plugins which support
860 a different layout of the bootloader partitions (for example grub2).
862 * During package installation (with `ninja install`), we would create
863 symlinks for getty@tty1.service, systemd-networkd.service,
864 systemd-networkd.socket, systemd-resolved.service,
865 remote-cryptsetup.target, remote-fs.target,
866 systemd-networkd-wait-online.service, and systemd-timesyncd.service
867 in /etc, as if `systemctl enable` was called for those units, to make
868 the system usable immediately after installation. Now this is not
869 done anymore, and instead calling `systemctl preset-all` is
870 recommended after the first installation of systemd.
872 * A new boolean sandboxing option RestrictSUIDSGID= has been added that
873 is built on seccomp. When turned on creation of SUID/SGID files is
876 * The NoNewPrivileges= and the new RestrictSUIDSGID= options are now
877 implied if DynamicUser= is turned on for a service. This hardens
878 these services, so that they neither can benefit from nor create
879 SUID/SGID executables. This is a minor compatibility breakage, given
880 that when DynamicUser= was first introduced SUID/SGID behaviour was
881 unaffected. However, the security benefit of these two options is
882 substantial, and the setting is still relatively new, hence we opted
883 to make it mandatory for services with dynamic users.
885 Contributions from: Adam Jackson, Alexander Tsoy, Andrey Yashkin,
886 Andrzej Pietrasiewicz, Anita Zhang, Balint Reczey, Beniamino Galvani,
887 Ben Iofel, Benjamin Berg, Benjamin Dahlhoff, Chris, Chris Morin,
888 Christopher Wong, Claudius Ellsel, Clemens Gruber, dana, Daniel Black,
889 Davide Cavalca, David Michael, David Rheinsberg, emersion, Evgeny
890 Vereshchagin, Filipe Brandenburger, Franck Bui, Frantisek Sumsal,
891 Giacinto Cifelli, Hans de Goede, Hugo Kindel, Ignat Korchagin, Insun
892 Pyo, Jan Engelhardt, Jonas Dorel, Jonathan Lebon, Jonathon Kowalski,
893 Jörg Sommer, Jörg Thalheim, Jussi Pakkanen, Kai-Heng Feng, Lennart
894 Poettering, Lubomir Rintel, Luís Ferreira, Martin Pitt, Matthias
895 Klumpp, Michael Biebl, Michael Niewöhner, Michael Olbrich, Michal
896 Sekletar, Mike Lothian, Paul Menzel, Piotr Drąg, Riccardo Schirone,
897 Robin Elvedi, Roman Kulikov, Ronald Tschalär, Ross Burton, Ryan
898 Gonzalez, Sebastian Krzyszkowiak, Stephane Chazelas, StKob, Susant
899 Sahani, Sylvain Plantefève, Szabolcs Fruhwald, Taro Yamada, Theo
900 Ouzhinski, Thomas Haller, Tobias Jungel, Tom Yan, Tony Asleson, Topi
901 Miettinen, unixsysadmin, Van Laser, Vesa Jääskeläinen, Yu, Li-Yu,
902 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
908 * The default locale can now be configured at compile time. Otherwise,
909 a suitable default will be selected automatically (one of C.UTF-8,
912 * The version string shown by systemd and other tools now includes the
913 git commit hash when built from git. An override may be specified
914 during compilation, which is intended to be used by distributions to
915 include the package release information.
917 * systemd-cat can now filter standard input and standard error streams
918 for different syslog priorities using the new --stderr-priority=
921 * systemd-journald and systemd-journal-remote reject entries which
922 contain too many fields (CVE-2018-16865) and set limits on the
923 process' command line length (CVE-2018-16864).
925 * $DBUS_SESSION_BUS_ADDRESS environment variable is set by pam_systemd
928 * A new network device NamePolicy "keep" is implemented for link files,
929 and used by default in 99-default.link (the fallback configuration
930 provided by systemd). With this policy, if the network device name
931 was already set by userspace, the device will not be renamed again.
932 This matches the naming scheme that was implemented before
933 systemd-240. If naming-scheme < 240 is specified, the "keep" policy
934 is also enabled by default, even if not specified. Effectively, this
935 means that if naming-scheme >= 240 is specified, network devices will
936 be renamed according to the configuration, even if they have been
937 renamed already, if "keep" is not specified as the naming policy in
938 the .link file. The 99-default.link file provided by systemd includes
939 "keep" for backwards compatibility, but it is recommended for user
940 installed .link files to *not* include it.
942 The "kernel" policy, which keeps kernel names declared to be
943 "persistent", now works again as documented.
945 * kernel-install script now optionally takes the paths to one or more
946 initrd files, and passes them to all plugins.
948 * The mincore() system call has been dropped from the @system-service
949 system call filter group, as it is pretty exotic and may potentially
950 used for side-channel attacks.
952 * -fPIE is dropped from compiler and linker options. Please specify
953 -Db_pie=true option to meson to build position-independent
954 executables. Note that the meson option is supported since meson-0.49.
956 * The fs.protected_regular and fs.protected_fifos sysctls, which were
957 added in Linux 4.19 to make some data spoofing attacks harder, are
958 now enabled by default. While this will hopefully improve the
959 security of most installations, it is technically a backwards
960 incompatible change; to disable these sysctls again, place the
961 following lines in /etc/sysctl.d/60-protected.conf or a similar file:
963 fs.protected_regular = 0
964 fs.protected_fifos = 0
966 Note that the similar hardlink and symlink protection has been
967 enabled since v199, and may be disabled likewise.
969 * The files read from the EnvironmentFile= setting in unit files now
970 parse backslashes inside quotes literally, matching the behaviour of
973 * udevadm trigger, udevadm control, udevadm settle and udevadm monitor
974 now automatically become NOPs when run in a chroot() environment.
976 * The tmpfiles.d/ "C" line type will now copy directory trees not only
977 when the destination is so far missing, but also if it already exists
978 as a directory and is empty. This is useful to cater for systems
979 where directory trees are put together from multiple separate mount
980 points but otherwise empty.
982 * A new function sd_bus_close_unref() (and the associated
983 sd_bus_close_unrefp()) has been added to libsystemd, that combines
984 sd_bus_close() and sd_bus_unref() in one.
986 * udevadm control learnt a new option for --ping for testing whether a
987 systemd-udevd instance is running and reacting.
989 * udevadm trigger learnt a new option for --wait-daemon for waiting
990 systemd-udevd daemon to be initialized.
992 Contributions from: Aaron Plattner, Alberts Muktupāvels, Alex Mayer,
993 Ayman Bagabas, Beniamino Galvani, Burt P, Chris Down, Chris Lamb, Chris
994 Morin, Christian Hesse, Claudius Ellsel, dana, Daniel Axtens, Daniele
995 Medri, Dave Reisner, David Santamaría Rogado, Diego Canuhe, Dimitri
996 John Ledkov, Evgeny Vereshchagin, Fabrice Fontaine, Filipe
997 Brandenburger, Franck Bui, Frantisek Sumsal, govwin, Hans de Goede,
998 James Hilliard, Jan Engelhardt, Jani Uusitalo, Jan Janssen, Jan
999 Synacek, Jonathan McDowell, Jonathan Roemer, Jonathon Kowalski, Joost
1000 Heitbrink, Jörg Thalheim, Lance, Lennart Poettering, Louis Taylor,
1001 Lucas Werkmeister, Mantas Mikulėnas, Marc-Antoine Perennou,
1002 marvelousblack, Michael Biebl, Michael Sloan, Michal Sekletar, Mike
1003 Auty, Mike Gilbert, Mikhail Kasimov, Neil Brown, Niklas Hambüchen,
1004 Patrick Williams, Paul Seyfert, Peter Hutterer, Philip Withnall, Roger
1005 James, Ronnie P. Thomas, Ryan Gonzalez, Sam Morris, Stephan Edel,
1006 Stephan Gerhold, Susant Sahani, Taro Yamada, Thomas Haller, Topi
1007 Miettinen, YiFei Zhu, YmrDtnJu, YunQiang Su, Yu Watanabe, Zbigniew
1008 Jędrzejewski-Szmek, zsergeant77, Дамјан Георгиевски
1010 — Berlin, 2019-02-14
1014 * NoNewPrivileges=yes has been set for all long-running services
1015 implemented by systemd. Previously, this was problematic due to
1016 SELinux (as this would also prohibit the transition from PID1's label
1017 to the service's label). This restriction has since been lifted, but
1018 an SELinux policy update is required.
1019 (See e.g. https://github.com/fedora-selinux/selinux-policy/pull/234.)
1021 * DynamicUser=yes is dropped from systemd-networkd.service,
1022 systemd-resolved.service and systemd-timesyncd.service, which was
1023 enabled in v239 for systemd-networkd.service and systemd-resolved.service,
1024 and since v236 for systemd-timesyncd.service. The users and groups
1025 systemd-network, systemd-resolve and systemd-timesync are created
1026 by systemd-sysusers again. Distributors or system administrators
1027 may need to create these users and groups if they not exist (or need
1028 to re-enable DynamicUser= for those units) while upgrading systemd.
1029 Also, the clock file for systemd-timesyncd may need to move from
1030 /var/lib/private/systemd/timesync/clock to /var/lib/systemd/timesync/clock.
1032 * When unit files are loaded from disk, previously systemd would
1033 sometimes (depending on the unit loading order) load units from the
1034 target path of symlinks in .wants/ or .requires/ directories of other
1035 units. This meant that unit could be loaded from different paths
1036 depending on whether the unit was requested explicitly or as a
1037 dependency of another unit, not honouring the priority of directories
1038 in search path. It also meant that it was possible to successfully
1039 load and start units which are not found in the unit search path, as
1040 long as they were requested as a dependency and linked to from
1041 .wants/ or .requires/. The target paths of those symlinks are not
1042 used for loading units anymore and the unit file must be found in
1045 * A new service type has been added: Type=exec. It's very similar to
1046 Type=simple but ensures the service manager will wait for both fork()
1047 and execve() of the main service binary to complete before proceeding
1048 with follow-up units. This is primarily useful so that the manager
1049 propagates any errors in the preparation phase of service execution
1050 back to the job that requested the unit to be started. For example,
1051 consider a service that has ExecStart= set to a file system binary
1052 that doesn't exist. With Type=simple starting the unit would be
1053 considered instantly successful, as only fork() has to complete
1054 successfully and the manager does not wait for execve(), and hence
1055 its failure is seen "too late". With the new Type=exec service type
1056 starting the unit will fail, as the manager will wait for the
1057 execve() and notice its failure, which is then propagated back to the
1060 NOTE: with the next release 241 of systemd we intend to change the
1061 systemd-run tool to default to Type=exec for transient services
1062 started by it. This should be mostly safe, but in specific corner
1063 cases might result in problems, as the systemd-run tool will then
1064 block on NSS calls (such as user name look-ups due to User=) done
1065 between the fork() and execve(), which under specific circumstances
1066 might cause problems. It is recommended to specify "-p Type=simple"
1067 explicitly in the few cases where this applies. For regular,
1068 non-transient services (i.e. those defined with unit files on disk)
1069 we will continue to default to Type=simple.
1071 * The Linux kernel's current default RLIMIT_NOFILE resource limit for
1072 userspace processes is set to 1024 (soft) and 4096
1073 (hard). Previously, systemd passed this on unmodified to all
1074 processes it forked off. With this systemd release the hard limit
1075 systemd passes on is increased to 512K, overriding the kernel's
1076 defaults and substantially increasing the number of simultaneous file
1077 descriptors unprivileged userspace processes can allocate. Note that
1078 the soft limit remains at 1024 for compatibility reasons: the
1079 traditional UNIX select() call cannot deal with file descriptors >=
1080 1024 and increasing the soft limit globally might thus result in
1081 programs unexpectedly allocating a high file descriptor and thus
1082 failing abnormally when attempting to use it with select() (of
1083 course, programs shouldn't use select() anymore, and prefer
1084 poll()/epoll, but the call unfortunately remains undeservedly popular
1085 at this time). This change reflects the fact that file descriptor
1086 handling in the Linux kernel has been optimized in more recent
1087 kernels and allocating large numbers of them should be much cheaper
1088 both in memory and in performance than it used to be. Programs that
1089 want to take benefit of the increased limit have to "opt-in" into
1090 high file descriptors explicitly by raising their soft limit. Of
1091 course, when they do that they must acknowledge that they cannot use
1092 select() anymore (and neither can any shared library they use — or
1093 any shared library used by any shared library they use and so on).
1094 Which default hard limit is most appropriate is of course hard to
1095 decide. However, given reports that ~300K file descriptors are used
1096 in real-life applications we believe 512K is sufficiently high as new
1097 default for now. Note that there are also reports that using very
1098 high hard limits (e.g. 1G) is problematic: some software allocates
1099 large arrays with one element for each potential file descriptor
1100 (Java, …) — a high hard limit thus triggers excessively large memory
1101 allocations in these applications. Hopefully, the new default of 512K
1102 is a good middle ground: higher than what real-life applications
1103 currently need, and low enough for avoid triggering excessively large
1104 allocations in problematic software. (And yes, somebody should fix
1107 * The fs.nr_open and fs.file-max sysctls are now automatically bumped
1108 to the highest possible values, as separate accounting of file
1109 descriptors is no longer necessary, as memcg tracks them correctly as
1110 part of the memory accounting anyway. Thus, from the four limits on
1111 file descriptors currently enforced (fs.file-max, fs.nr_open,
1112 RLIMIT_NOFILE hard, RLIMIT_NOFILE soft) we turn off the first two,
1113 and keep only the latter two. A set of build-time options
1114 (-Dbump-proc-sys-fs-file-max=false and -Dbump-proc-sys-fs-nr-open=false)
1115 has been added to revert this change in behaviour, which might be
1116 an option for systems that turn off memcg in the kernel.
1118 * When no /etc/locale.conf file exists (and hence no locale settings
1119 are in place), systemd will now use the "C.UTF-8" locale by default,
1120 and set LANG= to it. This locale is supported by various
1121 distributions including Fedora, with clear indications that upstream
1122 glibc is going to make it available too. This locale enables UTF-8
1123 mode by default, which appears appropriate for 2018.
1125 * The "net.ipv4.conf.all.rp_filter" sysctl will now be set to 2 by
1126 default. This effectively switches the RFC3704 Reverse Path filtering
1127 from Strict mode to Loose mode. This is more appropriate for hosts
1128 that have multiple links with routes to the same networks (e.g.
1129 a client with a Wi-Fi and Ethernet both connected to the internet).
1131 Consult the kernel documentation for details on this sysctl:
1132 https://www.kernel.org/doc/Documentation/networking/ip-sysctl.txt
1134 * CPUAccounting=yes no longer enables the CPU controller when using
1135 kernel 4.15+ and the unified cgroup hierarchy, as required accounting
1136 statistics are now provided independently from the CPU controller.
1138 * Support for disabling a particular cgroup controller within a sub-tree
1139 has been added through the DisableControllers= directive.
1141 * cgroup_no_v1=all on the kernel command line now also implies
1142 using the unified cgroup hierarchy, unless one explicitly passes
1143 systemd.unified_cgroup_hierarchy=0 on the kernel command line.
1145 * The new "MemoryMin=" unit file property may now be used to set the
1146 memory usage protection limit of processes invoked by the unit. This
1147 controls the cgroup v2 memory.min attribute. Similarly, the new
1148 "IODeviceLatencyTargetSec=" property has been added, wrapping the new
1149 cgroup v2 io.latency cgroup property for configuring per-service I/O
1152 * systemd now supports the cgroup v2 devices BPF logic, as counterpart
1153 to the cgroup v1 "devices" cgroup controller.
1155 * systemd-escape now is able to combine --unescape with --template. It
1156 also learnt a new option --instance for extracting and unescaping the
1157 instance part of a unit name.
1159 * sd-bus now provides the sd_bus_message_readv() which is similar to
1160 sd_bus_message_read() but takes a va_list object. The pair
1161 sd_bus_set_method_call_timeout() and sd_bus_get_method_call_timeout()
1162 has been added for configuring the default method call timeout to
1163 use. sd_bus_error_move() may be used to efficiently move the contents
1164 from one sd_bus_error structure to another, invalidating the
1165 source. sd_bus_set_close_on_exit() and sd_bus_get_close_on_exit() may
1166 be used to control whether a bus connection object is automatically
1167 flushed when an sd-event loop is exited.
1169 * When processing classic BSD syslog log messages, journald will now
1170 save the original time-stamp string supplied in the new
1171 SYSLOG_TIMESTAMP= journal field. This permits consumers to
1172 reconstruct the original BSD syslog message more correctly.
1174 * StandardOutput=/StandardError= in service files gained support for
1175 new "append:…" parameters, for connecting STDOUT/STDERR of a service
1176 to a file, and appending to it.
1178 * The signal to use as last step of killing of unit processes is now
1179 configurable. Previously it was hard-coded to SIGKILL, which may now
1180 be overridden with the new KillSignal= setting. Note that this is the
1181 signal used when regular termination (i.e. SIGTERM) does not suffice.
1182 Similarly, the signal used when aborting a program in case of a
1183 watchdog timeout may now be configured too (WatchdogSignal=).
1185 * The XDG_SESSION_DESKTOP environment variable may now be configured in
1186 the pam_systemd argument line, using the new desktop= switch. This is
1187 useful to initialize it properly from a display manager without
1188 having to touch C code.
1190 * Most configuration options that previously accepted percentage values
1191 now also accept permille values with the '‰' suffix (instead of '%').
1193 * systemd-resolved may now optionally use OpenSSL instead of GnuTLS for
1196 * systemd-resolved's configuration file resolved.conf gained a new
1197 option ReadEtcHosts= which may be used to turn off processing and
1198 honoring /etc/hosts entries.
1200 * The "--wait" switch may now be passed to "systemctl
1201 is-system-running", in which case the tool will synchronously wait
1202 until the system finished start-up.
1204 * hostnamed gained a new bus call to determine the DMI product UUID.
1206 * On x86-64 systemd will now prefer using the RDRAND processor
1207 instruction over /dev/urandom whenever it requires randomness that
1208 neither has to be crypto-grade nor should be reproducible. This
1209 should substantially reduce the amount of entropy systemd requests
1210 from the kernel during initialization on such systems, though not
1211 reduce it to zero. (Why not zero? systemd still needs to allocate
1212 UUIDs and such uniquely, which require high-quality randomness.)
1214 * networkd gained support for Foo-Over-UDP, ERSPAN and ISATAP
1215 tunnels. It also gained a new option ForceDHCPv6PDOtherInformation=
1216 for forcing the "Other Information" bit in IPv6 RA messages. The
1217 bonding logic gained four new options AdActorSystemPriority=,
1218 AdUserPortKey=, AdActorSystem= for configuring various 802.3ad
1219 aspects, and DynamicTransmitLoadBalancing= for enabling dynamic
1220 shuffling of flows. The tunnel logic gained a new
1221 IPv6RapidDeploymentPrefix= option for configuring IPv6 Rapid
1222 Deployment. The policy rule logic gained four new options IPProtocol=,
1223 SourcePort= and DestinationPort=, InvertRule=. The bridge logic gained
1224 support for the MulticastToUnicast= option. networkd also gained
1225 support for configuring static IPv4 ARP or IPv6 neighbor entries.
1227 * .preset files (as read by 'systemctl preset') may now be used to
1228 instantiate services.
1230 * /etc/crypttab now understands the sector-size= option to configure
1231 the sector size for an encrypted partition.
1233 * Key material for encrypted disks may now be placed on a formatted
1234 medium, and referenced from /etc/crypttab by the UUID of the file
1235 system, followed by "=" suffixed by the path to the key file.
1237 * The "collect" udev component has been removed without replacement, as
1238 it is neither used nor maintained.
1240 * When the RuntimeDirectory=, StateDirectory=, CacheDirectory=,
1241 LogsDirectory=, ConfigurationDirectory= settings are used in a
1242 service the executed processes will now receive a set of environment
1243 variables containing the full paths of these directories.
1244 Specifically, RUNTIME_DIRECTORY=, STATE_DIRECTORY, CACHE_DIRECTORY,
1245 LOGS_DIRECTORY, CONFIGURATION_DIRECTORY are now set if these options
1246 are used. Note that these options may be used multiple times per
1247 service in which case the resulting paths will be concatenated and
1248 separated by colons.
1250 * Predictable interface naming has been extended to cover InfiniBand
1251 NICs. They will be exposed with an "ib" prefix.
1253 * tmpfiles.d/ line types may now be suffixed with a '-' character, in
1254 which case the respective line failing is ignored.
1256 * .link files may now be used to configure the equivalent to the
1257 "ethtool advertise" commands.
1259 * The sd-device.h and sd-hwdb.h APIs are now exported, as an
1260 alternative to libudev.h. Previously, the latter was just an internal
1261 wrapper around the former, but now these two APIs are exposed
1264 * sd-id128.h gained a new function sd_id128_get_boot_app_specific()
1265 which calculates an app-specific boot ID similar to how
1266 sd_id128_get_machine_app_specific() generates an app-specific machine
1269 * A new tool systemd-id128 has been added that can be used to determine
1270 and generate various 128bit IDs.
1272 * /etc/os-release gained two new standardized fields DOCUMENTATION_URL=
1275 * systemd-hibernate-resume-generator will now honor the "noresume"
1276 kernel command line option, in which case it will bypass resuming
1277 from any hibernated image.
1279 * The systemd-sleep.conf configuration file gained new options
1280 AllowSuspend=, AllowHibernation=, AllowSuspendThenHibernate=,
1281 AllowHybridSleep= for prohibiting specific sleep modes even if the
1282 kernel exports them.
1284 * portablectl is now officially supported and has thus moved to
1287 * bootctl learnt the two new commands "set-default" and "set-oneshot"
1288 for setting the default boot loader item to boot to (either
1289 persistently or only for the next boot). This is currently only
1290 compatible with sd-boot, but may be implemented on other boot loaders
1291 too, that follow the boot loader interface. The updated interface is
1292 now documented here:
1294 https://systemd.io/BOOT_LOADER_INTERFACE
1296 * A new kernel command line option systemd.early_core_pattern= is now
1297 understood which may be used to influence the core_pattern PID 1
1298 installs during early boot.
1300 * busctl learnt two new options -j and --json= for outputting method
1301 call replies, properties and monitoring output in JSON.
1303 * journalctl's JSON output now supports simple ANSI coloring as well as
1304 a new "json-seq" mode for generating RFC7464 output.
1306 * Unit files now support the %g/%G specifiers that resolve to the UNIX
1307 group/GID of the service manager runs as, similar to the existing
1308 %u/%U specifiers that resolve to the UNIX user/UID.
1310 * systemd-logind learnt a new global configuration option
1311 UserStopDelaySec= that may be set in logind.conf. It specifies how
1312 long the systemd --user instance shall remain started after a user
1313 logs out. This is useful to speed up repetitive re-connections of the
1314 same user, as it means the user's service manager doesn't have to be
1315 stopped/restarted on each iteration, but can be reused between
1316 subsequent options. This setting defaults to 10s. systemd-logind also
1317 exports two new properties on its Manager D-Bus objects indicating
1318 whether the system's lid is currently closed, and whether the system
1321 * systemd gained support for a generic boot counting logic, which
1322 generically permits automatic reverting to older boot loader entries
1323 if newer updated ones don't work. The boot loader side is implemented
1324 in sd-boot, but is kept open for other boot loaders too. For details
1327 https://systemd.io/AUTOMATIC_BOOT_ASSESSMENT
1329 * The SuccessAction=/FailureAction= unit file settings now learnt two
1330 new parameters: "exit" and "exit-force", which result in immediate
1331 exiting of the service manager, and are only useful in systemd --user
1332 and container environments.
1334 * Unit files gained support for a pair of options
1335 FailureActionExitStatus=/SuccessActionExitStatus= for configuring the
1336 exit status to use as service manager exit status when
1337 SuccessAction=/FailureAction= is set to exit or exit-force.
1339 * A pair of LogRateLimitIntervalSec=/LogRateLimitBurst= per-service
1340 options may now be used to configure the log rate limiting applied by
1341 journald per-service.
1343 * systemd-analyze gained a new verb "timespan" for parsing and
1344 normalizing time span values (i.e. strings like "5min 7s 8us").
1346 * systemd-analyze also gained a new verb "security" for analyzing the
1347 security and sand-boxing settings of services in order to determine an
1348 "exposure level" for them, indicating whether a service would benefit
1349 from more sand-boxing options turned on for them.
1351 * "systemd-analyze syscall-filter" will now also show system calls
1352 supported by the local kernel but not included in any of the defined
1355 * .nspawn files now understand the Ephemeral= setting, matching the
1356 --ephemeral command line switch.
1358 * sd-event gained the new APIs sd_event_source_get_floating() and
1359 sd_event_source_set_floating() for controlling whether a specific
1360 event source is "floating", i.e. destroyed along with the even loop
1363 * Unit objects on D-Bus gained a new "Refs" property that lists all
1364 clients that currently have a reference on the unit (to ensure it is
1367 * The JoinControllers= option in system.conf is no longer supported, as
1368 it didn't work correctly, is hard to support properly, is legacy (as
1369 the concept only exists on cgroup v1) and apparently wasn't used.
1371 * Journal messages that are generated whenever a unit enters the failed
1372 state are now tagged with a unique MESSAGE_ID. Similarly, messages
1373 generated whenever a service process exits are now made recognizable,
1374 too. A tagged message is also emitted whenever a unit enters the
1375 "dead" state on success.
1377 * systemd-run gained a new switch --working-directory= for configuring
1378 the working directory of the service to start. A shortcut -d is
1379 equivalent, setting the working directory of the service to the
1380 current working directory of the invoking program. The new --shell
1381 (or just -S) option has been added for invoking the $SHELL of the
1382 caller as a service, and implies --pty --same-dir --wait --collect
1383 --service-type=exec. Or in other words, "systemd-run -S" is now the
1384 quickest way to quickly get an interactive in a fully clean and
1385 well-defined system service context.
1387 * machinectl gained a new verb "import-fs" for importing an OS tree
1388 from a directory. Moreover, when a directory or tarball is imported
1389 and single top-level directory found with the OS itself below the OS
1390 tree is automatically mangled and moved one level up.
1392 * systemd-importd will no longer set up an implicit btrfs loop-back
1393 file system on /var/lib/machines. If one is already set up, it will
1394 continue to be used.
1396 * A new generator "systemd-run-generator" has been added. It will
1397 synthesize a unit from one or more program command lines included in
1398 the kernel command line. This is very useful in container managers
1401 # systemd-nspawn -i someimage.raw -b systemd.run='"some command line"'
1403 This will run "systemd-nspawn" on an image, invoke the specified
1404 command line and immediately shut down the container again, returning
1405 the command line's exit code.
1407 * The block device locking logic is now documented:
1409 https://systemd.io/BLOCK_DEVICE_LOCKING
1411 * loginctl and machinectl now optionally output the various tables in
1412 JSON using the --output= switch. It is our intention to add similar
1413 support to systemctl and all other commands.
1415 * udevadm's query and trigger verb now optionally take a .device unit
1418 * systemd-udevd's network naming logic now understands a new
1419 net.naming-scheme= kernel command line switch, which may be used to
1420 pick a specific version of the naming scheme. This helps stabilizing
1421 interface names even as systemd/udev are updated and the naming logic
1424 * sd-id128.h learnt two new auxiliary helpers: sd_id128_is_allf() and
1425 SD_ID128_ALLF to test if a 128bit ID is set to all 0xFF bytes, and to
1426 initialize one to all 0xFF.
1428 * After loading the SELinux policy systemd will now recursively relabel
1429 all files and directories listed in
1430 /run/systemd/relabel-extra.d/*.relabel (which should be simple
1431 newline separated lists of paths) in addition to the ones it already
1432 implicitly relabels in /run, /dev and /sys. After the relabelling is
1433 completed the *.relabel files (and /run/systemd/relabel-extra.d/) are
1434 removed. This is useful to permit initrds (i.e. code running before
1435 the SELinux policy is in effect) to generate files in the host
1436 filesystem safely and ensure that the correct label is applied during
1437 the transition to the host OS.
1439 * KERNEL API BREAKAGE: Linux kernel 4.18 changed behaviour regarding
1440 mknod() handling in user namespaces. Previously mknod() would always
1441 fail with EPERM in user namespaces. Since 4.18 mknod() will succeed
1442 but device nodes generated that way cannot be opened, and attempts to
1443 open them result in EPERM. This breaks the "graceful fallback" logic
1444 in systemd's PrivateDevices= sand-boxing option. This option is
1445 implemented defensively, so that when systemd detects it runs in a
1446 restricted environment (such as a user namespace, or an environment
1447 where mknod() is blocked through seccomp or absence of CAP_SYS_MKNOD)
1448 where device nodes cannot be created the effect of PrivateDevices= is
1449 bypassed (following the logic that 2nd-level sand-boxing is not
1450 essential if the system systemd runs in is itself already sand-boxed
1451 as a whole). This logic breaks with 4.18 in container managers where
1452 user namespacing is used: suddenly PrivateDevices= succeeds setting
1453 up a private /dev/ file system containing devices nodes — but when
1454 these are opened they don't work.
1456 At this point it is recommended that container managers utilizing
1457 user namespaces that intend to run systemd in the payload explicitly
1458 block mknod() with seccomp or similar, so that the graceful fallback
1461 We are very sorry for the breakage and the requirement to change
1462 container configurations for newer kernels. It's purely caused by an
1463 incompatible kernel change. The relevant kernel developers have been
1464 notified about this userspace breakage quickly, but they chose to
1467 * PermissionsStartOnly= setting is deprecated (but is still supported
1468 for backwards compatibility). The same functionality is provided by
1469 the more flexible "+", "!", and "!!" prefixes to ExecStart= and other
1472 * $DBUS_SESSION_BUS_ADDRESS environment variable is not set by
1473 pam_systemd anymore.
1475 * The naming scheme for network devices was changed to always rename
1476 devices, even if they were already renamed by userspace. The "kernel"
1477 policy was changed to only apply as a fallback, if no other naming
1480 * The requirements to build systemd is bumped to meson-0.46 and
1483 Contributions from: afg, Alan Jenkins, Aleksei Timofeyev, Alexander
1484 Filippov, Alexander Kurtz, Alexey Bogdanenko, Andreas Henriksson,
1485 Andrew Jorgensen, Anita Zhang, apnix-uk, Arkan49, Arseny Maslennikov,
1486 asavah, Asbjørn Apeland, aszlig, Bastien Nocera, Ben Boeckel, Benedikt
1487 Morbach, Benjamin Berg, Bruce Zhang, Carlo Caione, Cedric Viou, Chen
1488 Qi, Chris Chiu, Chris Down, Chris Morin, Christian Rebischke, Claudius
1489 Ellsel, Colin Guthrie, dana, Daniel, Daniele Medri, Daniel Kahn
1490 Gillmor, Daniel Rusek, Daniel van Vugt, Dariusz Gadomski, Dave Reisner,
1491 David Anderson, Davide Cavalca, David Leeds, David Malcolm, David
1492 Strauss, David Tardon, Dimitri John Ledkov, Dmitry Torokhov, dj-kaktus,
1493 Dongsu Park, Elias Probst, Emil Soleyman, Erik Kooistra, Ervin Peters,
1494 Evgeni Golov, Evgeny Vereshchagin, Fabrice Fontaine, Faheel Ahmad,
1495 Faizal Luthfi, Felix Yan, Filipe Brandenburger, Franck Bui, Frank
1496 Schaefer, Frantisek Sumsal, Gautier Husson, Gianluca Boiano, Giuseppe
1497 Scrivano, glitsj16, Hans de Goede, Harald Hoyer, Harry Mallon, Harshit
1498 Jain, Helmut Grohne, Henry Tung, Hui Yiqun, imayoda, Insun Pyo, Iwan
1499 Timmer, Jan Janssen, Jan Pokorný, Jan Synacek, Jason A. Donenfeld,
1500 javitoom, Jérémy Nouhaud, Jeremy Su, Jiuyang Liu, João Paulo Rechi
1501 Vita, Joe Hershberger, Joe Rayhawk, Joerg Behrmann, Joerg Steffens,
1502 Jonas Dorel, Jon Ringle, Josh Soref, Julian Andres Klode, Jun Bo Bi,
1503 Jürg Billeter, Keith Busch, Khem Raj, Kirill Marinushkin, Larry
1504 Bernstone, Lennart Poettering, Lion Yang, Li Song, Lorenz
1505 Hübschle-Schneider, Lubomir Rintel, Lucas Werkmeister, Ludwin Janvier,
1506 Lukáš Nykrýn, Luke Shumaker, mal, Marc-Antoine Perennou, Marcin
1507 Skarbek, Marco Trevisan (Treviño), Marian Cepok, Mario Hros, Marko
1508 Myllynen, Markus Grimm, Martin Pitt, Martin Sobotka, Martin Wilck,
1509 Mathieu Trudel-Lapierre, Matthew Leeds, Michael Biebl, Michael Olbrich,
1510 Michael 'pbone' Pobega, Michael Scherer, Michal Koutný, Michal
1511 Sekletar, Michal Soltys, Mike Gilbert, Mike Palmer, Muhammet Kara, Neal
1512 Gompa, Neil Brown, Network Silence, Niklas Tibbling, Nikolas Nyby,
1513 Nogisaka Sadata, Oliver Smith, Patrik Flykt, Pavel Hrdina, Paweł
1514 Szewczyk, Peter Hutterer, Piotr Drąg, Ray Strode, Reinhold Mueller,
1515 Renaud Métrich, Roman Gushchin, Ronny Chevalier, Rubén Suárez Alvarez,
1516 Ruixin Bao, RussianNeuroMancer, Ryutaroh Matsumoto, Saleem Rashid, Sam
1517 Morris, Samuel Morris, Sandy Carter, scootergrisen, Sébastien Bacher,
1518 Sergey Ptashnick, Shawn Landden, Shengyao Xue, Shih-Yuan Lee
1519 (FourDollars), Silvio Knizek, Sjoerd Simons, Stasiek Michalski, Stephen
1520 Gallagher, Steven Allen, Steve Ramage, Susant Sahani, Sven Joachim,
1521 Sylvain Plantefève, Tanu Kaskinen, Tejun Heo, Thiago Macieira, Thomas
1522 Blume, Thomas Haller, Thomas H. P. Andersen, Tim Ruffing, TJ, Tobias
1523 Jungel, Todd Walton, Tommi Rantala, Tomsod M, Tony Novak, Tore
1524 Anderson, Trevonn, Victor Laskurain, Victor Tapia, Violet Halo, Vojtech
1525 Trefny, welaq, William A. Kennington III, William Douglas, Wyatt Ward,
1526 Xiang Fan, Xi Ruoyao, Xuanwo, Yann E. Morin, YmrDtnJu, Yu Watanabe,
1527 Zbigniew Jędrzejewski-Szmek, Zhang Xianwei, Zsolt Dollenstein
1529 — Warsaw, 2018-12-21
1533 * NETWORK INTERFACE DEVICE NAMING CHANGES: systemd-udevd's "net_id"
1534 builtin will name network interfaces differently than in previous
1535 versions for virtual network interfaces created with SR-IOV and NPAR
1536 and for devices where the PCI network controller device does not have
1537 a slot number associated.
1539 SR-IOV virtual devices are now named based on the name of the parent
1540 interface, with a suffix of "v<N>", where <N> is the virtual device
1541 number. Previously those virtual devices were named as if completely
1544 The ninth and later NPAR virtual devices will be named following the
1545 scheme used for the first eight NPAR partitions. Previously those
1546 devices were not renamed and the kernel default (eth<n>) was used.
1548 "net_id" will also generate names for PCI devices where the PCI
1549 network controller device does not have an associated slot number
1550 itself, but one of its parents does. Previously those devices were
1551 not renamed and the kernel default (eth<n>) was used.
1553 * AF_INET and AF_INET6 are dropped from RestrictAddressFamilies= in
1554 systemd-logind.service. Since v235, IPAddressDeny=any has been set to
1555 the unit. So, it is expected that the default behavior of
1556 systemd-logind is not changed. However, if distribution packagers or
1557 administrators disabled or modified IPAddressDeny= setting by a
1558 drop-in config file, then it may be necessary to update the file to
1559 re-enable AF_INET and AF_INET6 to support network user name services,
1562 * When the RestrictNamespaces= unit property is specified multiple
1563 times, then the specified types are merged now. Previously, only the
1564 last assignment was used. So, if distribution packagers or
1565 administrators modified the setting by a drop-in config file, then it
1566 may be necessary to update the file.
1568 * When OnFailure= is used in combination with Restart= on a service
1569 unit, then the specified units will no longer be triggered on
1570 failures that result in restarting. Previously, the specified units
1571 would be activated each time the unit failed, even when the unit was
1572 going to be restarted automatically. This behaviour contradicted the
1573 documentation. With this release the code is adjusted to match the
1576 * systemd-tmpfiles will now print a notice whenever it encounters
1577 tmpfiles.d/ lines referencing the /var/run/ directory. It will
1578 recommend reworking them to use the /run/ directory instead (for
1579 which /var/run/ is simply a symlinked compatibility alias). This way
1580 systemd-tmpfiles can properly detect line conflicts and merge lines
1581 referencing the same file by two paths, without having to access
1584 * systemctl disable/unmask/preset/preset-all cannot be used with
1585 --runtime. Previously this was allowed, but resulted in unintuitive
1586 behaviour that wasn't useful. systemctl disable/unmask will now undo
1587 both runtime and persistent enablement/masking, i.e. it will remove
1588 any relevant symlinks both in /run and /etc.
1590 * Note that all long-running system services shipped with systemd will
1591 now default to a system call whitelist (rather than a blacklist, as
1592 before). In particular, systemd-udevd will now enforce one too. For
1593 most cases this should be safe, however downstream distributions
1594 which disabled sandboxing of systemd-udevd (specifically the
1595 MountFlags= setting), might want to disable this security feature
1596 too, as the default whitelisting will prohibit all mount, swap,
1597 reboot and clock changing operations from udev rules.
1599 * sd-boot acquired new loader configuration settings to optionally turn
1600 off Windows and MacOS boot partition discovery as well as
1601 reboot-into-firmware menu items. It is also able to pick a better
1602 screen resolution for HiDPI systems, and now provides loader
1603 configuration settings to change the resolution explicitly.
1605 * systemd-resolved now supports DNS-over-TLS. It's still
1606 turned off by default, use DNSOverTLS=opportunistic to turn it on in
1607 resolved.conf. We intend to make this the default as soon as couple
1608 of additional techniques for optimizing the initial latency caused by
1609 establishing a TLS/TCP connection are implemented.
1611 * systemd-resolved.service and systemd-networkd.service now set
1612 DynamicUser=yes. The users systemd-resolve and systemd-network are
1613 not created by systemd-sysusers anymore.
1615 NOTE: This has a chance of breaking nss-ldap and similar NSS modules
1616 that embed a network facing module into any process using getpwuid()
1617 or related call: the dynamic allocation of the user ID for
1618 systemd-resolved.service means the service manager has to check NSS
1619 if the user name is already taken when forking off the service. Since
1620 the user in the common case won't be defined in /etc/passwd the
1621 lookup is likely to trigger nss-ldap which in turn might use NSS to
1622 ask systemd-resolved for hostname lookups. This will hence result in
1623 a deadlock: a user name lookup in order to start
1624 systemd-resolved.service will result in a host name lookup for which
1625 systemd-resolved.service needs to be started already. There are
1626 multiple ways to work around this problem: pre-allocate the
1627 "systemd-resolve" user on such systems, so that nss-ldap won't be
1628 triggered; or use a different NSS package that doesn't do networking
1629 in-process but provides a local asynchronous name cache; or configure
1630 the NSS package to avoid lookups for UIDs in the range `pkg-config
1631 systemd --variable=dynamicuidmin` … `pkg-config systemd
1632 --variable=dynamicuidmax`, so that it does not consider itself
1633 authoritative for the same UID range systemd allocates dynamic users
1636 * The systemd-resolve tool has been renamed to resolvectl (it also
1637 remains available under the old name, for compatibility), and its
1638 interface is now verb-based, similar in style to the other <xyz>ctl
1639 tools, such as systemctl or loginctl.
1641 * The resolvectl/systemd-resolve tool also provides 'resolvconf'
1642 compatibility. It may be symlinked under the 'resolvconf' name, in
1643 which case it will take arguments and input compatible with the
1644 Debian and FreeBSD resolvconf tool.
1646 * Support for suspend-then-hibernate has been added, i.e. a sleep mode
1647 where the system initially suspends, and after a timeout resumes and
1650 * networkd's ClientIdentifier= now accepts a new option "duid-only". If
1651 set the client will only send a DUID as client identifier.
1653 * The nss-systemd glibc NSS module will now enumerate dynamic users and
1654 groups in effect. Previously, it could resolve UIDs/GIDs to user
1655 names/groups and vice versa, but did not support enumeration.
1657 * journald's Compress= configuration setting now optionally accepts a
1658 byte threshold value. All journal objects larger than this threshold
1659 will be compressed, smaller ones will not. Previously this threshold
1660 was not configurable and set to 512.
1662 * A new system.conf setting NoNewPrivileges= is now available which may
1663 be used to turn off acquisition of new privileges system-wide
1664 (i.e. set Linux' PR_SET_NO_NEW_PRIVS for PID 1 itself, and thus also
1665 for all its children). Note that turning this option on means setuid
1666 binaries and file system capabilities lose their special powers.
1667 While turning on this option is a big step towards a more secure
1668 system, doing so is likely to break numerous pre-existing UNIX tools,
1669 in particular su and sudo.
1671 * A new service systemd-time-sync-wait.service has been added. If
1672 enabled it will delay the time-sync.target unit at boot until time
1673 synchronization has been received from the network. This
1674 functionality is useful on systems lacking a local RTC or where it is
1675 acceptable that the boot process shall be delayed by external network
1678 * When hibernating, systemd will now inform the kernel of the image
1679 write offset, on kernels new enough to support this. This means swap
1680 files should work for hibernation now.
1682 * When loading unit files, systemd will now look for drop-in unit files
1683 extensions in additional places. Previously, for a unit file name
1684 "foo-bar-baz.service" it would look for dropin files in
1685 "foo-bar-baz.service.d/*.conf". Now, it will also look in
1686 "foo-bar-.service.d/*.conf" and "foo-.service.d/", i.e. at the
1687 service name truncated after all inner dashes. This scheme allows
1688 writing drop-ins easily that apply to a whole set of unit files at
1689 once. It's particularly useful for mount and slice units (as their
1690 naming is prefix based), but is also useful for service and other
1691 units, for packages that install multiple unit files at once,
1692 following a strict naming regime of beginning the unit file name with
1693 the package's name. Two new specifiers are now supported in unit
1694 files to match this: %j and %J are replaced by the part of the unit
1695 name following the last dash.
1697 * Unit files and other configuration files that support specifier
1698 expansion now understand another three new specifiers: %T and %V will
1699 resolve to /tmp and /var/tmp respectively, or whatever temporary
1700 directory has been set for the calling user. %E will expand to either
1701 /etc (for system units) or $XDG_CONFIG_HOME (for user units).
1703 * The ExecStart= lines of unit files are no longer required to
1704 reference absolute paths. If non-absolute paths are specified the
1705 specified binary name is searched within the service manager's
1706 built-in $PATH, which may be queried with 'systemd-path
1707 search-binaries-default'. It's generally recommended to continue to
1708 use absolute paths for all binaries specified in unit files.
1710 * Units gained a new load state "bad-setting", which is used when a
1711 unit file was loaded, but contained fatal errors which prevent it
1712 from being started (for example, a service unit has been defined
1713 lacking both ExecStart= and ExecStop= lines).
1715 * coredumpctl's "gdb" verb has been renamed to "debug", in order to
1716 support alternative debuggers, for example lldb. The old name
1717 continues to be available however, for compatibility reasons. Use the
1718 new --debugger= switch or the $SYSTEMD_DEBUGGER environment variable
1719 to pick an alternative debugger instead of the default gdb.
1721 * systemctl and the other tools will now output escape sequences that
1722 generate proper clickable hyperlinks in various terminal emulators
1723 where useful (for example, in the "systemctl status" output you can
1724 now click on the unit file name to quickly open it in the
1725 editor/viewer of your choice). Note that not all terminal emulators
1726 support this functionality yet, but many do. Unfortunately, the
1727 "less" pager doesn't support this yet, hence this functionality is
1728 currently automatically turned off when a pager is started (which
1729 happens quite often due to auto-paging). We hope to remove this
1730 limitation as soon as "less" learns these escape sequences. This new
1731 behaviour may also be turned off explicitly with the $SYSTEMD_URLIFY
1732 environment variable. For details on these escape sequences see:
1733 https://gist.github.com/egmontkob/eb114294efbcd5adb1944c9f3cb5feda
1735 * networkd's .network files now support a new IPv6MTUBytes= option for
1736 setting the MTU used by IPv6 explicitly as well as a new MTUBytes=
1737 option in the [Route] section to configure the MTU to use for
1738 specific routes. It also gained support for configuration of the DHCP
1739 "UserClass" option through the new UserClass= setting. It gained
1740 three new options in the new [CAN] section for configuring CAN
1741 networks. The MULTICAST and ALLMULTI interface flags may now be
1742 controlled explicitly with the new Multicast= and AllMulticast=
1745 * networkd will now automatically make use of the kernel's route
1746 expiration feature, if it is available.
1748 * udevd's .link files now support setting the number of receive and
1749 transmit channels, using the RxChannels=, TxChannels=,
1750 OtherChannels=, CombinedChannels= settings.
1752 * Support for UDPSegmentationOffload= has been removed, given its
1753 limited support in hardware, and waning software support.
1755 * networkd's .netdev files now support creating "netdevsim" interfaces.
1757 * PID 1 learnt a new bus call GetUnitByControlGroup() which may be used
1758 to query the unit belonging to a specific kernel control group.
1760 * systemd-analyze gained a new verb "cat-config", which may be used to
1761 dump the contents of any configuration file, with all its matching
1762 drop-in files added in, and honouring the usual search and masking
1763 logic applied to systemd configuration files. For example use
1764 "systemd-analyze cat-config systemd/system.conf" to get the complete
1765 system configuration file of systemd how it would be loaded by PID 1
1766 itself. Similar to this, various tools such as systemd-tmpfiles or
1767 systemd-sysusers, gained a new option "--cat-config", which does the
1768 corresponding operation for their own configuration settings. For
1769 example, "systemd-tmpfiles --cat-config" will now output the full
1770 list of tmpfiles.d/ lines in place.
1772 * timedatectl gained three new verbs: "show" shows bus properties of
1773 systemd-timedated, "timesync-status" shows the current NTP
1774 synchronization state of systemd-timesyncd, and "show-timesync"
1775 shows bus properties of systemd-timesyncd.
1777 * systemd-timesyncd gained a bus interface on which it exposes details
1780 * A new environment variable $SYSTEMD_TIMEDATED_NTP_SERVICES is now
1781 understood by systemd-timedated. It takes a colon-separated list of
1782 unit names of NTP client services. The list is used by
1783 "timedatectl set-ntp".
1785 * systemd-nspawn gained a new --rlimit= switch for setting initial
1786 resource limits for the container payload. There's a new switch
1787 --hostname= to explicitly override the container's hostname. A new
1788 --no-new-privileges= switch may be used to control the
1789 PR_SET_NO_NEW_PRIVS flag for the container payload. A new
1790 --oom-score-adjust= switch controls the OOM scoring adjustment value
1791 for the payload. The new --cpu-affinity= switch controls the CPU
1792 affinity of the container payload. The new --resolv-conf= switch
1793 allows more detailed control of /etc/resolv.conf handling of the
1794 container. Similarly, the new --timezone= switch allows more detailed
1795 control of /etc/localtime handling of the container.
1797 * systemd-detect-virt gained a new --list switch, which will print a
1798 list of all currently known VM and container environments.
1800 * Support for "Portable Services" has been added, see
1801 doc/PORTABLE_SERVICES.md for details. Currently, the support is still
1802 experimental, but this is expected to change soon. Reflecting this
1803 experimental state, the "portablectl" binary is not installed into
1804 /usr/bin yet. The binary has to be called with the full path
1805 /usr/lib/systemd/portablectl instead.
1807 * journalctl's and systemctl's -o switch now knows a new log output
1808 mode "with-unit". The output it generates is very similar to the
1809 regular "short" mode, but displays the unit name instead of the
1810 syslog tag for each log line. Also, the date is shown with timezone
1811 information. This mode is probably more useful than the classic
1812 "short" output mode for most purposes, except where pixel-perfect
1813 compatibility with classic /var/log/messages formatting is required.
1815 * A new --dump-bus-properties switch has been added to the systemd
1816 binary, which may be used to dump all supported D-Bus properties.
1817 (Options which are still supported, but are deprecated, are *not*
1820 * sd-bus gained a set of new calls:
1821 sd_bus_slot_set_floating()/sd_bus_slot_get_floating() may be used to
1822 enable/disable the "floating" state of a bus slot object,
1823 i.e. whether the slot object pins the bus it is allocated for into
1824 memory or if the bus slot object gets disconnected when the bus goes
1825 away. sd_bus_open_with_description(),
1826 sd_bus_open_user_with_description(),
1827 sd_bus_open_system_with_description() may be used to allocate bus
1828 objects and set their description string already during allocation.
1830 * sd-event gained support for watching inotify events from the event
1831 loop, in an efficient way, sharing inotify handles between multiple
1832 users. For this a new function sd_event_add_inotify() has been added.
1834 * sd-event and sd-bus gained support for calling special user-supplied
1835 destructor functions for userdata pointers associated with
1836 sd_event_source, sd_bus_slot, and sd_bus_track objects. For this new
1837 functions sd_bus_slot_set_destroy_callback,
1838 sd_bus_slot_get_destroy_callback, sd_bus_track_set_destroy_callback,
1839 sd_bus_track_get_destroy_callback,
1840 sd_event_source_set_destroy_callback,
1841 sd_event_source_get_destroy_callback have been added.
1843 * The "net.ipv4.tcp_ecn" sysctl will now be turned on by default.
1845 * PID 1 will now automatically reschedule .timer units whenever the
1846 local timezone changes. (They previously got rescheduled
1847 automatically when the system clock changed.)
1849 * New documentation has been added to document cgroups delegation,
1850 portable services and the various code quality tools we have set up:
1852 https://github.com/systemd/systemd/blob/master/docs/CGROUP_DELEGATION.md
1853 https://github.com/systemd/systemd/blob/master/docs/PORTABLE_SERVICES.md
1854 https://github.com/systemd/systemd/blob/master/docs/CODE_QUALITY.md
1856 * The Boot Loader Specification has been added to the source tree.
1858 https://github.com/systemd/systemd/blob/master/docs/BOOT_LOADER_SPECIFICATION.md
1860 While moving it into our source tree we have updated it and further
1861 changes are now accepted through the usual github PR workflow.
1863 * pam_systemd will now look for PAM userdata fields systemd.memory_max,
1864 systemd.tasks_max, systemd.cpu_weight, systemd.io_weight set by
1865 earlier PAM modules. The data in these fields is used to initialize
1866 the session scope's resource properties. Thus external PAM modules
1867 may now configure per-session limits, for example sourced from
1868 external user databases.
1870 * socket units with Accept=yes will now maintain a "refused" counter in
1871 addition to the existing "accepted" counter, counting connections
1872 refused due to the enforced limits.
1874 * The "systemd-path search-binaries-default" command may now be use to
1875 query the default, built-in $PATH PID 1 will pass to the services it
1878 * A new unit file setting PrivateMounts= has been added. It's a boolean
1879 option. If enabled the unit's processes are invoked in their own file
1880 system namespace. Note that this behaviour is also implied if any
1881 other file system namespacing options (such as PrivateTmp=,
1882 PrivateDevices=, ProtectSystem=, …) are used. This option is hence
1883 primarily useful for services that do not use any of the other file
1884 system namespacing options. One such service is systemd-udevd.service
1885 where this is now used by default.
1887 * ConditionSecurity= gained a new value "uefi-secureboot" that is true
1888 when the system is booted in UEFI "secure mode".
1890 * A new unit "system-update-pre.target" is added, which defines an
1891 optional synchronization point for offline system updates, as
1892 implemented by the pre-existing "system-update.target" unit. It
1893 allows ordering services before the service that executes the actual
1894 update process in a generic way.
1896 * Systemd now emits warnings whenever .include syntax is used.
1898 Contributions from: Adam Duskett, Alan Jenkins, Alessandro Casale,
1899 Alexander Kurtz, Alex Gartrell, Anssi Hannula, Arnaud Rebillout, Brian
1900 J. Murrell, Bruno Vernay, Chris Lamb, Chris Lesiak, Christian Brauner,
1901 Christian Hesse, Christian Rebischke, Colin Guthrie, Daniel Dao, Daniel
1902 Lin, Danylo Korostil, Davide Cavalca, David Tardon, Dimitri John
1903 Ledkov, Dmitriy Geels, Douglas Christman, Elia Geretto, emelenas, Emil
1904 Velikov, Evgeny Vereshchagin, Felipe Sateler, Feng Sun, Filipe
1905 Brandenburger, Franck Bui, futpib, Giuseppe Scrivano, Guillem Jover,
1906 guixxx, Hannes Reinecke, Hans de Goede, Harald Hoyer, Henrique Dante de
1907 Almeida, Hiram van Paassen, Ian Miell, Igor Gnatenko, Ivan Shapovalov,
1908 Iwan Timmer, James Cowgill, Jan Janssen, Jan Synacek, Jared Kazimir,
1909 Jérémy Rosen, João Paulo Rechi Vita, Joost Heitbrink, Jui-Chi Ricky
1910 Liang, Jürg Billeter, Kai-Heng Feng, Karol Augustin, Kay Sievers,
1911 Krzysztof Nowicki, Lauri Tirkkonen, Lennart Poettering, Leonard König,
1912 Long Li, Luca Boccassi, Lucas Werkmeister, Marcel Hoppe, Marc
1913 Kleine-Budde, Mario Limonciello, Martin Jansa, Martin Wilck, Mathieu
1914 Malaterre, Matteo F. Vescovi, Matthew McGinn, Matthias-Christian Ott,
1915 Michael Biebl, Michael Olbrich, Michael Prokop, Michal Koutný, Michal
1916 Sekletar, Mike Gilbert, Mikhail Kasimov, Milan Broz, Milan Pässler,
1917 Mladen Pejaković, Muhammet Kara, Nicolas Boichat, Omer Katz, Paride
1918 Legovini, Paul Menzel, Paul Milliken, Pavel Hrdina, Peter A. Bigot,
1919 Peter D'Hoye, Peter Hutterer, Peter Jones, Philip Sequeira, Philip
1920 Withnall, Piotr Drąg, Radostin Stoyanov, Ricardo Salveti de Araujo,
1921 Ronny Chevalier, Rosen Penev, Rubén Suárez Alvarez, Ryan Gonzalez,
1922 Salvo Tomaselli, Sebastian Reichel, Sergey Ptashnick, Sergio Lindo
1923 Mansilla, Stefan Schweter, Stephen Hemminger, Stuart Hayes, Susant
1924 Sahani, Sylvain Plantefève, Thomas H. P. Andersen, Tobias Jungel,
1925 Tomasz Torcz, Vito Caputo, Will Dietz, Will Thompson, Wim van Mourik,
1926 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
1928 — Berlin, 2018-06-22
1932 * The MemoryAccounting= unit property now defaults to on. After
1933 discussions with the upstream control group maintainers we learnt
1934 that the negative impact of cgroup memory accounting on current
1935 kernels is finally relatively minimal, so that it should be safe to
1936 enable this by default without affecting system performance. Besides
1937 memory accounting only task accounting is turned on by default, all
1938 other forms of resource accounting (CPU, IO, IP) remain off for now,
1939 because it's not clear yet that their impact is small enough to move
1940 from opt-in to opt-out. We recommend downstreams to leave memory
1941 accounting on by default if kernel 4.14 or higher is primarily
1942 used. On very resource constrained systems or when support for old
1943 kernels is a necessity, -Dmemory-accounting-default=false can be used
1944 to revert this change.
1946 * rpm scriptlets to update the udev hwdb and rules (%udev_hwdb_update,
1947 %udev_rules_update) and the journal catalog (%journal_catalog_update)
1948 from the upgrade scriptlets of individual packages now do nothing.
1949 Transfiletriggers have been added which will perform those updates
1950 once at the end of the transaction.
1952 Similar transfiletriggers have been added to execute any sysctl.d
1953 and binfmt.d rules. Thus, it should be unnecessary to provide any
1954 scriptlets to execute this configuration from package installation
1957 * systemd-sysusers gained a mode where the configuration to execute is
1958 specified on the command line, but this configuration is not executed
1959 directly, but instead it is merged with the configuration on disk,
1960 and the result is executed. This is useful for package installation
1961 scripts which want to create the user before installing any files on
1962 disk (in case some of those files are owned by that user), while
1963 still allowing local admin overrides.
1965 This functionality is exposed to rpm scriptlets through a new
1966 %sysusers_create_package macro. Old %sysusers_create and
1967 %sysusers_create_inline macros are deprecated.
1969 A transfiletrigger for sysusers.d configuration is now installed,
1970 which means that it should be unnecessary to call systemd-sysusers from
1971 package installation scripts, unless the package installs any files
1972 owned by those newly-created users, in which case
1973 %sysusers_create_package should be used.
1975 * Analogous change has been done for systemd-tmpfiles: it gained a mode
1976 where the command-line configuration is merged with the configuration
1977 on disk. This is exposed as the new %tmpfiles_create_package macro,
1978 and %tmpfiles_create is deprecated. A transfiletrigger is installed
1979 for tmpfiles.d, hence it should be unnecessary to call systemd-tmpfiles
1980 from package installation scripts.
1982 * sysusers.d configuration for a user may now also specify the group
1983 number, in addition to the user number ("u username 123:456"), or
1984 without the user number ("u username -:456").
1986 * Configution items for systemd-sysusers can now be specified as
1987 positional arguments when the new --inline switch is used.
1989 * The login shell of users created through sysusers.d may now be
1990 specified (previously, it was always /bin/sh for root and
1991 /sbin/nologin for other users).
1993 * systemd-analyze gained a new --global switch to look at global user
1994 configuration. It also gained a unit-paths verb to list the unit load
1995 paths that are compiled into systemd (which can be used with
1996 --systemd, --user, or --global).
1998 * udevadm trigger gained a new --settle/-w option to wait for any
1999 triggered events to finish (but just those, and not any other events
2000 which are triggered meanwhile).
2002 * The action that systemd-logind takes when the lid is closed and the
2003 machine is connected to external power can now be configured using
2004 HandleLidSwitchExternalPower= in logind.conf. Previously, this action
2005 was determined by HandleLidSwitch=, and, for backwards compatibility,
2006 is still is, if HandleLidSwitchExternalPower= is not explicitly set.
2008 * journalctl will periodically call sd_journal_process() to make it
2009 resilient against inotify queue overruns when journal files are
2010 rotated very quickly.
2012 * Two new functions in libsystemd — sd_bus_get_n_queued_read and
2013 sd_bus_get_n_queued_write — may be used to check the number of
2014 pending bus messages.
2016 * systemd gained a new
2017 org.freedesktop.systemd1.Manager.AttachProcessesToUnit dbus call
2018 which can be used to migrate foreign processes to scope and service
2019 units. The primary user for this new API is systemd itself: the
2020 systemd --user instance uses this call of the systemd --system
2021 instance to migrate processes if it itself gets the request to
2022 migrate processes and the kernel refuses this due to access
2023 restrictions. Thanks to this "systemd-run --scope --user …" works
2024 again in pure cgroup v2 environments when invoked from the user
2027 * A new TemporaryFileSystem= setting can be used to mask out part of
2028 the real file system tree with tmpfs mounts. This may be combined
2029 with BindPaths= and BindReadOnlyPaths= to hide files or directories
2030 not relevant to the unit, while still allowing some paths lower in
2031 the tree to be accessed.
2033 ProtectHome=tmpfs may now be used to hide user home and runtime
2034 directories from units, in a way that is mostly equivalent to
2035 "TemporaryFileSystem=/home /run/user /root".
2037 * Non-service units are now started with KeyringMode=shared by default.
2038 This means that mount and swapon and other mount tools have access
2039 to keys in the main keyring.
2041 * /sys/fs/bpf is now mounted automatically.
2043 * QNX virtualization is now detected by systemd-detect-virt and may
2044 be used in ConditionVirtualization=.
2046 * IPAccounting= may now be enabled also for slice units.
2048 * A new -Dsplit-bin= build configuration switch may be used to specify
2049 whether bin and sbin directories are merged, or if they should be
2050 included separately in $PATH and various listings of executable
2051 directories. The build configuration scripts will try to autodetect
2052 the proper values of -Dsplit-usr= and -Dsplit-bin= based on build
2053 system, but distributions are encouraged to configure this
2056 * A new -Dok-color= build configuration switch may be used to change
2057 the colour of "OK" status messages.
2059 * UPGRADE ISSUE: serialization of units using JoinsNamespaceOf= with
2060 PrivateNetwork=yes was buggy in previous versions of systemd. This
2061 means that after the upgrade and daemon-reexec, any such units must
2064 * INCOMPATIBILITY: as announced in the NEWS for 237, systemd-tmpfiles
2065 will not exclude read-only files owned by root from cleanup.
2067 Contributions from: Alan Jenkins, Alexander F Rødseth, Alexis Jeandet,
2068 Andika Triwidada, Andrei Gherzan, Ansgar Burchardt, antizealot1337,
2069 Batuhan Osman Taşkaya, Beniamino Galvani, Bill Yodlowsky, Caio Marcelo
2070 de Oliveira Filho, CuBiC, Daniele Medri, Daniel Mouritzen, Daniel
2071 Rusek, Davide Cavalca, Dimitri John Ledkov, Douglas Christman, Evgeny
2072 Vereshchagin, Faalagorn, Filipe Brandenburger, Franck Bui, futpib,
2073 Giacomo Longo, Gunnar Hjalmarsson, Hans de Goede, Hermann Gausterer,
2074 Iago López Galeiras, Jakub Filak, Jan Synacek, Jason A. Donenfeld,
2075 Javier Martinez Canillas, Jérémy Rosen, Lennart Poettering, Lucas
2076 Werkmeister, Mao Huang, Marco Gulino, Michael Biebl, Michael Vogt,
2077 MilhouseVH, Neal Gompa (ニール・ゴンパ), Oleander Reis, Olof Mogren,
2078 Patrick Uiterwijk, Peter Hutterer, Peter Portante, Piotr Drąg, Robert
2079 Antoni Buj Gelonch, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
2080 Fowler, SjonHortensius, snorreflorre, Susant Sahani, Sylvain
2081 Plantefève, Thomas Blume, Thomas Haller, Vito Caputo, Yu Watanabe,
2082 Zbigniew Jędrzejewski-Szmek, Марко М. Костић (Marko M. Kostić)
2084 — Warsaw, 2018-03-05
2088 * Some keyboards come with a zoom see-saw or rocker which until now got
2089 mapped to the Linux "zoomin/out" keys in hwdb. However, these
2090 keycodes are not recognized by any major desktop. They now produce
2091 Up/Down key events so that they can be used for scrolling.
2093 * INCOMPATIBILITY: systemd-tmpfiles' "f" lines changed behaviour
2094 slightly: previously, if an argument was specified for lines of this
2095 type (i.e. the right-most column was set) this string was appended to
2096 existing files each time systemd-tmpfiles was run. This behaviour was
2097 different from what the documentation said, and not particularly
2098 useful, as repeated systemd-tmpfiles invocations would not be
2099 idempotent and grow such files without bounds. With this release
2100 behaviour has been altered to match what the documentation says:
2101 lines of this type only have an effect if the indicated files don't
2102 exist yet, and only then the argument string is written to the file.
2104 * FUTURE INCOMPATIBILITY: In systemd v238 we intend to slightly change
2105 systemd-tmpfiles behaviour: previously, read-only files owned by root
2106 were always excluded from the file "aging" algorithm (i.e. the
2107 automatic clean-up of directories like /tmp based on
2108 atime/mtime/ctime). We intend to drop this restriction, and age files
2109 by default even when owned by root and read-only. This behaviour was
2110 inherited from older tools, but there have been requests to remove
2111 it, and it's not obvious why this restriction was made in the first
2112 place. Please speak up now, if you are aware of software that reqires
2113 this behaviour, otherwise we'll remove the restriction in v238.
2115 * A new environment variable $SYSTEMD_OFFLINE is now understood by
2116 systemctl. It takes a boolean argument. If on, systemctl assumes it
2117 operates on an "offline" OS tree, and will not attempt to talk to the
2118 service manager. Previously, this mode was implicitly enabled if a
2119 chroot() environment was detected, and this new environment variable
2120 now provides explicit control.
2122 * .path and .socket units may now be created transiently, too.
2123 Previously only service, mount, automount and timer units were
2124 supported as transient units. The systemd-run tool has been updated
2125 to expose this new functionality, you may hence use it now to bind
2126 arbitrary commands to path or socket activation on-the-fly from the
2127 command line. Moreover, almost all properties are now exposed for the
2128 unit types that already supported transient operation.
2130 * The systemd-mount command gained support for a new --owner= parameter
2131 which takes a user name, which is then resolved and included in uid=
2132 and gid= mount options string of the file system to mount.
2134 * A new unit condition ConditionControlGroupController= has been added
2135 that checks whether a specific cgroup controller is available.
2137 * Unit files, udev's .link files, and systemd-networkd's .netdev and
2138 .network files all gained support for a new condition
2139 ConditionKernelVersion= for checking against specific kernel
2142 * In systemd-networkd, the [IPVLAN] section in .netdev files gained
2143 support for configuring device flags in the Flags= setting. In the
2144 same files, the [Tunnel] section gained support for configuring
2145 AllowLocalRemote=. The [Route] section in .network files gained
2146 support for configuring InitialCongestionWindow=,
2147 InitialAdvertisedReceiveWindow= and QuickAck=. The [DHCP] section now
2148 understands RapidCommit=.
2150 * systemd-networkd's DHCPv6 support gained support for Prefix
2153 * sd-bus gained support for a new "watch-bind" feature. When this
2154 feature is enabled, an sd_bus connection may be set up to connect to
2155 an AF_UNIX socket in the file system as soon as it is created. This
2156 functionality is useful for writing early-boot services that
2157 automatically connect to the system bus as soon as it is started,
2158 without ugly time-based polling. systemd-networkd and
2159 systemd-resolved have been updated to make use of this
2160 functionality. busctl exposes this functionality in a new
2161 --watch-bind= command line switch.
2163 * sd-bus will now optionally synthesize a local "Connected" signal as
2164 soon as a D-Bus connection is set up fully. This message mirrors the
2165 already existing "Disconnected" signal which is synthesized when the
2166 connection is terminated. This signal is generally useful but
2167 particularly handy in combination with the "watch-bind" feature
2168 described above. Synthesizing of this message has to be requested
2169 explicitly through the new API call sd_bus_set_connected_signal(). In
2170 addition a new call sd_bus_is_ready() has been added that checks
2171 whether a connection is fully set up (i.e. between the "Connected" and
2172 "Disconnected" signals).
2174 * sd-bus gained two new calls sd_bus_request_name_async() and
2175 sd_bus_release_name_async() for asynchronously registering bus
2176 names. Similar, there is now sd_bus_add_match_async() for installing
2177 a signal match asynchronously. All of systemd's own services have
2178 been updated to make use of these calls. Doing these operations
2179 asynchronously has two benefits: it reduces the risk of deadlocks in
2180 case of cyclic dependencies between bus services, and it speeds up
2181 service initialization since synchronization points for bus
2182 round-trips are removed.
2184 * sd-bus gained two new calls sd_bus_match_signal() and
2185 sd_bus_match_signal_async(), which are similar to sd_bus_add_match()
2186 and sd_bus_add_match_async() but instead of taking a D-Bus match
2187 string take match fields as normal function parameters.
2189 * sd-bus gained two new calls sd_bus_set_sender() and
2190 sd_bus_message_set_sender() for setting the sender name of outgoing
2191 messages (either for all outgoing messages or for just one specific
2192 one). These calls are only useful in direct connections as on
2193 brokered connections the broker fills in the sender anyway,
2194 overwriting whatever the client filled in.
2196 * sd-event gained a new pseudo-handle that may be specified on all API
2197 calls where an "sd_event*" object is expected: SD_EVENT_DEFAULT. When
2198 used this refers to the default event loop object of the calling
2199 thread. Note however that this does not implicitly allocate one —
2200 which has to be done prior by using sd_event_default(). Similarly
2201 sd-bus gained three new pseudo-handles SD_BUS_DEFAULT,
2202 SD_BUS_DEFAULT_USER, SD_BUS_DEFAULT_SYSTEM that may be used to refer
2203 to the default bus of the specified type of the calling thread. Here
2204 too this does not implicitly allocate bus connection objects, this
2205 has to be done prior with sd_bus_default() and friends.
2207 * sd-event gained a new call pair
2208 sd_event_source_{get|set}_io_fd_own(). This may be used to request
2209 automatic closure of the file descriptor an IO event source watches
2210 when the event source is destroyed.
2212 * systemd-networkd gained support for natively configuring WireGuard
2215 * In previous versions systemd synthesized user records both for the
2216 "nobody" (UID 65534) and "root" (UID 0) users in nss-systemd and
2217 internally. In order to simplify distribution-wide renames of the
2218 "nobody" user (like it is planned in Fedora: nfsnobody → nobody), a
2219 new transitional flag file has been added: if
2220 /etc/systemd/dont-synthesize-nobody exists synthesizing of the 65534
2221 user and group record within the systemd codebase is disabled.
2223 * systemd-notify gained a new --uid= option for selecting the source
2224 user/UID to use for notification messages sent to the service
2227 * journalctl gained a new --grep= option to list only entries in which
2228 the message matches a certain pattern. By default matching is case
2229 insensitive if the pattern is lowercase, and case sensitive
2230 otherwise. Option --case-sensitive=yes|no can be used to override
2231 this an specify case sensitivity or case insensitivity.
2233 * There's now a "systemd-analyze service-watchdogs" command for printing
2234 the current state of the service runtime watchdog, and optionally
2235 enabling or disabling the per-service watchdogs system-wide if given a
2236 boolean argument (i.e. the concept you configure in WatchdogSec=), for
2237 debugging purposes. There's also a kernel command line option
2238 systemd.service_watchdogs= for controlling the same.
2240 * Two new "log-level" and "log-target" options for systemd-analyze were
2241 added that merge the now deprecated get-log-level, set-log-level and
2242 get-log-target, set-log-target pairs. The deprecated options are still
2243 understood for backwards compatibility. The two new options print the
2244 current value when no arguments are given, and set them when a
2245 level/target is given as an argument.
2247 * sysusers.d's "u" lines now optionally accept both a UID and a GID
2248 specification, separated by a ":" character, in order to create users
2249 where UID and GID do not match.
2251 Contributions from: Adam Duskett, Alan Jenkins, Alexander Kuleshov,
2252 Alexis Deruelle, Andrew Jeddeloh, Armin Widegreen, Batuhan Osman
2253 Taşkaya, Björn Esser, bleep_blop, Bruce A. Johnson, Chris Down, Clinton
2254 Roy, Colin Walters, Daniel Rusek, Dimitri John Ledkov, Dmitry Rozhkov,
2255 Evgeny Vereshchagin, Ewout van Mansom, Felipe Sateler, Franck Bui,
2256 Frantisek Sumsal, George Gaydarov, Gianluca Boiano, Hans-Christian
2257 Noren Egtvedt, Hans de Goede, Henrik Grindal Bakken, Jan Alexander
2258 Steffens, Jan Klötzke, Jason A. Donenfeld, jdkbx, Jérémy Rosen,
2259 Jerónimo Borque, John Lin, John Paul Herold, Jonathan Rudenberg, Jörg
2260 Thalheim, Ken (Bitsko) MacLeod, Larry Bernstone, Lennart Poettering,
2261 Lucas Werkmeister, Maciej S. Szmigiero, Marek Čermák, Martin Pitt,
2262 Mathieu Malaterre, Matthew Thode, Matthias-Christian Ott, Max Harmathy,
2263 Michael Biebl, Michael Vogt, Michal Koutný, Michal Sekletar, Michał
2264 Szczepański, Mike Gilbert, Nathaniel McCallum, Nicolas Chauvet, Olaf
2265 Hering, Olivier Schwander, Patrik Flykt, Paul Cercueil, Peter Hutterer,
2266 Piotr Drąg, Raphael Vogelgsang, Reverend Homer, Robert Kolchmeyer,
2267 Samuel Dionne-Riel, Sergey Ptashnick, Shawn Landden, Susant Sahani,
2268 Sylvain Plantefève, Thomas H. P. Andersen, Thomas Huth, Tomasz
2269 Bachorski, Vladislav Vishnyakov, Wieland Hoffmann, Yu Watanabe, Zachary
2270 Winnerman, Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски, Дилян
2277 * The modprobe.d/ drop-in for the bonding.ko kernel module introduced
2278 in v235 has been extended to also set the dummy.ko module option
2279 numdummies=0, preventing the kernel from automatically creating
2280 dummy0. All dummy interfaces must now be explicitly created.
2282 * Unknown '%' specifiers in configuration files are now rejected. This
2283 applies to units and tmpfiles.d configuration. Any percent characters
2284 that are followed by a letter or digit that are not supposed to be
2285 interpreted as the beginning of a specifier should be escaped by
2286 doubling ("%%"). (So "size=5%" is still accepted, as well as
2287 "size=5%,foo=bar", but not "LABEL=x%y%z" since %y and %z are not
2288 valid specifiers today.)
2290 * systemd-resolved now maintains a new dynamic
2291 /run/systemd/resolve/stub-resolv.conf compatibility file. It is
2292 recommended to make /etc/resolv.conf a symlink to it. This file
2293 points at the systemd-resolved stub DNS 127.0.0.53 resolver and
2294 includes dynamically acquired search domains, achieving more correct
2295 DNS resolution by software that bypasses local DNS APIs such as NSS.
2297 * The "uaccess" udev tag has been dropped from /dev/kvm and
2298 /dev/dri/renderD*. These devices now have the 0666 permissions by
2299 default (but this may be changed at build-time). /dev/dri/renderD*
2300 will now be owned by the "render" group along with /dev/kfd.
2302 * "DynamicUser=yes" has been enabled for systemd-timesyncd.service,
2303 systemd-journal-gatewayd.service and
2304 systemd-journal-upload.service. This means "nss-systemd" must be
2305 enabled in /etc/nsswitch.conf to ensure the UIDs assigned to these
2306 services are resolved properly.
2308 * In /etc/fstab two new mount options are now understood:
2309 x-systemd.makefs and x-systemd.growfs. The former has the effect that
2310 the configured file system is formatted before it is mounted, the
2311 latter that the file system is resized to the full block device size
2312 after it is mounted (i.e. if the file system is smaller than the
2313 partition it resides on, it's grown). This is similar to the fsck
2314 logic in /etc/fstab, and pulls in systemd-makefs@.service and
2315 systemd-growfs@.service as necessary, similar to
2316 systemd-fsck@.service. Resizing is currently only supported on ext4
2319 * In systemd-networkd, the IPv6 RA logic now optionally may announce
2320 DNS server and domain information.
2322 * Support for the LUKS2 on-disk format for encrypted partitions has
2323 been added. This requires libcryptsetup2 during compilation and
2326 * The systemd --user instance will now signal "readiness" when its
2327 basic.target unit has been reached, instead of when the run queue ran
2328 empty for the first time.
2330 * Tmpfiles.d with user configuration are now also supported.
2331 systemd-tmpfiles gained a new --user switch, and snippets placed in
2332 ~/.config/user-tmpfiles.d/ and corresponding directories will be
2333 executed by systemd-tmpfiles --user running in the new
2334 systemd-tmpfiles-setup.service and systemd-tmpfiles-clean.service
2335 running in the user session.
2337 * Unit files and tmpfiles.d snippets learnt three new % specifiers:
2338 %S resolves to the top-level state directory (/var/lib for the system
2339 instance, $XDG_CONFIG_HOME for the user instance), %C resolves to the
2340 top-level cache directory (/var/cache for the system instance,
2341 $XDG_CACHE_HOME for the user instance), %L resolves to the top-level
2342 logs directory (/var/log for the system instance,
2343 $XDG_CONFIG_HOME/log/ for the user instance). This matches the
2344 existing %t specifier, that resolves to the top-level runtime
2345 directory (/run for the system instance, and $XDG_RUNTIME_DIR for the
2348 * journalctl learnt a new parameter --output-fields= for limiting the
2349 set of journal fields to output in verbose and JSON output modes.
2351 * systemd-timesyncd's configuration file gained a new option
2352 RootDistanceMaxSec= for setting the maximum root distance of servers
2353 it'll use, as well as the new options PollIntervalMinSec= and
2354 PollIntervalMaxSec= to tweak the minimum and maximum poll interval.
2356 * bootctl gained a new command "list" for listing all available boot
2357 menu items on systems that follow the boot loader specification.
2359 * systemctl gained a new --dry-run switch that shows what would be done
2360 instead of doing it, and is currently supported by the shutdown and
2363 * ConditionSecurity= can now detect the TOMOYO security module.
2365 * Unit file [Install] sections are now also respected in unit drop-in
2366 files. This is intended to be used by drop-ins under /usr/lib/.
2368 * systemd-firstboot may now also set the initial keyboard mapping.
2370 * Udev "changed" events for devices which are exposed as systemd
2371 .device units are now propagated to units specified in
2372 ReloadPropagatedFrom= as reload requests.
2374 * If a udev device has a SYSTEMD_WANTS= property containing a systemd
2375 unit template name (i.e. a name in the form of 'foobar@.service',
2376 without the instance component between the '@' and - the '.'), then
2377 the escaped sysfs path of the device is automatically used as the
2380 * SystemCallFilter= in unit files has been extended so that an "errno"
2381 can be specified individually for each system call. Example:
2382 SystemCallFilter=~uname:EILSEQ.
2384 * The cgroup delegation logic has been substantially updated. Delegate=
2385 now optionally takes a list of controllers (instead of a boolean, as
2386 before), which lists the controllers to delegate at least.
2388 * The networkd DHCPv6 client now implements the FQDN option (RFC 4704).
2390 * A new LogLevelMax= setting configures the maximum log level any
2391 process of the service may log at (i.e. anything with a lesser
2392 priority than what is specified is automatically dropped). A new
2393 LogExtraFields= setting allows configuration of additional journal
2394 fields to attach to all log records generated by any of the unit's
2397 * New StandardInputData= and StandardInputText= settings along with the
2398 new option StandardInput=data may be used to configure textual or
2399 binary data that shall be passed to the executed service process via
2400 standard input, encoded in-line in the unit file.
2402 * StandardInput=, StandardOutput= and StandardError= may now be used to
2403 connect stdin/stdout/stderr of executed processes directly with a
2404 file or AF_UNIX socket in the file system, using the new "file:" option.
2406 * A new unit file option CollectMode= has been added, that allows
2407 tweaking the garbage collection logic for units. It may be used to
2408 tell systemd to garbage collect units that have failed automatically
2409 (normally it only GCs units that exited successfully). systemd-run
2410 and systemd-mount expose this new functionality with a new -G option.
2412 * "machinectl bind" may now be used to bind mount non-directories
2413 (i.e. regularfiles, devices, fifos, sockets).
2415 * systemd-analyze gained a new verb "calendar" for validating and
2416 testing calendar time specifications to use for OnCalendar= in timer
2417 units. Besides validating the expression it will calculate the next
2418 time the specified expression would elapse.
2420 * In addition to the pre-existing FailureAction= unit file setting
2421 there's now SuccessAction=, for configuring a shutdown action to
2422 execute when a unit completes successfully. This is useful in
2423 particular inside containers that shall terminate after some workload
2424 has been completed. Also, both options are now supported for all unit
2425 types, not just services.
2427 * networkds's IP rule support gained two new options
2428 IncomingInterface= and OutgoingInterface= for configuring the incoming
2429 and outgoing interfaces of configured rules. systemd-networkd also
2430 gained support for "vxcan" network devices.
2432 * networkd gained a new setting RequiredForOnline=, taking a
2433 boolean. If set, systemd-wait-online will take it into consideration
2434 when determining that the system is up, otherwise it will ignore the
2435 interface for this purpose.
2437 * The sd_notify() protocol gained support for a new operation: with
2438 FDSTOREREMOVE=1 file descriptors may be removed from the per-service
2439 store again, ahead of POLLHUP or POLLERR when they are removed
2442 * A new document doc/UIDS-GIDS.md has been added to the source tree,
2443 that documents the UID/GID range and assignment assumptions and
2444 requirements of systemd.
2446 * The watchdog device PID 1 will ping may now be configured through the
2447 WatchdogDevice= configuration file setting, or by setting the
2448 systemd.watchdog_service= kernel commandline option.
2450 * systemd-resolved's gained support for registering DNS-SD services on
2451 the local network using MulticastDNS. Services may either be
2452 registered by dropping in a .dnssd file in /etc/systemd/dnssd/ (or
2453 the same dir below /run, /usr/lib), or through its D-Bus API.
2455 * The sd_notify() protocol can now with EXTEND_TIMEOUT_USEC=microsecond
2456 extend the effective start, runtime, and stop time. The service must
2457 continue to send EXTEND_TIMEOUT_USEC within the period specified to
2458 prevent the service manager from making the service as timedout.
2460 * systemd-resolved's DNSSEC support gained support for RFC 8080
2461 (Ed25519 keys and signatures).
2463 * The systemd-resolve command line tool gained a new set of options
2464 --set-dns=, --set-domain=, --set-llmnr=, --set-mdns=, --set-dnssec=,
2465 --set-nta= and --revert to configure per-interface DNS configuration
2466 dynamically during runtime. It's useful for pushing DNS information
2467 into systemd-resolved from DNS hook scripts that various interface
2468 managing software supports (such as pppd).
2470 * systemd-nspawn gained a new --network-namespace-path= command line
2471 option, which may be used to make a container join an existing
2472 network namespace, by specifying a path to a "netns" file.
2474 Contributions from: Alan Jenkins, Alan Robertson, Alessandro Ghedini,
2475 Andrew Jeddeloh, Antonio Rojas, Ari, asavah, bleep_blop, Carsten
2476 Strotmann, Christian Brauner, Christian Hesse, Clinton Roy, Collin
2477 Eggert, Cong Wang, Daniel Black, Daniel Lockyer, Daniel Rusek, Dimitri
2478 John Ledkov, Dmitry Rozhkov, Dongsu Park, Edward A. James, Evgeny
2479 Vereshchagin, Florian Klink, Franck Bui, Gwendal Grignou, Hans de
2480 Goede, Harald Hoyer, Hristo Venev, Iago López Galeiras, Ikey Doherty,
2481 Jakub Wilk, Jérémy Rosen, Jiahui Xie, John Lin, José Bollo, Josef
2482 Andersson, juga0, Krzysztof Nowicki, Kyle Walker, Lars Karlitski, Lars
2483 Kellogg-Stedman, Lauri Tirkkonen, Lennart Poettering, Lubomir Rintel,
2484 Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn, Lukáš Říha, Lukasz
2485 Rubaszewski, Maciej S. Szmigiero, Mantas Mikulėnas, Marcus Folkesson,
2486 Martin Steuer, Mathieu Trudel-Lapierre, Matija Skala,
2487 Matthias-Christian Ott, Max Resch, Michael Biebl, Michael Vogt, Michal
2488 Koutný, Michal Sekletar, Mike Gilbert, Muhammet Kara, Neil Brown, Olaf
2489 Hering, Ondrej Kozina, Patrik Flykt, Patryk Kocielnik, Peter Hutterer,
2490 Piotr Drąg, Razvan Cojocaru, Robin McCorkell, Roland Hieber, Saran
2491 Tunyasuvunakool, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
2492 Arlott, Simon Peeters, Stanislav Angelovič, Stefan Agner, Susant
2493 Sahani, Sylvain Plantefève, Thomas Blume, Thomas Haller, Tiago Salem
2494 Herrmann, Tinu Weber, Tom Stellard, Topi Miettinen, Torsten Hilbrich,
2495 Vito Caputo, Vladislav Vishnyakov, WaLyong Cho, Yu Watanabe, Zbigniew
2496 Jędrzejewski-Szmek, Zeal Jagannatha
2498 — Berlin, 2017-12-14
2502 * INCOMPATIBILITY: systemd-logind.service and other long-running
2503 services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP
2504 communication with the outside. This generally improves security of
2505 the system, and is in almost all cases a safe and good choice, as
2506 these services do not and should not provide any network-facing
2507 functionality. However, systemd-logind uses the glibc NSS API to
2508 query the user database. This creates problems on systems where NSS
2509 is set up to directly consult network services for user database
2510 lookups. In particular, this creates incompatibilities with the
2511 "nss-nis" module, which attempts to directly contact the NIS/YP
2512 network servers it is configured for, and will now consistently
2513 fail. In such cases, it is possible to turn off IP sandboxing for
2514 systemd-logind.service (set IPAddressDeny= in its [Service] section
2515 to the empty string, via a .d/ unit file drop-in). Downstream
2516 distributions might want to update their nss-nis packaging to include
2517 such a drop-in snippet, accordingly, to hide this incompatibility
2518 from the user. Another option is to make use of glibc's nscd service
2519 to proxy such network requests through a privilege-separated, minimal
2520 local caching daemon, or to switch to more modern technologies such
2521 sssd, whose NSS hook-ups generally do not involve direct network
2522 access. In general, we think it's definitely time to question the
2523 implementation choices of nss-nis, i.e. whether it's a good idea
2524 today to embed a network-facing loadable module into all local
2525 processes that need to query the user database, including the most
2526 trivial and benign ones, such as "ls". For more details about
2527 IPAddressDeny= see below.
2529 * A new modprobe.d drop-in is now shipped by default that sets the
2530 bonding module option max_bonds=0. This overrides the kernel default,
2531 to avoid conflicts and ambiguity as to whether or not bond0 should be
2532 managed by systemd-networkd or not. This resolves multiple issues
2533 with bond0 properties not being applied, when bond0 is configured
2534 with systemd-networkd. Distributors may choose to not package this,
2535 however in that case users will be prevented from correctly managing
2536 bond0 interface using systemd-networkd.
2538 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
2539 which print the logging level and target of the system manager. They
2540 complement the existing "set-log-level" and "set-log-target" verbs
2541 used to change those values.
2543 * journald.conf gained a new boolean setting ReadKMsg= which defaults
2544 to on. If turned off kernel log messages will not be read by
2545 systemd-journald or included in the logs. It also gained a new
2546 setting LineMax= for configuring the maximum line length in
2547 STDOUT/STDERR log streams. The new default for this value is 48K, up
2548 from the previous hardcoded 2048.
2550 * A new unit setting RuntimeDirectoryPreserve= has been added, which
2551 allows more detailed control of what to do with a runtime directory
2552 configured with RuntimeDirectory= (i.e. a directory below /run or
2553 $XDG_RUNTIME_DIR) after a unit is stopped.
2555 * The RuntimeDirectory= setting for units gained support for creating
2556 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
2557 one top-level directory.
2559 * Units gained new options StateDirectory=, CacheDirectory=,
2560 LogsDirectory= and ConfigurationDirectory= which are closely related
2561 to RuntimeDirectory= but manage per-service directories below
2562 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
2563 possible to write unit files which when activated automatically gain
2564 properly owned service specific directories in these locations, thus
2565 making unit files self-contained and increasing compatibility with
2566 stateless systems and factory reset where /etc or /var are
2567 unpopulated at boot. Matching these new settings there's also
2568 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
2569 ConfigurationDirectoryMode= for configuring the access mode of these
2570 directories. These settings are particularly useful in combination
2571 with DynamicUser=yes as they provide secure, properly-owned,
2572 writable, and stateful locations for storage, excluded from the
2573 sandbox that such services live in otherwise.
2575 * Automake support has been removed from this release. systemd is now
2578 * systemd-journald will now aggressively cache client metadata during
2579 runtime, speeding up log write performance under pressure. This comes
2580 at a small price though: as much of the metadata is read
2581 asynchronously from /proc/ (and isn't implicitly attached to log
2582 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
2583 metadata stored alongside a log entry might be slightly
2584 out-of-date. Previously it could only be slightly newer than the log
2585 message. The time window is small however, and given that the kernel
2586 is unlikely to be improved anytime soon in this regard, this appears
2589 * nss-myhostname/systemd-resolved will now by default synthesize an
2590 A/AAAA resource record for the "_gateway" hostname, pointing to the
2591 current default IP gateway. Previously it did that for the "gateway"
2592 name, hampering adoption, as some distributions wanted to leave that
2593 host name open for local use. The old behaviour may still be
2594 requested at build time.
2596 * systemd-networkd's [Address] section in .network files gained a new
2597 Scope= setting for configuring the IP address scope. The [Network]
2598 section gained a new boolean setting ConfigureWithoutCarrier= that
2599 tells systemd-networkd to ignore link sensing when configuring the
2600 device. The [DHCP] section gained a new Anonymize= boolean option for
2601 turning on a number of options suggested in RFC 7844. A new
2602 [RoutingPolicyRule] section has been added for configuring the IP
2603 routing policy. The [Route] section has gained support for a new
2604 Type= setting which permits configuring
2605 blackhole/unreachable/prohibit routes.
2607 * The [VRF] section in .netdev files gained a new Table= setting for
2608 configuring the routing table to use. The [Tunnel] section gained a
2609 new Independent= boolean field for configuring tunnels independent of
2610 an underlying network interface. The [Bridge] section gained a new
2611 GroupForwardMask= option for configuration of propagation of link
2612 local frames between bridge ports.
2614 * The WakeOnLan= setting in .link files gained support for a number of
2615 new modes. A new TCP6SegmentationOffload= setting has been added for
2616 configuring TCP/IPv6 hardware segmentation offload.
2618 * The IPv6 RA sender implementation may now optionally send out RDNSS
2619 and RDNSSL records to supply DNS configuration to peers.
2621 * systemd-nspawn gained support for a new --system-call-filter= command
2622 line option for adding and removing entries in the default system
2623 call filter it applies. Moreover systemd-nspawn has been changed to
2624 implement a system call whitelist instead of a blacklist.
2626 * systemd-run gained support for a new --pipe command line option. If
2627 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
2628 are directly passed on to the activated transient service
2629 executable. This allows invoking arbitrary processes as systemd
2630 services (for example to take benefit of dependency management,
2631 accounting management, resource management or log management that is
2632 done automatically for services) — while still allowing them to be
2633 integrated in a classic UNIX shell pipeline.
2635 * When a service sends RELOAD=1 via sd_notify() and reload propagation
2636 using ReloadPropagationTo= is configured, a reload is now propagated
2637 to configured units. (Previously this was only done on explicitly
2638 requested reloads, using "systemctl reload" or an equivalent
2641 * For each service unit a restart counter is now kept: it is increased
2642 each time the service is restarted due to Restart=, and may be
2643 queried using "systemctl show -p NRestarts …".
2645 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
2646 @signal and @timer have been added, for usage with SystemCallFilter=
2647 in unit files and the new --system-call-filter= command line option
2648 of systemd-nspawn (see above).
2650 * ExecStart= lines in unit files gained two new modifiers: when a
2651 command line is prefixed with "!" the command will be executed as
2652 configured, except for the credentials applied by
2653 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
2654 "+", but does still apply namespacing options unlike "+". There's
2655 also "!!" now, which is mostly identical, but becomes a NOP on
2656 systems that support ambient capabilities. This is useful to write
2657 unit files that work with ambient capabilities where possible but
2658 automatically fall back to traditional privilege dropping mechanisms
2659 on systems where this is not supported.
2661 * ListenNetlink= settings in socket units now support RDMA netlink
2664 * A new unit file setting LockPersonality= has been added which permits
2665 locking down the chosen execution domain ("personality") of a service
2668 * A new special target "getty-pre.target" has been added, which is
2669 ordered before all text logins, and may be used to order services
2670 before textual logins acquire access to the console.
2672 * systemd will now attempt to load the virtio-rng.ko kernel module very
2673 early on if a VM environment supporting this is detected. This should
2674 improve entropy during early boot in virtualized environments.
2676 * A _netdev option is now supported in /etc/crypttab that operates in a
2677 similar way as the same option in /etc/fstab: it permits configuring
2678 encrypted devices that need to be ordered after the network is up.
2679 Following this logic, two new special targets
2680 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
2681 added that are to cryptsetup.target what remote-fs.target and
2682 remote-fs-pre.target are to local-fs.target.
2684 * Service units gained a new UnsetEnvironment= setting which permits
2685 unsetting specific environment variables for services that are
2686 normally passed to it (for example in order to mask out locale
2687 settings for specific services that can't deal with it).
2689 * Units acquired a new boolean option IPAccounting=. When turned on, IP
2690 traffic accounting (packet count as well as byte count) is done for
2691 the service, and shown as part of "systemctl status" or "systemd-run
2694 * Service units acquired two new options IPAddressAllow= and
2695 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
2696 for configuring a simple IP access control list for all sockets of
2697 the unit. These options are available also on .slice and .socket
2698 units, permitting flexible access list configuration for individual
2699 services as well as groups of services (as defined by a slice unit),
2700 including system-wide. Note that IP ACLs configured this way are
2701 enforced on every single IPv4 and IPv6 socket created by any process
2702 of the service unit, and apply to ingress as well as egress traffic.
2704 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
2705 structured log message is generated each time the unit is stopped,
2706 containing information about the consumed resources of this
2709 * A new setting KeyringMode= has been added to unit files, which may be
2710 used to control how the kernel keyring is set up for executed
2713 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
2714 "systemctl kexec" and "systemctl exit" are now always asynchronous in
2715 behaviour (that is: these commands return immediately after the
2716 operation was enqueued instead of waiting for the operation to
2717 complete). Previously, "systemctl poweroff" and "systemctl reboot"
2718 were asynchronous on systems using systemd-logind (i.e. almost
2719 always, and like they were on sysvinit), and the other three commands
2720 were unconditionally synchronous. With this release this is cleaned
2721 up, and callers will see the same asynchronous behaviour on all
2722 systems for all five operations.
2724 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
2727 * .timer units now accept calendar specifications in other timezones
2728 than UTC or the local timezone.
2730 * The tmpfiles snippet var.conf has been changed to create
2731 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
2732 the "utmp" group already, and it appears to be generally understood
2733 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
2734 databases. Previously this was implemented correctly for all these
2735 databases excepts btmp, which has been opened up like this now
2736 too. Note that while the other databases are world-readable
2737 (i.e. 0644), btmp is not and remains more restrictive.
2739 * The systemd-resolve tool gained a new --reset-server-features
2740 switch. When invoked like this systemd-resolved will forget
2741 everything it learnt about the features supported by the configured
2742 upstream DNS servers, and restarts the feature probing logic on the
2743 next resolver look-up for them at the highest feature level
2746 * The status dump systemd-resolved sends to the logs upon receiving
2747 SIGUSR1 now also includes information about all DNS servers it is
2748 configured to use, and the features levels it probed for them.
2750 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
2751 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
2752 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
2753 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
2754 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
2755 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
2756 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
2757 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
2758 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
2759 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
2760 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
2761 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
2762 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
2763 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
2764 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
2765 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
2766 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
2767 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
2768 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
2769 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
2771 — Berlin, 2017-10-06
2775 * Meson is now supported as build system in addition to Automake. It is
2776 our plan to remove Automake in one of our next releases, so that
2777 Meson becomes our exclusive build system. Hence, please start using
2778 the Meson build system in your downstream packaging. There's plenty
2779 of documentation around how to use Meson, the extremely brief
2782 ./autogen.sh && ./configure && make && sudo make install
2786 meson build && ninja -C build && sudo ninja -C build install
2788 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
2789 which permits configuring a timeout on the time a job is
2790 running. This is particularly useful for setting timeouts on jobs for
2793 * Unit files gained two new options ConditionUser= and ConditionGroup=
2794 for conditionalizing units based on the identity of the user/group
2795 running a systemd user instance.
2797 * systemd-networkd now understands a new FlowLabel= setting in the
2798 [VXLAN] section of .network files, as well as a Priority= in
2799 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
2800 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
2801 gained support for configuration of GENEVE links, and IPv6 address
2802 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
2804 * .link files now understand a new Port= setting.
2806 * systemd-networkd's DHCP support gained support for DHCP option 119
2807 (domain search list).
2809 * systemd-networkd gained support for serving IPv6 address ranges using
2810 the Router Advertisement protocol. The new .network configuration
2811 section [IPv6Prefix] may be used to configure the ranges to
2812 serve. This is implemented based on a new, minimal, native server
2813 implementation of RA.
2815 * journalctl's --output= switch gained support for a new parameter
2816 "short-iso-precise" for a mode where timestamps are shown as precise
2819 * systemd-udevd's "net_id" builtin may now generate stable network
2820 interface names from IBM PowerVM VIO devices as well as ACPI platform
2823 * MulticastDNS support in systemd-resolved may now be explicitly
2824 enabled/disabled using the new MulticastDNS= configuration file
2827 * systemd-resolved may now optionally use libidn2 instead of the libidn
2828 for processing internationalized domain names. Support for libidn2
2829 should be considered experimental and should not be enabled by
2832 * "machinectl pull-tar" and related call may now do verification of
2833 downloaded images using SUSE-style .sha256 checksum files in addition
2834 to the already existing support for validating using Ubuntu-style
2837 * sd-bus gained support for a new sd_bus_message_appendv() call which
2838 is va_list equivalent of sd_bus_message_append().
2840 * sd-boot gained support for validating images using SHIM/MOK.
2842 * The SMACK code learnt support for "onlycap".
2844 * systemd-mount --umount is now much smarter in figuring out how to
2845 properly unmount a device given its mount or device path.
2847 * The code to call libnss_dns as a fallback from libnss_resolve when
2848 the communication with systemd-resolved fails was removed. This
2849 fallback was redundant and interfered with the [!UNAVAIL=return]
2850 suffix. See nss-resolve(8) for the recommended configuration.
2852 * systemd-logind may now be restarted without losing state. It stores
2853 the file descriptors for devices it manages in the system manager
2854 using the FDSTORE= mechanism. Please note that further changes in
2855 other components may be required to make use of this (for example
2856 Xorg has code to listen for stops of systemd-logind and terminate
2857 itself when logind is stopped or restarted, in order to avoid using
2858 stale file descriptors for graphical devices, which is now
2859 counterproductive and must be reverted in order for restarts of
2860 systemd-logind to be safe. See
2861 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
2863 * All kernel-install plugins are called with the environment variable
2864 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
2865 /etc/machine-id. If the machine ID could not be determined,
2866 $KERNEL_INSTALL_MACHINE_ID will be empty. Plugins should not put
2867 anything in the entry directory (passed as the second argument) if
2868 $KERNEL_INSTALL_MACHINE_ID is empty. For backwards compatibility, a
2869 temporary directory is passed as the entry directory and removed
2870 after all the plugins exit.
2872 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
2873 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
2874 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
2875 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
2876 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
2877 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
2878 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
2879 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
2880 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
2881 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
2882 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
2883 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
2884 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
2885 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
2886 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
2887 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
2888 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
2889 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
2890 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
2891 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
2892 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
2893 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
2894 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
2895 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
2896 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
2897 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
2898 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
2899 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
2900 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
2903 — Berlin, 2017-07-12
2907 * The "hybrid" control group mode has been modified to improve
2908 compatibility with "legacy" cgroups-v1 setups. Specifically, the
2909 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
2910 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
2911 cgroups-v1 hierarchy), the only externally visible change being that
2912 the cgroups-v2 hierarchy is also mounted, to
2913 /sys/fs/cgroup/unified. This should provide a large degree of
2914 compatibility with "legacy" cgroups-v1, while taking benefit of the
2915 better management capabilities of cgroups-v2.
2917 * The default control group setup mode may be selected both a boot-time
2918 via a set of kernel command line parameters (specifically:
2919 systemd.unified_cgroup_hierarchy= and
2920 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
2921 default selected on the configure command line
2922 (--with-default-hierarchy=). The upstream default is "hybrid"
2923 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
2924 this will change in a future systemd version to be "unified" (pure
2925 cgroups-v2 mode). The third option for the compile time option is
2926 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
2927 distributions to default to "hybrid" mode for release distributions,
2928 starting with v233. We recommend "unified" for development
2929 distributions (specifically: distributions such as Fedora's rawhide)
2930 as that's where things are headed in the long run. Use "legacy" for
2931 greatest stability and compatibility only.
2933 * Note one current limitation of "unified" and "hybrid" control group
2934 setup modes: the kernel currently does not permit the systemd --user
2935 instance (i.e. unprivileged code) to migrate processes between two
2936 disconnected cgroup subtrees, even if both are managed and owned by
2937 the user. This effectively means "systemd-run --user --scope" doesn't
2938 work when invoked from outside of any "systemd --user" service or
2939 scope. Specifically, it is not supported from session scopes. We are
2940 working on fixing this in a future systemd version. (See #3388 for
2941 further details about this.)
2943 * DBus policy files are now installed into /usr rather than /etc. Make
2944 sure your system has dbus >= 1.9.18 running before upgrading to this
2945 version, or override the install path with --with-dbuspolicydir= .
2947 * All python scripts shipped with systemd (specifically: the various
2948 tests written in Python) now require Python 3.
2950 * systemd unit tests can now run standalone (without the source or
2951 build directories), and can be installed into /usr/lib/systemd/tests/
2952 with 'make install-tests'.
2954 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
2955 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
2958 * Support for the %c, %r, %R specifiers in unit files has been
2959 removed. Specifiers are not supposed to be dependent on configuration
2960 in the unit file itself (so that they resolve the same regardless
2961 where used in the unit files), but these specifiers were influenced
2962 by the Slice= option.
2964 * The shell invoked by debug-shell.service now defaults to /bin/sh in
2965 all cases. If distributions want to use a different shell for this
2966 purpose (for example Fedora's /sbin/sushell) they need to specify
2967 this explicitly at configure time using --with-debug-shell=.
2969 * The confirmation spawn prompt has been reworked to offer the
2972 (c)ontinue, proceed without asking anymore
2973 (D)ump, show the state of the unit
2974 (f)ail, don't execute the command and pretend it failed
2976 (i)nfo, show a short summary of the unit
2977 (j)obs, show jobs that are in progress
2978 (s)kip, don't execute the command and pretend it succeeded
2979 (y)es, execute the command
2981 The 'n' choice for the confirmation spawn prompt has been removed,
2982 because its meaning was confusing.
2984 The prompt may now also be redirected to an alternative console by
2985 specifying the console as parameter to systemd.confirm_spawn=.
2987 * Services of Type=notify require a READY=1 notification to be sent
2988 during startup. If no such message is sent, the service now fails,
2989 even if the main process exited with a successful exit code.
2991 * Services that fail to start up correctly now always have their
2992 ExecStopPost= commands executed. Previously, they'd enter "failed"
2993 state directly, without executing these commands.
2995 * The option MulticastDNS= of network configuration files has acquired
2996 an actual implementation. With MulticastDNS=yes a host can resolve
2997 names of remote hosts and reply to mDNS A and AAAA requests.
2999 * When units are about to be started an additional check is now done to
3000 ensure that all dependencies of type BindsTo= (when used in
3001 combination with After=) have been started.
3003 * systemd-analyze gained a new verb "syscall-filter" which shows which
3004 system call groups are defined for the SystemCallFilter= unit file
3005 setting, and which system calls they contain.
3007 * A new system call filter group "@filesystem" has been added,
3008 consisting of various file system related system calls. Group
3009 "@reboot" has been added, covering reboot, kexec and shutdown related
3010 calls. Finally, group "@swap" has been added covering swap
3011 configuration related calls.
3013 * A new unit file option RestrictNamespaces= has been added that may be
3014 used to restrict access to the various process namespace types the
3015 Linux kernel provides. Specifically, it may be used to take away the
3016 right for a service unit to create additional file system, network,
3017 user, and other namespaces. This sandboxing option is particularly
3018 relevant due to the high amount of recently discovered namespacing
3019 related vulnerabilities in the kernel.
3021 * systemd-udev's .link files gained support for a new AutoNegotiation=
3022 setting for configuring Ethernet auto-negotiation.
3024 * systemd-networkd's .network files gained support for a new
3025 ListenPort= setting in the [DHCP] section to explicitly configure the
3026 UDP client port the DHCP client shall listen on.
3028 * .network files gained a new Unmanaged= boolean setting for explicitly
3029 excluding one or more interfaces from management by systemd-networkd.
3031 * The systemd-networkd ProxyARP= option has been renamed to
3032 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
3033 renamed to ReduceARPProxy=. The old names continue to be available
3036 * systemd-networkd gained support for configuring IPv6 Proxy NDP
3037 addresses via the new IPv6ProxyNDPAddress= .network file setting.
3039 * systemd-networkd's bonding device support gained support for two new
3040 configuration options ActiveSlave= and PrimarySlave=.
3042 * The various options in the [Match] section of .network files gained
3043 support for negative matching.
3045 * New systemd-specific mount options are now understood in /etc/fstab:
3047 x-systemd.mount-timeout= may be used to configure the maximum
3048 permitted runtime of the mount command.
3050 x-systemd.device-bound may be set to bind a mount point to its
3051 backing device unit, in order to automatically remove a mount point
3052 if its backing device is unplugged. This option may also be
3053 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
3054 on the block device, which is now automatically set for all CDROM
3055 drives, so that mounted CDs are automatically unmounted when they are
3056 removed from the drive.
3058 x-systemd.after= and x-systemd.before= may be used to explicitly
3059 order a mount after or before another unit or mount point.
3061 * Enqueued start jobs for device units are now automatically garbage
3062 collected if there are no jobs waiting for them anymore.
3064 * systemctl list-jobs gained two new switches: with --after, for every
3065 queued job the jobs it's waiting for are shown; with --before the
3066 jobs which it's blocking are shown.
3068 * systemd-nspawn gained support for ephemeral boots from disk images
3069 (or in other words: --ephemeral and --image= may now be
3070 combined). Moreover, ephemeral boots are now supported for normal
3071 directories, even if the backing file system is not btrfs. Of course,
3072 if the file system does not support file system snapshots or
3073 reflinks, the initial copy operation will be relatively expensive, but
3074 this should still be suitable for many use cases.
3076 * Calendar time specifications in .timer units now support
3077 specifications relative to the end of a month by using "~" instead of
3078 "-" as separator between month and day. For example, "*-02~03" means
3079 "the third last day in February". In addition a new syntax for
3080 repeated events has been added using the "/" character. For example,
3081 "9..17/2:00" means "every two hours from 9am to 5pm".
3083 * systemd-socket-proxyd gained a new parameter --connections-max= for
3084 configuring the maximum number of concurrent connections.
3086 * sd-id128 gained a new API for generating unique IDs for the host in a
3087 way that does not leak the machine ID. Specifically,
3088 sd_id128_get_machine_app_specific() derives an ID based on the
3089 machine ID a in well-defined, non-reversible, stable way. This is
3090 useful whenever an identifier for the host is needed but where the
3091 identifier shall not be useful to identify the system beyond the
3092 scope of the application itself. (Internally this uses HMAC-SHA256 as
3093 keyed hash function using the machine ID as input.)
3095 * NotifyAccess= gained a new supported value "exec". When set
3096 notifications are accepted from all processes systemd itself invoked,
3097 including all control processes.
3099 * .nspawn files gained support for defining overlay mounts using the
3100 Overlay= and OverlayReadOnly= options. Previously this functionality
3101 was only available on the systemd-nspawn command line.
3103 * systemd-nspawn's --bind= and --overlay= options gained support for
3104 bind/overlay mounts whose source lies within the container tree by
3105 prefixing the source path with "+".
3107 * systemd-nspawn's --bind= and --overlay= options gained support for
3108 automatically allocating a temporary source directory in /var/tmp
3109 that is removed when the container dies. Specifically, if the source
3110 directory is specified as empty string this mechanism is selected. An
3111 example usage is --overlay=+/var::/var, which creates an overlay
3112 mount based on the original /var contained in the image, overlayed
3113 with a temporary directory in the host's /var/tmp. This way changes
3114 to /var are automatically flushed when the container shuts down.
3116 * systemd-nspawn --image= option does now permit raw file system block
3117 devices (in addition to images containing partition tables, as
3120 * The disk image dissection logic in systemd-nspawn gained support for
3121 automatically setting up LUKS encrypted as well as Verity protected
3122 partitions. When a container is booted from an encrypted image the
3123 passphrase is queried at start-up time. When a container with Verity
3124 data is started, the root hash is search in a ".roothash" file
3125 accompanying the disk image (alternatively, pass the root hash via
3126 the new --root-hash= command line option).
3128 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
3129 be used to dissect disk images the same way as systemd-nspawn does
3130 it, following the Bootable Partition Specification. It may even be
3131 used to mount disk images with complex partition setups (including
3132 LUKS and Verity partitions) to a local host directory, in order to
3133 inspect them. This tool is not considered public API (yet), and is
3134 thus not installed into /usr/bin. Please do not rely on its
3135 existence, since it might go away or be changed in later systemd
3138 * A new generator "systemd-verity-generator" has been added, similar in
3139 style to "systemd-cryptsetup-generator", permitting automatic setup of
3140 Verity root partitions when systemd boots up. In order to make use of
3141 this your partition setup should follow the Discoverable Partitions
3142 Specification, and the GPT partition ID of the root file system
3143 partition should be identical to the upper 128bit of the Verity root
3144 hash. The GPT partition ID of the Verity partition protecting it
3145 should be the lower 128bit of the Verity root hash. If the partition
3146 image follows this model it is sufficient to specify a single
3147 "roothash=" kernel command line argument to both configure which root
3148 image and verity partition to use as well as the root hash for
3149 it. Note that systemd-nspawn's Verity support follows the same
3150 semantics, meaning that disk images with proper Verity data in place
3151 may be booted in containers with systemd-nspawn as well as on
3152 physical systems via the verity generator. Also note that the "mkosi"
3153 tool available at https://github.com/systemd/mkosi has been updated
3154 to generate Verity protected disk images following this scheme. In
3155 fact, it has been updated to generate disk images that optionally
3156 implement a complete UEFI SecureBoot trust chain, involving a signed
3157 kernel and initrd image that incorporates such a root hash as well as
3158 a Verity-enabled root partition.
3160 * The hardware database (hwdb) udev supports has been updated to carry
3161 accelerometer quirks.
3163 * All system services are now run with a fresh kernel keyring set up
3164 for them. The invocation ID is stored by default in it, thus
3165 providing a safe, non-overridable way to determine the invocation
3168 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
3169 options for bind mounting arbitrary paths in a service-specific
3170 way. When these options are used, arbitrary host or service files and
3171 directories may be mounted to arbitrary locations in the service's
3174 * Documentation has been added that lists all of systemd's low-level
3175 environment variables:
3177 https://github.com/systemd/systemd/blob/master/docs/ENVIRONMENT.md
3179 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
3180 whether a specific socket file descriptor matches a specified socket
3183 * systemd-firstboot has been updated to check for the
3184 systemd.firstboot= kernel command line option. It accepts a boolean
3185 and when set to false the first boot questions are skipped.
3187 * systemd-fstab-generator has been updated to check for the
3188 systemd.volatile= kernel command line option, which either takes an
3189 optional boolean parameter or the special value "state". If used the
3190 system may be booted in a "volatile" boot mode. Specifically,
3191 "systemd.volatile" is used, the root directory will be mounted as
3192 tmpfs, and only /usr is mounted from the actual root file system. If
3193 "systemd.volatile=state" is used, the root directory will be mounted
3194 as usual, but /var is mounted as tmpfs. This concept provides similar
3195 functionality as systemd-nspawn's --volatile= option, but provides it
3196 on physical boots. Use this option for implementing stateless
3197 systems, or testing systems with all state and/or configuration reset
3198 to the defaults. (Note though that many distributions are not
3199 prepared to boot up without a populated /etc or /var, though.)
3201 * systemd-gpt-auto-generator gained support for LUKS encrypted root
3202 partitions. Previously it only supported LUKS encrypted partitions
3203 for all other uses, except for the root partition itself.
3205 * Socket units gained support for listening on AF_VSOCK sockets for
3206 communication in virtualized QEMU environments.
3208 * The "configure" script gained a new option --with-fallback-hostname=
3209 for specifying the fallback hostname to use if none is configured in
3210 /etc/hostname. For example, by specifying
3211 --with-fallback-hostname=fedora it is possible to default to a
3212 hostname of "fedora" on pristine installations.
3214 * systemd-cgls gained support for a new --unit= switch for listing only
3215 the control groups of a specific unit. Similar --user-unit= has been
3216 added for listing only the control groups of a specific user unit.
3218 * systemd-mount gained a new --umount switch for unmounting a mount or
3219 automount point (and all mount/automount points below it).
3221 * systemd will now refuse full configuration reloads (via systemctl
3222 daemon-reload and related calls) unless at least 16MiB of free space
3223 are available in /run. This is a safety precaution in order to ensure
3224 that generators can safely operate after the reload completed.
3226 * A new unit file option RootImage= has been added, which has a similar
3227 effect as RootDirectory= but mounts the service's root directory from
3228 a disk image instead of plain directory. This logic reuses the same
3229 image dissection and mount logic that systemd-nspawn already uses,
3230 and hence supports any disk images systemd-nspawn supports, including
3231 those following the Discoverable Partition Specification, as well as
3232 Verity enabled images. This option enables systemd to run system
3233 services directly off disk images acting as resource bundles,
3234 possibly even including full integrity data.
3236 * A new MountAPIVFS= unit file option has been added, taking a boolean
3237 argument. If enabled /proc, /sys and /dev (collectively called the
3238 "API VFS") will be mounted for the service. This is only relevant if
3239 RootDirectory= or RootImage= is used for the service, as these mounts
3240 are of course in place in the host mount namespace anyway.
3242 * systemd-nspawn gained support for a new --pivot-root= switch. If
3243 specified the root directory within the container image is pivoted to
3244 the specified mount point, while the original root disk is moved to a
3245 different place. This option enables booting of ostree images
3246 directly with systemd-nspawn.
3248 * The systemd build scripts will no longer complain if the NTP server
3249 addresses are not changed from the defaults. Google now supports
3250 these NTP servers officially. We still recommend downstreams to
3251 properly register an NTP pool with the NTP pool project though.
3253 * coredumpctl gained a new "--reverse" option for printing the list
3254 of coredumps in reverse order.
3256 * coredumpctl will now show additional information about truncated and
3257 inaccessible coredumps, as well as coredumps that are still being
3258 processed. It also gained a new --quiet switch for suppressing
3259 additional informational message in its output.
3261 * coredumpctl gained support for only showing coredumps newer and/or
3262 older than specific timestamps, using the new --since= and --until=
3263 options, reminiscent of journalctl's options by the same name.
3265 * The systemd-coredump logic has been improved so that it may be reused
3266 to collect backtraces in non-compiled languages, for example in
3267 scripting languages such as Python.
3269 * machinectl will now show the UID shift of local containers, if user
3270 namespacing is enabled for them.
3272 * systemd will now optionally run "environment generator" binaries at
3273 configuration load time. They may be used to add environment
3274 variables to the environment block passed to services invoked. One
3275 user environment generator is shipped by default that sets up
3276 environment variables based on files dropped into /etc/environment.d
3277 and ~/.config/environment.d/.
3279 * systemd-resolved now includes the new, recently published 2017 DNSSEC
3282 * hostnamed has been updated to report a new chassis type of
3283 "convertible" to cover "foldable" laptops that can both act as a
3284 tablet and as a laptop, such as various Lenovo Yoga devices.
3286 Contributions from: Adrián López, Alexander Galanin, Alexander
3287 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
3288 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
3289 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
3290 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
3291 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
3292 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
3293 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
3294 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
3295 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
3296 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
3297 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
3298 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
3299 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
3300 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
3301 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
3302 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
3303 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
3304 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
3305 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
3306 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
3307 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
3308 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
3309 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
3310 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
3311 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
3312 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
3313 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
3316 — Berlin, 2017-03-01
3320 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
3321 RestrictAddressFamilies= enabled. These sandboxing options should
3322 generally be compatible with the various external udev call-out
3323 binaries we are aware of, however there may be exceptions, in
3324 particular when exotic languages for these call-outs are used. In
3325 this case, consider turning off these settings locally.
3327 * The new RemoveIPC= option can be used to remove IPC objects owned by
3328 the user or group of a service when that service exits.
3330 * The new ProtectKernelModules= option can be used to disable explicit
3331 load and unload operations of kernel modules by a service. In
3332 addition access to /usr/lib/modules is removed if this option is set.
3334 * ProtectSystem= option gained a new value "strict", which causes the
3335 whole file system tree with the exception of /dev, /proc, and /sys,
3336 to be remounted read-only for a service.
3338 * The new ProtectKernelTunables= option can be used to disable
3339 modification of configuration files in /sys and /proc by a service.
3340 Various directories and files are remounted read-only, so access is
3341 restricted even if the file permissions would allow it.
3343 * The new ProtectControlGroups= option can be used to disable write
3344 access by a service to /sys/fs/cgroup.
3346 * Various systemd services have been hardened with
3347 ProtectKernelTunables=yes, ProtectControlGroups=yes,
3348 RestrictAddressFamilies=.
3350 * Support for dynamically creating users for the lifetime of a service
3351 has been added. If DynamicUser=yes is specified, user and group IDs
3352 will be allocated from the range 61184..65519 for the lifetime of the
3353 service. They can be resolved using the new nss-systemd.so NSS
3354 module. The module must be enabled in /etc/nsswitch.conf. Services
3355 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
3356 any resources allocated by the service will be cleaned up when the
3357 service exits. They also have ProtectHome=read-only and
3358 ProtectSystem=strict enabled, so they are not able to make any
3359 permanent modifications to the system.
3361 * The nss-systemd module also always resolves root and nobody, making
3362 it possible to have no /etc/passwd or /etc/group files in minimal
3363 container or chroot environments.
3365 * Services may be started with their own user namespace using the new
3366 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
3367 under which the service is running are mapped. All other users are
3370 * Support for the cgroup namespace has been added to systemd-nspawn. If
3371 supported by kernel, the container system started by systemd-nspawn
3372 will have its own view of the cgroup hierarchy. This new behaviour
3373 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
3375 * The new MemorySwapMax= option can be used to limit the maximum swap
3376 usage under the unified cgroup hierarchy.
3378 * Support for the CPU controller in the unified cgroup hierarchy has
3379 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
3380 options. This controller requires out-of-tree patches for the kernel
3381 and the support is provisional.
3383 * Mount and automount units may now be created transiently
3384 (i.e. dynamically at runtime via the bus API, instead of requiring
3385 unit files in the file system).
3387 * systemd-mount is a new tool which may mount file systems – much like
3388 mount(8), optionally pulling in additional dependencies through
3389 transient .mount and .automount units. For example, this tool
3390 automatically runs fsck on a backing block device before mounting,
3391 and allows the automount logic to be used dynamically from the
3392 command line for establishing mount points. This tool is particularly
3393 useful when dealing with removable media, as it will ensure fsck is
3394 run – if necessary – before the first access and that the file system
3395 is quickly unmounted after each access by utilizing the automount
3396 logic. This maximizes the chance that the file system on the
3397 removable media stays in a clean state, and if it isn't in a clean
3398 state is fixed automatically.
3400 * LazyUnmount=yes option for mount units has been added to expose the
3401 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
3404 * /efi will be used as the mount point of the EFI boot partition, if
3405 the directory is present, and the mount point was not configured
3406 through other means (e.g. fstab). If /efi directory does not exist,
3407 /boot will be used as before. This makes it easier to automatically
3408 mount the EFI partition on systems where /boot is used for something
3411 * When operating on GPT disk images for containers, systemd-nspawn will
3412 now mount the ESP to /boot or /efi according to the same rules as PID
3413 1 running on a host. This allows tools like "bootctl" to operate
3414 correctly within such containers, in order to make container images
3415 bootable on physical systems.
3417 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
3419 * Two new user session targets have been added to support running
3420 graphical sessions under the systemd --user instance:
3421 graphical-session.target and graphical-session-pre.target. See
3422 systemd.special(7) for a description of how those targets should be
3425 * The vconsole initialization code has been significantly reworked to
3426 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
3427 support unicode keymaps. Font and keymap configuration will now be
3428 copied to all allocated virtual consoles.
3430 * FreeBSD's bhyve virtualization is now detected.
3432 * Information recorded in the journal for core dumps now includes the
3433 contents of /proc/mountinfo and the command line of the process at
3434 the top of the process hierarchy (which is usually the init process
3437 * systemd-journal-gatewayd learned the --directory= option to serve
3438 files from the specified location.
3440 * journalctl --root=… can be used to peruse the journal in the
3441 /var/log/ directories inside of a container tree. This is similar to
3442 the existing --machine= option, but does not require the container to
3445 * The hardware database has been extended to support
3446 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
3449 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
3450 specify the click rate for mice which include a horizontal wheel with
3451 a click rate that is different than the one for the vertical wheel.
3453 * systemd-run gained a new --wait option that makes service execution
3454 synchronous. (Specifically, the command will not return until the
3455 specified service binary exited.)
3457 * systemctl gained a new --wait option that causes the start command to
3458 wait until the units being started have terminated again.
3460 * A new journal output mode "short-full" has been added which displays
3461 timestamps with abbreviated English day names and adds a timezone
3462 suffix. Those timestamps include more information than the default
3463 "short" output mode, and can be passed directly to journalctl's
3464 --since= and --until= options.
3466 * /etc/resolv.conf will be bind-mounted into containers started by
3467 systemd-nspawn, if possible, so any changes to resolv.conf contents
3468 are automatically propagated to the container.
3470 * The number of instances for socket-activated services originating
3471 from a single IP address can be limited with
3472 MaxConnectionsPerSource=, extending the existing setting of
3475 * systemd-networkd gained support for vcan ("Virtual CAN") interface
3478 * .netdev and .network configuration can now be extended through
3481 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
3482 Segmentation Offload, Generic Receive Offload, Large Receive Offload
3483 can be enabled and disabled using the new UDPSegmentationOffload=,
3484 TCPSegmentationOffload=, GenericSegmentationOffload=,
3485 GenericReceiveOffload=, LargeReceiveOffload= options in the
3486 [Link] section of .link files.
3488 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
3489 Port VLAN ID can be configured for bridge devices using the new STP=,
3490 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
3491 section of .netdev files.
3493 * The route table to which routes received over DHCP or RA should be
3494 added can be configured with the new RouteTable= option in the [DHCP]
3495 and [IPv6AcceptRA] sections of .network files.
3497 * The Address Resolution Protocol can be disabled on links managed by
3498 systemd-networkd using the ARP=no setting in the [Link] section of
3501 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
3502 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
3503 encode information about the result and exit codes of the current
3504 service runtime cycle.
3506 * systemd-sysctl will now configure kernel parameters in the order
3507 they occur in the configuration files. This matches what sysctl
3508 has been traditionally doing.
3510 * kernel-install "plugins" that are executed to perform various
3511 tasks after a new kernel is added and before an old one is removed
3512 can now return a special value to terminate the procedure and
3513 prevent any later plugins from running.
3515 * Journald's SplitMode=login setting has been deprecated. It has been
3516 removed from documentation, and its use is discouraged. In a future
3517 release it will be completely removed, and made equivalent to current
3518 default of SplitMode=uid.
3520 * Storage=both option setting in /etc/systemd/coredump.conf has been
3521 removed. With fast LZ4 compression storing the core dump twice is not
3524 * The --share-system systemd-nspawn option has been replaced with an
3525 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
3526 this functionality is discouraged. In addition the variables
3527 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
3528 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
3529 individual namespaces.
3531 * "machinectl list" now shows the IP address of running containers in
3532 the output, as well as OS release information.
3534 * "loginctl list" now shows the TTY of each session in the output.
3536 * sd-bus gained new API calls sd_bus_track_set_recursive(),
3537 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
3538 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
3539 tracking objects in a "recursive" mode, where a single client can be
3540 counted multiple times, if it takes multiple references.
3542 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
3543 sd_bus_get_exit_on_disconnect(). They may be used to make a
3544 process using sd-bus automatically exit if the bus connection is
3547 * Bus clients of the service manager may now "pin" loaded units into
3548 memory, by taking an explicit reference on them. This is useful to
3549 ensure the client can retrieve runtime data about the service even
3550 after the service completed execution. Taking such a reference is
3551 available only for privileged clients and should be helpful to watch
3552 running services in a race-free manner, and in particular collect
3553 information about exit statuses and results.
3555 * The nss-resolve module has been changed to strictly return UNAVAIL
3556 when communication via D-Bus with resolved failed, and NOTFOUND when
3557 a lookup completed but was negative. This means it is now possible to
3558 neatly configure fallbacks using nsswitch.conf result checking
3559 expressions. Taking benefit of this, the new recommended
3560 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
3562 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
3564 * A new setting CtrlAltDelBurstAction= has been added to
3565 /etc/systemd/system.conf which may be used to configure the precise
3566 behaviour if the user on the console presses Ctrl-Alt-Del more often
3567 than 7 times in 2s. Previously this would unconditionally result in
3568 an expedited, immediate reboot. With this new setting the precise
3569 operation may be configured in more detail, and also turned off
3572 * In .netdev files two new settings RemoteChecksumTx= and
3573 RemoteChecksumRx= are now understood that permit configuring the
3574 remote checksumming logic for VXLAN networks.
3576 * The service manager learnt a new "invocation ID" concept for invoked
3577 services. Each runtime cycle of a service will get a new invocation
3578 ID (a 128bit random UUID) assigned that identifies the current
3579 run of the service uniquely and globally. A new invocation ID
3580 is generated each time a service starts up. The journal will store
3581 the invocation ID of a service along with any logged messages, thus
3582 making the invocation ID useful for matching the online runtime of a
3583 service with the offline log data it generated in a safe way without
3584 relying on synchronized timestamps. In many ways this new service
3585 invocation ID concept is similar to the kernel's boot ID concept that
3586 uniquely and globally identifies the runtime of each boot. The
3587 invocation ID of a service is passed to the service itself via an
3588 environment variable ($INVOCATION_ID). A new bus call
3589 GetUnitByInvocationID() has been added that is similar to GetUnit()
3590 but instead of retrieving the bus path for a unit by its name
3591 retrieves it by its invocation ID. The returned path is valid only as
3592 long as the passed invocation ID is current.
3594 * systemd-resolved gained a new "DNSStubListener" setting in
3595 resolved.conf. It either takes a boolean value or the special values
3596 "udp" and "tcp", and configures whether to enable the stub DNS
3597 listener on 127.0.0.53:53.
3599 * IP addresses configured via networkd may now carry additional
3600 configuration settings supported by the kernel. New options include:
3601 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
3602 PrefixRoute=, AutoJoin=.
3604 * The PAM configuration fragment file for "user@.service" shipped with
3605 systemd (i.e. the --user instance of systemd) has been stripped to
3606 the minimum necessary to make the system boot. Previously, it
3607 contained Fedora-specific stanzas that did not apply to other
3608 distributions. It is expected that downstream distributions add
3609 additional configuration lines, matching their needs to this file,
3610 using it only as rough template of what systemd itself needs. Note
3611 that this reduced fragment does not even include an invocation of
3612 pam_limits which most distributions probably want to add, even though
3613 systemd itself does not need it. (There's also the new build time
3614 option --with-pamconfdir=no to disable installation of the PAM
3617 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
3618 capability is now also dropped from its set (in addition to
3619 CAP_SYS_MKNOD as before).
3621 * In service unit files it is now possible to connect a specific named
3622 file descriptor with stdin/stdout/stdout of an executed service. The
3623 name may be specified in matching .socket units using the
3624 FileDescriptorName= setting.
3626 * A number of journal settings may now be configured on the kernel
3627 command line. Specifically, the following options are now understood:
3628 systemd.journald.max_level_console=,
3629 systemd.journald.max_level_store=,
3630 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
3631 systemd.journald.max_level_wall=.
3633 * "systemctl is-enabled --full" will now show by which symlinks a unit
3634 file is enabled in the unit dependency tree.
3636 * Support for VeraCrypt encrypted partitions has been added to the
3637 "cryptsetup" logic and /etc/crypttab.
3639 * systemd-detect-virt gained support for a new --private-users switch
3640 that checks whether the invoking processes are running inside a user
3641 namespace. Similar, a new special value "private-users" for the
3642 existing ConditionVirtualization= setting has been added, permitting
3643 skipping of specific units in user namespace environments.
3645 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
3646 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
3647 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
3648 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
3649 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
3650 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
3651 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
3652 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
3653 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
3654 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
3655 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
3656 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
3657 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
3658 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
3659 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
3660 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
3661 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
3662 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
3663 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
3664 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
3665 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
3666 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
3667 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
3668 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
3669 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
3670 Jędrzejewski-Szmek, Zeal Jagannatha
3672 — Santa Fe, 2016-11-03
3676 * In service units the various ExecXYZ= settings have been extended
3677 with an additional special character as first argument of the
3678 assigned value: if the character '+' is used the specified command
3679 line it will be run with full privileges, regardless of User=,
3680 Group=, CapabilityBoundingSet= and similar options. The effect is
3681 similar to the existing PermissionsStartOnly= option, but allows
3682 configuration of this concept for each executed command line
3685 * Services may now alter the service watchdog timeout at runtime by
3686 sending a WATCHDOG_USEC= message via sd_notify().
3688 * MemoryLimit= and related unit settings now optionally take percentage
3689 specifications. The percentage is taken relative to the amount of
3690 physical memory in the system (or in case of containers, the assigned
3691 amount of memory). This allows scaling service resources neatly with
3692 the amount of RAM available on the system. Similarly, systemd-logind's
3693 RuntimeDirectorySize= option now also optionally takes percentage
3696 * In similar fashion TasksMax= takes percentage values now, too. The
3697 value is taken relative to the configured maximum number of processes
3698 on the system. The per-service task maximum has been changed to 15%
3699 using this functionality. (Effectively this is an increase of 512 →
3700 4915 for service units, given the kernel's default pid_max setting.)
3702 * Calendar time specifications in .timer units now understand a ".."
3703 syntax for time ranges. Example: "4..7:10" may now be used for
3704 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
3707 * The InaccessableDirectories=, ReadOnlyDirectories= and
3708 ReadWriteDirectories= unit file settings have been renamed to
3709 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
3710 applied to all kinds of file nodes, and not just directories, with
3711 the exception of symlinks. Specifically these settings may now be
3712 used on block and character device nodes, UNIX sockets and FIFOS as
3713 well as regular files. The old names of these settings remain
3714 available for compatibility.
3716 * systemd will now log about all service processes it kills forcibly
3717 (using SIGKILL) because they remained after the clean shutdown phase
3718 of the service completed. This should help identifying services that
3719 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
3720 systemd-logind's configuration a similar log message is generated for
3721 processes killed at the end of each session due to this setting.
3723 * systemd will now set the $JOURNAL_STREAM environment variable for all
3724 services whose stdout/stderr are connected to the Journal (which
3725 effectively means by default: all services). The variable contains
3726 the device and inode number of the file descriptor used for
3727 stdout/stderr. This may be used by invoked programs to detect whether
3728 their stdout/stderr is connected to the Journal, in which case they
3729 can switch over to direct Journal communication, thus being able to
3730 pass extended, structured metadata along with their log messages. As
3731 one example, this is now used by glib's logging primitives.
3733 * When using systemd's default tmp.mount unit for /tmp, the mount point
3734 will now be established with the "nosuid" and "nodev" options. This
3735 avoids privilege escalation attacks that put traps and exploits into
3736 /tmp. However, this might cause problems if you e. g. put container
3737 images or overlays into /tmp; if you need this, override tmp.mount's
3738 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
3741 * systemd now supports the "memory" cgroup controller also on
3744 * The systemd-cgtop tool now optionally takes a control group path as
3745 command line argument. If specified, the control group list shown is
3746 limited to subgroups of that group.
3748 * The SystemCallFilter= unit file setting gained support for
3749 pre-defined, named system call filter sets. For example
3750 SystemCallFilter=@clock is now an effective way to make all clock
3751 changing-related system calls unavailable to a service. A number of
3752 similar pre-defined groups are defined. Writing system call filters
3753 for system services is simplified substantially with this new
3754 concept. Accordingly, all of systemd's own, long-running services now
3755 enable system call filtering based on this, by default.
3757 * A new service setting MemoryDenyWriteExecute= has been added, taking
3758 a boolean value. If turned on, a service may no longer create memory
3759 mappings that are writable and executable at the same time. This
3760 enhances security for services where this is enabled as it becomes
3761 harder to dynamically write and then execute memory in exploited
3762 service processes. This option has been enabled for all of systemd's
3763 own long-running services.
3765 * A new RestrictRealtime= service setting has been added, taking a
3766 boolean argument. If set the service's processes may no longer
3767 acquire realtime scheduling. This improves security as realtime
3768 scheduling may otherwise be used to easily freeze the system.
3770 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
3771 value. This may be used for requesting that the system manager inside
3772 of the container reports start-up completion to nspawn which then
3773 propagates this notification further to the service manager
3774 supervising nspawn itself. A related option NotifyReady= in .nspawn
3775 files has been added too. This functionality allows ordering of the
3776 start-up of multiple containers using the usual systemd ordering
3779 * machinectl gained a new command "stop" that is an alias for
3782 * systemd-resolved gained support for contacting DNS servers on
3783 link-local IPv6 addresses.
3785 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
3786 its caches. A method call for requesting the same operation has been
3787 added to the bus API too, and is made available via "systemd-resolve
3790 * systemd-resolve gained a new --status switch. If passed a brief
3791 summary of the used DNS configuration with per-interface information
3794 * resolved.conf gained a new Cache= boolean option, defaulting to
3795 on. If turned off local DNS caching is disabled. This comes with a
3796 performance penalty in particular when DNSSEC is enabled. Note that
3797 resolved disables its internal caching implicitly anyway, when the
3798 configured DNS server is on a host-local IP address such as ::1 or
3799 127.0.0.1, thus automatically avoiding double local caching.
3801 * systemd-resolved now listens on the local IP address 127.0.0.53:53
3802 for DNS requests. This improves compatibility with local programs
3803 that do not use the libc NSS or systemd-resolved's bus APIs for name
3804 resolution. This minimal DNS service is only available to local
3805 programs and does not implement the full DNS protocol, but enough to
3806 cover local DNS clients. A new, static resolv.conf file, listing just
3807 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
3808 now recommended to make /etc/resolv.conf a symlink to this file in
3809 order to route all DNS lookups to systemd-resolved, regardless if
3810 done via NSS, the bus API or raw DNS packets. Note that this local
3811 DNS service is not as fully featured as the libc NSS or
3812 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
3813 used to deliver link-local address information (as this implies
3814 sending a local interface index along), LLMNR/mDNS support via this
3815 interface is severely restricted. It is thus strongly recommended for
3816 all applications to use the libc NSS API or native systemd-resolved
3819 * systemd-networkd's bridge support learned a new setting
3820 VLANFiltering= for controlling VLAN filtering. Moreover a new section
3821 in .network files has been added for configuring VLAN bridging in
3822 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
3824 * systemd-networkd's IPv6 Router Advertisement code now makes use of
3825 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
3826 now be acquired without relying on DHCPv6. Two new options
3827 UseDomains= and UseDNS= have been added to configure this behaviour.
3829 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
3830 renamed IPv6AcceptRA=, without altering its behaviour. The old
3831 setting name remains available for compatibility reasons.
3833 * The systemd-networkd VTI/VTI6 tunneling support gained new options
3834 Key=, InputKey= and OutputKey=.
3836 * systemd-networkd gained support for VRF ("Virtual Routing Function")
3837 interface configuration.
3839 * "systemctl edit" may now be used to create new unit files by
3840 specifying the --force switch.
3842 * sd-event gained a new function sd_event_get_iteration() for
3843 requesting the current iteration counter of the event loop. It starts
3844 at zero and is increased by one with each event loop iteration.
3846 * A new rpm macro %systemd_ordering is provided by the macros.systemd
3847 file. It can be used in lieu of %systemd_requires in packages which
3848 don't use any systemd functionality and are intended to be installed
3849 in minimal containers without systemd present. This macro provides
3850 ordering dependencies to ensure that if the package is installed in
3851 the same rpm transaction as systemd, systemd will be installed before
3852 the scriptlets for the package are executed, allowing unit presets
3855 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
3856 been added to simplify packaging of generators.
3858 * The os-release file gained VERSION_CODENAME field for the
3859 distribution nickname (e.g. VERSION_CODENAME=woody).
3861 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
3862 can be set to disable parsing of metadata and the creation
3863 of persistent symlinks for that device.
3865 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
3866 to make them available to logged-in users has been reverted.
3868 * Much of the common code of the various systemd components is now
3869 built into an internal shared library libsystemd-shared-231.so
3870 (incorporating the systemd version number in the name, to be updated
3871 with future releases) that the components link to. This should
3872 decrease systemd footprint both in memory during runtime and on
3873 disk. Note that the shared library is not for public use, and is
3874 neither API nor ABI stable, but is likely to change with every new
3875 released update. Packagers need to make sure that binaries
3876 linking to libsystemd-shared.so are updated in step with the
3879 * Configuration for "mkosi" is now part of the systemd
3880 repository. mkosi is a tool to easily build legacy-free OS images,
3881 and is available on github: https://github.com/systemd/mkosi. If
3882 "mkosi" is invoked in the build tree a new raw OS image is generated
3883 incorporating the systemd sources currently being worked on and a
3884 clean, fresh distribution installation. The generated OS image may be
3885 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
3886 UEFI PC. This functionality is particularly useful to easily test
3887 local changes made to systemd in a pristine, defined environment. See
3888 doc/HACKING for details.
3890 * configure learned the --with-support-url= option to specify the
3891 distribution's bugtracker.
3893 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
3894 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
3895 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
3896 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
3897 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
3898 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
3899 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
3900 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
3901 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
3902 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
3903 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
3904 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
3905 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
3906 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
3907 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
3908 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
3909 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
3910 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
3911 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
3913 — Berlin, 2016-07-25
3917 * DNSSEC is now turned on by default in systemd-resolved (in
3918 "allow-downgrade" mode), but may be turned off during compile time by
3919 passing "--with-default-dnssec=no" to "configure" (and of course,
3920 during runtime with DNSSEC= in resolved.conf). We recommend
3921 downstreams to leave this on at least during development cycles and
3922 report any issues with the DNSSEC logic upstream. We are very
3923 interested in collecting feedback about the DNSSEC validator and its
3924 limitations in the wild. Note however, that DNSSEC support is
3925 probably nothing downstreams should turn on in stable distros just
3926 yet, as it might create incompatibilities with a few DNS servers and
3927 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
3928 automatically whenever we detect such incompatible setups, but there
3929 might be systems we do not cover yet. Hence: please help us testing
3930 the DNSSEC code, leave this on where you can, report back, but then
3931 again don't consider turning this on in your stable, LTS or
3932 production release just yet. (Note that you have to enable
3933 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
3934 and its DNSSEC mode for host name resolution from local
3937 * systemd-resolve conveniently resolves DANE records with the --tlsa
3938 option and OPENPGPKEY records with the --openpgp option. It also
3939 supports dumping raw DNS record data via the new --raw= switch.
3941 * systemd-logind will now by default terminate user processes that are
3942 part of the user session scope unit (session-XX.scope) when the user
3943 logs out. This behavior is controlled by the KillUserProcesses=
3944 setting in logind.conf, and the previous default of "no" is now
3945 changed to "yes". This means that user sessions will be properly
3946 cleaned up after, but additional steps are necessary to allow
3947 intentionally long-running processes to survive logout.
3949 While the user is logged in at least once, user@.service is running,
3950 and any service that should survive the end of any individual login
3951 session can be started at a user service or scope using systemd-run.
3952 systemd-run(1) man page has been extended with an example which shows
3953 how to run screen in a scope unit underneath user@.service. The same
3954 command works for tmux.
3956 After the user logs out of all sessions, user@.service will be
3957 terminated too, by default, unless the user has "lingering" enabled.
3958 To effectively allow users to run long-term tasks even if they are
3959 logged out, lingering must be enabled for them. See loginctl(1) for
3960 details. The default polkit policy was modified to allow users to
3961 set lingering for themselves without authentication.
3963 Previous defaults can be restored at compile time by the
3964 --without-kill-user-processes option to "configure".
3966 * systemd-logind gained new configuration settings SessionsMax= and
3967 InhibitorsMax=, both with a default of 8192. It will not register new
3968 user sessions or inhibitors above this limit.
3970 * systemd-logind will now reload configuration on SIGHUP.
3972 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
3973 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
3974 enable. Also, support for the "io" cgroup controller in the unified
3975 hierarchy has been added, so that the "memory", "pids" and "io" are
3976 now the controllers that are supported on the unified hierarchy.
3978 WARNING: it is not possible to use previous systemd versions with
3979 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
3980 is necessary to also update systemd in the initramfs if using the
3981 unified hierarchy. An updated SELinux policy is also required.
3983 * LLDP support has been extended, and both passive (receive-only) and
3984 active (sender) modes are supported. Passive mode ("routers-only") is
3985 enabled by default in systemd-networkd. Active LLDP mode is enabled
3986 by default for containers on the internal network. The "networkctl
3987 lldp" command may be used to list information gathered. "networkctl
3988 status" will also show basic LLDP information on connected peers now.
3990 * The IAID and DUID unique identifier sent in DHCP requests may now be
3991 configured for the system and each .network file managed by
3992 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
3994 * systemd-networkd gained support for configuring proxy ARP support for
3995 each interface, via the ProxyArp= setting in .network files. It also
3996 gained support for configuring the multicast querier feature of
3997 bridge devices, via the new MulticastQuerier= setting in .netdev
3998 files. Similarly, snooping on the IGMP traffic can be controlled
3999 via the new setting MulticastSnooping=.
4001 A new setting PreferredLifetime= has been added for addresses
4002 configured in .network file to configure the lifetime intended for an
4005 The systemd-networkd DHCP server gained the option EmitRouter=, which
4006 defaults to yes, to configure whether the DHCP Option 3 (Router)
4009 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
4010 systemd-socket-activate and installed into /usr/bin. It is now fully
4013 * systemd-journald now uses separate threads to flush changes to disk
4014 when closing journal files, thus reducing impact of slow disk I/O on
4015 logging performance.
4017 * The sd-journal API gained two new calls
4018 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
4019 can be used to open journal files using file descriptors instead of
4020 file or directory paths. sd_journal_open_container() has been
4021 deprecated, sd_journal_open_directory_fd() should be used instead
4022 with the flag SD_JOURNAL_OS_ROOT.
4024 * journalctl learned a new output mode "-o short-unix" that outputs log
4025 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
4026 UTC). It also gained support for a new --no-hostname setting to
4027 suppress the hostname column in the family of "short" output modes.
4029 * systemd-ask-password now optionally skips printing of the password to
4030 stdout with --no-output which can be useful in scripts.
4032 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
4033 (devices tagged with ID_MAKER_TOOL) are now tagged with
4034 "uaccess" and are available to logged in users.
4036 * The DeviceAllow= unit setting now supports specifiers (with "%").
4038 * "systemctl show" gained a new --value switch, which allows print a
4039 only the contents of a specific unit property, without also printing
4040 the property's name. Similar support was added to "show*" verbs
4041 of loginctl and machinectl that output "key=value" lists.
4043 * A new unit type "generated" was added for files dynamically generated
4044 by generator tools. Similarly, a new unit type "transient" is used
4045 for unit files created using the runtime API. "systemctl enable" will
4046 refuse to operate on such files.
4048 * A new command "systemctl revert" has been added that may be used to
4049 revert to the vendor version of a unit file, in case local changes
4050 have been made by adding drop-ins or overriding the unit file.
4052 * "machinectl clean" gained a new verb to automatically remove all or
4053 just hidden container images.
4055 * systemd-tmpfiles gained support for a new line type "e" for emptying
4056 directories, if they exist, without creating them if they don't.
4058 * systemd-nspawn gained support for automatically patching the UID/GIDs
4059 of the owners and the ACLs of all files and directories in a
4060 container tree to match the UID/GID user namespacing range selected
4061 for the container invocation. This mode is enabled via the new
4062 --private-users-chown switch. It also gained support for
4063 automatically choosing a free, previously unused UID/GID range when
4064 starting a container, via the new --private-users=pick setting (which
4065 implies --private-users-chown). Together, these options for the first
4066 time make user namespacing for nspawn containers fully automatic and
4067 thus deployable. The systemd-nspawn@.service template unit file has
4068 been changed to use this functionality by default.
4070 * systemd-nspawn gained a new --network-zone= switch, that allows
4071 creating ad-hoc virtual Ethernet links between multiple containers,
4072 that only exist as long as at least one container referencing them is
4073 running. This allows easy connecting of multiple containers with a
4074 common link that implements an Ethernet broadcast domain. Each of
4075 these network "zones" may be named relatively freely by the user, and
4076 may be referenced by any number of containers, but each container may
4077 only reference one of these "zones". On the lower level, this is
4078 implemented by an automatically managed bridge network interface for
4079 each zone, that is created when the first container referencing its
4080 zone is created and removed when the last one referencing its zone
4083 * The default start timeout may now be configured on the kernel command
4084 line via systemd.default_timeout_start_sec=. It was already
4085 configurable via the DefaultTimeoutStartSec= option in
4086 /etc/systemd/system.conf.
4088 * Socket units gained a new TriggerLimitIntervalSec= and
4089 TriggerLimitBurst= setting to configure a limit on the activation
4090 rate of the socket unit.
4092 * The LimitNICE= setting now optionally takes normal UNIX nice values
4093 in addition to the raw integer limit value. If the specified
4094 parameter is prefixed with "+" or "-" and is in the range -20..19 the
4095 value is understood as UNIX nice value. If not prefixed like this it
4096 is understood as raw RLIMIT_NICE limit.
4098 * Note that the effect of the PrivateDevices= unit file setting changed
4099 slightly with this release: the per-device /dev file system will be
4100 mounted read-only from this version on, and will have "noexec"
4101 set. This (minor) change of behavior might cause some (exceptional)
4102 legacy software to break, when PrivateDevices=yes is set for its
4103 service. Please leave PrivateDevices= off if you run into problems
4106 * systemd-bootchart has been split out to a separate repository:
4107 https://github.com/systemd/systemd-bootchart
4109 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
4110 merged into the kernel in its current form.
4112 * The compatibility libraries libsystemd-daemon.so,
4113 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
4114 which have been deprecated since systemd-209 have been removed along
4115 with the corresponding pkg-config files. All symbols provided by
4116 those libraries are provided by libsystemd.so.
4118 * The Capabilities= unit file setting has been removed (it is ignored
4119 for backwards compatibility). AmbientCapabilities= and
4120 CapabilityBoundingSet= should be used instead.
4122 * A new special target has been added, initrd-root-device.target,
4123 which creates a synchronization point for dependencies of the root
4124 device in early userspace. Initramfs builders must ensure that this
4125 target is now included in early userspace.
4127 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
4128 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
4129 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
4130 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
4131 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
4132 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
4133 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
4134 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
4135 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
4136 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
4137 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
4138 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
4139 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
4140 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
4141 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
4142 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
4143 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
4144 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
4145 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
4146 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
4147 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
4148 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
4149 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
4150 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
4151 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
4154 — Fairfax, 2016-05-21
4158 * The systemd-resolved DNS resolver service has gained a substantial
4159 set of new features, most prominently it may now act as a DNSSEC
4160 validating stub resolver. DNSSEC mode is currently turned off by
4161 default, but is expected to be turned on by default in one of the
4162 next releases. For now, we invite everybody to test the DNSSEC logic
4163 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
4164 service also gained a full set of D-Bus interfaces, including calls
4165 to configure DNS and DNSSEC settings per link (for use by external
4166 network management software). systemd-resolved and systemd-networkd
4167 now distinguish between "search" and "routing" domains. The former
4168 are used to qualify single-label names, the latter are used purely
4169 for routing lookups within certain domains to specific links.
4170 resolved now also synthesizes RRs for all entries from /etc/hosts.
4172 * The systemd-resolve tool (which is a client utility for
4173 systemd-resolved) has been improved considerably and is now fully
4174 supported and documented. Hence it has moved from /usr/lib/systemd to
4177 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
4180 * The coredump collection logic has been reworked: when a coredump is
4181 collected it is now written to disk, compressed and processed
4182 (including stacktrace extraction) from a new instantiated service
4183 systemd-coredump@.service, instead of directly from the
4184 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
4185 processing large coredumps can take up a substantial amount of
4186 resources and time, and this previously happened entirely outside of
4187 systemd's service supervision. With the new logic the core_pattern
4188 hook only does minimal metadata collection before passing off control
4189 to the new instantiated service, which is configured with a time
4190 limit, a nice level and other settings to minimize negative impact on
4191 the rest of the system. Also note that the new logic will honour the
4192 RLIMIT_CORE setting of the crashed process, which now allows users
4193 and processes to turn off coredumping for their processes by setting
4196 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
4197 and all forked processes by default. Previously, PID 1 would leave
4198 the setting at "0" for all processes, as set by the kernel. Note that
4199 the resource limit traditionally has no effect on the generated
4200 coredumps on the system if the /proc/sys/kernel/core_pattern hook
4201 logic is used. Since the limit is now honoured (see above) its
4202 default has been changed so that the coredumping logic is enabled by
4203 default for all processes, while allowing specific opt-out.
4205 * When the stacktrace is extracted from processes of system users, this
4206 is now done as "systemd-coredump" user, in order to sandbox this
4207 potentially security sensitive parsing operation. (Note that when
4208 processing coredumps of normal users this is done under the user ID
4209 of process that crashed, as before.) Packagers should take notice
4210 that it is now necessary to create the "systemd-coredump" system user
4211 and group at package installation time.
4213 * The systemd-activate socket activation testing tool gained support
4214 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
4215 and --seqpacket switches. It also has been extended to support both
4216 new-style and inetd-style file descriptor passing. Use the new
4217 --inetd switch to request inetd-style file descriptor passing.
4219 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
4220 variable, which takes a boolean value. If set to false, ANSI color
4221 output is disabled in the tools even when run on a terminal that
4224 * The VXLAN support in networkd now supports two new settings
4225 DestinationPort= and PortRange=.
4227 * A new systemd.machine_id= kernel command line switch has been added,
4228 that may be used to set the machine ID in /etc/machine-id if it is
4229 not initialized yet. This command line option has no effect if the
4230 file is already initialized.
4232 * systemd-nspawn gained a new --as-pid2 switch that invokes any
4233 specified command line as PID 2 rather than PID 1 in the
4234 container. In this mode PID 1 is a minimal stub init process that
4235 implements the special POSIX and Linux semantics of PID 1 regarding
4236 signal and child process management. Note that this stub init process
4237 is implemented in nspawn itself and requires no support from the
4238 container image. This new logic is useful to support running
4239 arbitrary commands in the container, as normal processes are
4240 generally not prepared to run as PID 1.
4242 * systemd-nspawn gained a new --chdir= switch for setting the current
4243 working directory for the process started in the container.
4245 * "journalctl /dev/sda" will now output all kernel log messages for
4246 specified device from the current boot, in addition to all devices
4247 that are parents of it. This should make log output about devices
4248 pretty useful, as long as kernel drivers attach enough metadata to
4249 the log messages. (The usual SATA drivers do.)
4251 * The sd-journal API gained two new calls
4252 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
4253 that report whether log data from /run or /var has been found.
4255 * journalctl gained a new switch "--fields" that prints all journal
4256 record field names currently in use in the journal. This is backed
4257 by two new sd-journal API calls sd_journal_enumerate_fields() and
4258 sd_journal_restart_fields().
4260 * Most configurable timeouts in systemd now expect an argument of
4261 "infinity" to turn them off, instead of "0" as before. The semantics
4262 from now on is that a timeout of "0" means "now", and "infinity"
4263 means "never". To maintain backwards compatibility, "0" continues to
4264 turn off previously existing timeout settings.
4266 * "systemctl reload-or-try-restart" has been renamed to "systemctl
4267 try-reload-or-restart" to clarify what it actually does: the "try"
4268 logic applies to both reloading and restarting, not just restarting.
4269 The old name continues to be accepted for compatibility.
4271 * On boot-up, when PID 1 detects that the system clock is behind the
4272 release date of the systemd version in use, the clock is now set
4273 to the latter. Previously, this was already done in timesyncd, in order
4274 to avoid running with clocks set to the various clock epochs such as
4275 1902, 1938 or 1970. With this change the logic is now done in PID 1
4276 in addition to timesyncd during early boot-up, so that it is enforced
4277 before the first process is spawned by systemd. Note that the logic
4278 in timesyncd remains, as it is more comprehensive and ensures
4279 clock monotonicity by maintaining a persistent timestamp file in
4280 /var. Since /var is generally not available in earliest boot or the
4281 initrd, this part of the logic remains in timesyncd, and is not done
4284 * Support for tweaking details in net_cls.class_id through the
4285 NetClass= configuration directive has been removed, as the kernel
4286 people have decided to deprecate that controller in cgroup v2.
4287 Userspace tools such as nftables are moving over to setting rules
4288 that are specific to the full cgroup path of a task, which obsoletes
4289 these controllers anyway. The NetClass= directive is kept around for
4290 legacy compatibility reasons. For a more in-depth description of the
4291 kernel change, please refer to the respective upstream commit:
4293 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
4295 * A new service setting RuntimeMaxSec= has been added that may be used
4296 to specify a maximum runtime for a service. If the timeout is hit, the
4297 service is terminated and put into a failure state.
4299 * A new service setting AmbientCapabilities= has been added. It allows
4300 configuration of additional Linux process capabilities that are
4301 passed to the activated processes. This is only available on very
4304 * The process resource limit settings in service units may now be used
4305 to configure hard and soft limits individually.
4307 * The various libsystemd APIs such as sd-bus or sd-event now publicly
4308 expose support for gcc's __attribute__((cleanup())) C extension.
4309 Specifically, for many object destructor functions alternative
4310 versions have been added that have names suffixed with "p" and take a
4311 pointer to a pointer to the object to destroy, instead of just a
4312 pointer to the object itself. This is useful because these destructor
4313 functions may be used directly as parameters to the cleanup
4314 construct. Internally, systemd has been a heavy user of this GCC
4315 extension for a long time, and with this change similar support is
4316 now available to consumers of the library outside of systemd. Note
4317 that by using this extension in your sources compatibility with old
4318 and strictly ANSI compatible C compilers is lost. However, all gcc or
4319 LLVM versions of recent years support this extension.
4321 * Timer units gained support for a new setting RandomizedDelaySec= that
4322 allows configuring some additional randomized delay to the configured
4323 time. This is useful to spread out timer events to avoid load peaks in
4324 clusters or larger setups.
4326 * Calendar time specifications now support sub-second accuracy.
4328 * Socket units now support listening on SCTP and UDP-lite protocol
4331 * The sd-event API now comes with a full set of man pages.
4333 * Older versions of systemd contained experimental support for
4334 compressing journal files and coredumps with the LZ4 compressor that
4335 was not compatible with the lz4 binary (due to API limitations of the
4336 lz4 library). This support has been removed; only support for files
4337 compatible with the lz4 binary remains. This LZ4 logic is now
4338 officially supported and no longer considered experimental.
4340 * The dkr image import logic has been removed again from importd. dkr's
4341 micro-services focus doesn't fit into the machine image focus of
4342 importd, and quickly got out of date with the upstream dkr API.
4344 * Creation of the /run/lock/lockdev/ directory was dropped from
4345 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
4346 been available for many years. If you still need this, you need to
4347 create your own tmpfiles.d config file with:
4349 d /run/lock/lockdev 0775 root lock -
4351 * The settings StartLimitBurst=, StartLimitInterval=, StartLimitAction=
4352 and RebootArgument= have been moved from the [Service] section of
4353 unit files to [Unit], and they are now supported on all unit types,
4354 not just service units. Of course, systemd will continue to
4355 understand these settings also at the old location, in order to
4356 maintain compatibility.
4358 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
4359 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
4360 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
4361 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
4362 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
4363 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
4364 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
4365 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
4366 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
4367 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
4368 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
4369 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
4370 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
4371 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
4372 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
4373 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
4374 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
4375 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
4376 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
4378 — Berlin, 2016-02-11
4382 * A number of properties previously only settable in unit
4383 files are now also available as properties to set when
4384 creating transient units programmatically via the bus, as it
4385 is exposed with systemd-run's --property=
4386 setting. Specifically, these are: SyslogIdentifier=,
4387 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
4388 EnvironmentFile=, ReadWriteDirectories=,
4389 ReadOnlyDirectories=, InaccessibleDirectories=,
4390 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
4392 * When creating transient services via the bus API it is now
4393 possible to pass in a set of file descriptors to use as
4394 STDIN/STDOUT/STDERR for the invoked process.
4396 * Slice units may now be created transiently via the bus APIs,
4397 similar to the way service and scope units may already be
4398 created transiently.
4400 * Wherever systemd expects a calendar timestamp specification
4401 (like in journalctl's --since= and --until= switches) UTC
4402 timestamps are now supported. Timestamps suffixed with "UTC"
4403 are now considered to be in Universal Time Coordinated
4404 instead of the local timezone. Also, timestamps may now
4405 optionally be specified with sub-second accuracy. Both of
4406 these additions also apply to recurring calendar event
4407 specification, such as OnCalendar= in timer units.
4409 * journalctl gained a new "--sync" switch that asks the
4410 journal daemon to write all so far unwritten log messages to
4411 disk and sync the files, before returning.
4413 * systemd-tmpfiles learned two new line types "q" and "Q" that
4414 operate like "v", but also set up a basic btrfs quota
4415 hierarchy when used on a btrfs file system with quota
4418 * tmpfiles' "v", "q" and "Q" will now create a plain directory
4419 instead of a subvolume (even on a btrfs file system) if the
4420 root directory is a plain directory, and not a
4421 subvolume. This should simplify things with certain chroot()
4422 environments which are not aware of the concept of btrfs
4425 * systemd-detect-virt gained a new --chroot switch to detect
4426 whether execution takes place in a chroot() environment.
4428 * CPUAffinity= now takes CPU index ranges in addition to
4431 * The various memory-related resource limit settings (such as
4432 LimitAS=) now understand the usual K, M, G, ... suffixes to
4433 the base of 1024 (IEC). Similar, the time-related resource
4434 limit settings understand the usual min, h, day, ...
4437 * There's a new system.conf setting DefaultTasksMax= to
4438 control the default TasksMax= setting for services and
4439 scopes running on the system. (TasksMax= is the primary
4440 setting that exposes the "pids" cgroup controller on systemd
4441 and was introduced in the previous systemd release.) The
4442 setting now defaults to 512, which means services that are
4443 not explicitly configured otherwise will only be able to
4444 create 512 processes or threads at maximum, from this
4445 version on. Note that this means that thread- or
4446 process-heavy services might need to be reconfigured to set
4447 TasksMax= to a higher value. It is sufficient to set
4448 TasksMax= in these specific unit files to a higher value, or
4449 even "infinity". Similar, there's now a logind.conf setting
4450 UserTasksMax= that defaults to 4096 and limits the total
4451 number of processes or tasks each user may own
4452 concurrently. nspawn containers also have the TasksMax=
4453 value set by default now, to 8192. Note that all of this
4454 only has an effect if the "pids" cgroup controller is
4455 enabled in the kernel. The general benefit of these changes
4456 should be a more robust and safer system, that provides a
4457 certain amount of per-service fork() bomb protection.
4459 * systemd-nspawn gained the new --network-veth-extra= switch
4460 to define additional and arbitrarily-named virtual Ethernet
4461 links between the host and the container.
4463 * A new service execution setting PassEnvironment= has been
4464 added that allows importing select environment variables
4465 from PID1's environment block into the environment block of
4468 * Timer units gained support for a new RemainAfterElapse=
4469 setting which takes a boolean argument. It defaults to on,
4470 exposing behaviour unchanged to previous releases. If set to
4471 off, timer units are unloaded after they elapsed if they
4472 cannot elapse again. This is particularly useful for
4473 transient timer units, which shall not stay around longer
4474 than until they first elapse.
4476 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
4477 default now (the kernel default is 16). This is beneficial
4478 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
4479 allows substantially larger numbers of queued
4480 datagrams. This should increase the capability of systemd to
4481 parallelize boot-up, as logging and sd_notify() are unlikely
4482 to stall execution anymore. If you need to change the value
4483 from the new defaults, use the usual sysctl.d/ snippets.
4485 * The compression framing format used by the journal or
4486 coredump processing has changed to be in line with what the
4487 official LZ4 tools generate. LZ4 compression support in
4488 systemd was considered unsupported previously, as the format
4489 was not compatible with the normal tools. With this release
4490 this has changed now, and it is hence safe for downstream
4491 distributions to turn it on. While not compressing as well
4492 as the XZ, LZ4 is substantially faster, which makes
4493 it a good default choice for the compression logic in the
4494 journal and in coredump handling.
4496 * Any reference to /etc/mtab has been dropped from
4497 systemd. The file has been obsolete since a while, but
4498 systemd refused to work on systems where it was incorrectly
4499 set up (it should be a symlink or non-existent). Please make
4500 sure to update to util-linux 2.27.1 or newer in conjunction
4501 with this systemd release, which also drops any reference to
4502 /etc/mtab. If you maintain a distribution make sure that no
4503 software you package still references it, as this is a
4504 likely source of bugs. There's also a glibc bug pending,
4505 asking for removal of any reference to this obsolete file:
4507 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
4509 Note that only util-linux versions built with
4510 --enable-libmount-force-mountinfo are supported.
4512 * Support for the ".snapshot" unit type has been removed. This
4513 feature turned out to be little useful and little used, and
4514 has now been removed from the core and from systemctl.
4516 * The dependency types RequiresOverridable= and
4517 RequisiteOverridable= have been removed from systemd. They
4518 have been used only very sparingly to our knowledge and
4519 other options that provide a similar effect (such as
4520 systemctl --mode=ignore-dependencies) are much more useful
4521 and commonly used. Moreover, they were only half-way
4522 implemented as the option to control behaviour regarding
4523 these dependencies was never added to systemctl. By removing
4524 these dependency types the execution engine becomes a bit
4525 simpler. Unit files that use these dependencies should be
4526 changed to use the non-Overridable dependency types
4527 instead. In fact, when parsing unit files with these
4528 options, that's what systemd will automatically convert them
4529 too, but it will also warn, asking users to fix the unit
4530 files accordingly. Removal of these dependency types should
4531 only affect a negligible number of unit files in the wild.
4533 * Behaviour of networkd's IPForward= option changed
4534 (again). It will no longer maintain a per-interface setting,
4535 but propagate one way from interfaces where this is enabled
4536 to the global kernel setting. The global setting will be
4537 enabled when requested by a network that is set up, but
4538 never be disabled again. This change was made to make sure
4539 IPv4 and IPv6 behaviour regarding packet forwarding is
4540 similar (as the Linux IPv6 stack does not support
4541 per-interface control of this setting) and to minimize
4544 * In unit files the behaviour of %u, %U, %h, %s has
4545 changed. These specifiers will now unconditionally resolve
4546 to the various user database fields of the user that the
4547 systemd instance is running as, instead of the user
4548 configured in the specific unit via User=. Note that this
4549 effectively doesn't change much, as resolving of these
4550 specifiers was already turned off in the --system instance
4551 of systemd, as we cannot do NSS lookups from PID 1. In the
4552 --user instance of systemd these specifiers where correctly
4553 resolved, but hardly made any sense, since the user instance
4554 lacks privileges to do user switches anyway, and User= is
4555 hence useless. Moreover, even in the --user instance of
4556 systemd behaviour was awkward as it would only take settings
4557 from User= assignment placed before the specifier into
4558 account. In order to unify and simplify the logic around
4559 this the specifiers will now always resolve to the
4560 credentials of the user invoking the manager (which in case
4561 of PID 1 is the root user).
4563 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
4564 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
4565 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
4566 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
4567 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
4568 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
4569 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
4570 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
4571 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
4572 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
4573 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
4574 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
4575 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
4576 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
4579 — Berlin, 2015-11-18
4583 * systemd now depends on util-linux v2.27. More specifically,
4584 the newly added mount monitor feature in libmount now
4585 replaces systemd's former own implementation.
4587 * libmount mandates /etc/mtab not to be regular file, and
4588 systemd now enforces this condition at early boot.
4589 /etc/mtab has been deprecated and warned about for a very
4590 long time, so systems running systemd should already have
4591 stopped having this file around as anything else than a
4592 symlink to /proc/self/mounts.
4594 * Support for the "pids" cgroup controller has been added. It
4595 allows accounting the number of tasks in a cgroup and
4596 enforcing limits on it. This adds two new setting
4597 TasksAccounting= and TasksMax= to each unit, as well as a
4598 global option DefaultTasksAccounting=.
4600 * Support for the "net_cls" cgroup controller has been added.
4601 It allows assigning a net class ID to each task in the
4602 cgroup, which can then be used in firewall rules and traffic
4603 shaping configurations. Note that the kernel netfilter net
4604 class code does not currently work reliably for ingress
4605 packets on unestablished sockets.
4607 This adds a new config directive called NetClass= to CGroup
4608 enabled units. Allowed values are positive numbers for fixed
4609 assignments and "auto" for picking a free value
4612 * 'systemctl is-system-running' now returns 'offline' if the
4613 system is not booted with systemd. This command can now be
4614 used as a substitute for 'systemd-notify --booted'.
4616 * Watchdog timeouts have been increased to 3 minutes for all
4617 in-tree service files. Apparently, disk IO issues are more
4618 frequent than we hoped, and user reported >1 minute waiting
4621 * 'machine-id-commit' functionality has been merged into
4622 'machine-id-setup --commit'. The separate binary has been
4625 * The WorkingDirectory= directive in unit files may now be set
4626 to the special value '~'. In this case, the working
4627 directory is set to the home directory of the user
4628 configured in User=.
4630 * "machinectl shell" will now open the shell in the home
4631 directory of the selected user by default.
4633 * The CrashChVT= configuration file setting is renamed to
4634 CrashChangeVT=, following our usual logic of not
4635 abbreviating unnecessarily. The old directive is still
4636 supported for compat reasons. Also, this directive now takes
4637 an integer value between 1 and 63, or a boolean value. The
4638 formerly supported '-1' value for disabling stays around for
4641 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
4642 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
4643 RootDirectory= properties can now be set for transient
4646 * The systemd-analyze tool gained a new "set-log-target" verb
4647 to change the logging target the system manager logs to
4648 dynamically during runtime. This is similar to how
4649 "systemd-analyze set-log-level" already changes the log
4652 * In nspawn /sys is now mounted as tmpfs, with only a selected
4653 set of subdirectories mounted in from the real sysfs. This
4654 enhances security slightly, and is useful for ensuring user
4655 namespaces work correctly.
4657 * Support for USB FunctionFS activation has been added. This
4658 allows implementation of USB gadget services that are
4659 activated as soon as they are requested, so that they don't
4660 have to run continuously, similar to classic socket
4663 * The "systemctl exit" command now optionally takes an
4664 additional parameter that sets the exit code to return from
4665 the systemd manager when exiting. This is only relevant when
4666 running the systemd user instance, or when running the
4667 system instance in a container.
4669 * sd-bus gained the new API calls sd_bus_path_encode_many()
4670 and sd_bus_path_decode_many() that allow easy encoding and
4671 decoding of multiple identifier strings inside a D-Bus
4672 object path. Another new call sd_bus_default_flush_close()
4673 has been added to flush and close per-thread default
4676 * systemd-cgtop gained support for a -M/--machine= switch to
4677 show the control groups within a certain container only.
4679 * "systemctl kill" gained support for an optional --fail
4680 switch. If specified the requested operation will fail of no
4681 processes have been killed, because the unit had no
4682 processes attached, or similar.
4684 * A new systemd.crash_reboot=1 kernel command line option has
4685 been added that triggers a reboot after crashing. This can
4686 also be set through CrashReboot= in systemd.conf.
4688 * The RuntimeDirectory= setting now understands unit
4689 specifiers like %i or %f.
4691 * A new (still internal) library API sd-ipv4acd has been added,
4692 that implements address conflict detection for IPv4. It's
4693 based on code from sd-ipv4ll, and will be useful for
4694 detecting DHCP address conflicts.
4696 * File descriptors passed during socket activation may now be
4697 named. A new API sd_listen_fds_with_names() is added to
4698 access the names. The default names may be overridden,
4699 either in the .socket file using the FileDescriptorName=
4700 parameter, or by passing FDNAME= when storing the file
4701 descriptors using sd_notify().
4703 * systemd-networkd gained support for:
4705 - Setting the IPv6 Router Advertisement settings via
4706 IPv6AcceptRouterAdvertisements= in .network files.
4708 - Configuring the HelloTimeSec=, MaxAgeSec= and
4709 ForwardDelaySec= bridge parameters in .netdev files.
4711 - Configuring PreferredSource= for static routes in
4714 * The "ask-password" framework used to query for LUKS harddisk
4715 passwords or SSL passwords during boot gained support for
4716 caching passwords in the kernel keyring, if it is
4717 available. This makes sure that the user only has to type in
4718 a passphrase once if there are multiple objects to unlock
4719 with the same one. Previously, such password caching was
4720 available only when Plymouth was used; this moves the
4721 caching logic into the systemd codebase itself. The
4722 "systemd-ask-password" utility gained a new --keyname=
4723 switch to control which kernel keyring key to use for
4724 caching a password in. This functionality is also useful for
4725 enabling display managers such as gdm to automatically
4726 unlock the user's GNOME keyring if its passphrase, the
4727 user's password and the harddisk password are the same, if
4728 gdm-autologin is used.
4730 * When downloading tar or raw images using "machinectl
4731 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
4732 file is now also downloaded, if it is available and stored
4733 next to the image file.
4735 * Units of type ".socket" gained a new boolean setting
4736 Writable= which is only useful in conjunction with
4737 ListenSpecial=. If true, enables opening the specified
4738 special file in O_RDWR mode rather than O_RDONLY mode.
4740 * systemd-rfkill has been reworked to become a singleton
4741 service that is activated through /dev/rfkill on each rfkill
4742 state change and saves the settings to disk. This way,
4743 systemd-rfkill is now compatible with devices that exist
4744 only intermittendly, and even restores state if the previous
4745 system shutdown was abrupt rather than clean.
4747 * The journal daemon gained support for vacuuming old journal
4748 files controlled by the number of files that shall remain,
4749 in addition to the already existing control by size and by
4750 date. This is useful as journal interleaving performance
4751 degrades with too many separate journal files, and allows
4752 putting an effective limit on them. The new setting defaults
4753 to 100, but this may be changed by setting SystemMaxFiles=
4754 and RuntimeMaxFiles= in journald.conf. Also, the
4755 "journalctl" tool gained the new --vacuum-files= switch to
4756 manually vacuum journal files to leave only the specified
4757 number of files in place.
4759 * udev will now create /dev/disk/by-path links for ATA devices
4760 on kernels where that is supported.
4762 * Galician, Serbian, Turkish and Korean translations were added.
4764 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
4765 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
4766 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
4767 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
4768 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
4769 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
4770 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
4771 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
4772 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
4773 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
4774 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
4775 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
4776 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
4777 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
4778 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
4779 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
4780 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
4781 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
4783 — Berlin, 2015-10-07
4787 * The DHCP implementation of systemd-networkd gained a set of
4790 - The DHCP server now supports emitting DNS and NTP
4791 information. It may be enabled and configured via
4792 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
4793 and NTP information is enabled, but no servers are
4794 configured, the corresponding uplink information (if there
4795 is any) is propagated.
4797 - Server and client now support transmission and reception
4798 of timezone information. It can be configured via the
4799 newly introduced network options UseTimezone=,
4800 EmitTimezone=, and Timezone=. Transmission of timezone
4801 information is enabled between host and containers by
4802 default now: the container will change its local timezone
4803 to what the host has set.
4805 - Lease timeouts can now be configured via
4806 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
4808 - The DHCP server improved on the stability of
4809 leases. Clients are more likely to get the same lease
4810 information back, even if the server loses state.
4812 - The DHCP server supports two new configuration options to
4813 control the lease address pool metrics, PoolOffset= and
4816 * The encapsulation limit of tunnels in systemd-networkd may
4817 now be configured via 'EncapsulationLimit='. It allows
4818 modifying the maximum additional levels of encapsulation
4819 that are permitted to be prepended to a packet.
4821 * systemd now supports the concept of user buses replacing
4822 session buses, if used with dbus-1.10 (and enabled via dbus
4823 --enable-user-session). It previously only supported this on
4824 kdbus-enabled systems, and this release expands this to
4825 'dbus-daemon' systems.
4827 * systemd-networkd now supports predictable interface names
4830 * systemd now optionally supports the new Linux kernel
4831 "unified" control group hierarchy. If enabled via the kernel
4832 command-line option 'systemd.unified_cgroup_hierarchy=1',
4833 systemd will try to mount the unified cgroup hierarchy
4834 directly on /sys/fs/cgroup. If not enabled, or not
4835 available, systemd will fall back to the legacy cgroup
4836 hierarchy setup, as before. Host system and containers can
4837 mix and match legacy and unified hierarchies as they
4838 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
4839 environment variable to individually select the hierarchy to
4840 use for executed containers. By default, nspawn will use the
4841 unified hierarchy for the containers if the host uses the
4842 unified hierarchy, and the legacy hierarchy otherwise.
4843 Please note that at this point the unified hierarchy is an
4844 experimental kernel feature and is likely to change in one
4845 of the next kernel releases. Therefore, it should not be
4846 enabled by default in downstream distributions yet. The
4847 minimum required kernel version for the unified hierarchy to
4848 work is 4.2. Note that when the unified hierarchy is used
4849 for the first time delegated access to controllers is
4850 safe. Because of this systemd-nspawn containers will get
4851 access to controllers now, as will systemd user
4852 sessions. This means containers and user sessions may now
4853 manage their own resources, partitioning up what the system
4856 * A new special scope unit "init.scope" has been introduced
4857 that encapsulates PID 1 of the system. It may be used to
4858 determine resource usage and enforce resource limits on PID
4859 1 itself. PID 1 hence moved out of the root of the control
4862 * The cgtop tool gained support for filtering out kernel
4863 threads when counting tasks in a control group. Also, the
4864 count of processes is now recursively summed up by
4865 default. Two options -k and --recursive= have been added to
4866 revert to old behaviour. The tool has also been updated to
4867 work correctly in containers now.
4869 * systemd-nspawn's --bind= and --bind-ro= options have been
4870 extended to allow creation of non-recursive bind mounts.
4872 * libsystemd gained two new calls sd_pid_get_cgroup() and
4873 sd_peer_get_cgroup() which return the control group path of
4874 a process or peer of a connected AF_UNIX socket. This
4875 function call is particularly useful when implementing
4876 delegated subtrees support in the control group hierarchy.
4878 * The "sd-event" event loop API of libsystemd now supports
4879 correct dequeuing of real-time signals, without losing
4882 * When systemd requests a polkit decision when managing units it
4883 will now add additional fields to the request, including unit
4884 name and desired operation. This enables more powerful polkit
4885 policies, that make decisions depending on these parameters.
4887 * nspawn learnt support for .nspawn settings files, that may
4888 accompany the image files or directories of containers, and
4889 may contain additional settings for the container. This is
4890 an alternative to configuring container parameters via the
4891 nspawn command line.
4893 Contributions from: Cristian Rodríguez, Daniel Mack, David
4894 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
4895 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
4896 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
4897 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
4898 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
4899 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
4900 Andersen, Tom Gundersen, Torstein Husebø
4902 — Berlin, 2015-09-08
4906 * machinectl gained a new verb 'shell' which opens a fresh
4907 shell on the target container or the host. It is similar to
4908 the existing 'login' command of machinectl, but spawns the
4909 shell directly without prompting for username or
4910 password. The pseudo machine '.host' now refers to the local
4911 host and is used by default. Hence, 'machinectl shell' can
4912 be used as replacement for 'su -' which spawns a session as
4913 a fresh systemd unit in a way that is fully isolated from
4914 the originating session.
4916 * systemd-networkd learned to cope with private-zone DHCP
4917 options and allows other programs to query the values.
4919 * SELinux access control when enabling/disabling units is no
4920 longer enforced with this release. The previous implementation
4921 was incorrect, and a new corrected implementation is not yet
4922 available. As unit file operations are still protected via
4923 polkit and D-Bus policy this is not a security problem. Yet,
4924 distributions which care about optimal SELinux support should
4925 probably not stabilize on this release.
4927 * sd-bus gained support for matches of type "arg0has=", that
4928 test for membership of strings in string arrays sent in bus
4931 * systemd-resolved now dumps the contents of its DNS and LLMNR
4932 caches to the logs on reception of the SIGUSR1 signal. This
4933 is useful to debug DNS behaviour.
4935 * The coredumpctl tool gained a new --directory= option to
4936 operate on journal files in a specific directory.
4938 * "systemctl reboot" and related commands gained a new
4939 "--message=" option which may be used to set a free-text
4940 wall message when shutting down or rebooting the
4941 system. This message is also logged, which is useful for
4942 figuring out the reason for a reboot or shutdown a
4945 * The "systemd-resolve-host" tool's -i switch now takes
4946 network interface numbers as alternative to interface names.
4948 * A new unit file setting for services has been introduced:
4949 UtmpMode= allows configuration of how precisely systemd
4950 handles utmp and wtmp entries for the service if this is
4951 enabled. This allows writing services that appear similar to
4952 user sessions in the output of the "w", "who", "last" and
4955 * systemd-resolved will now locally synthesize DNS resource
4956 records for the "localhost" and "gateway" domains as well as
4957 the local hostname. This should ensure that clients querying
4958 RRs via resolved will get similar results as those going via
4959 NSS, if nss-myhostname is enabled.
4961 Contributions from: Alastair Hughes, Alex Crawford, Daniel
4962 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
4963 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
4964 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
4965 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
4966 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
4967 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
4968 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
4969 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
4970 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
4971 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
4972 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
4974 — Berlin, 2015-08-27
4978 * The systemd-efi-boot-generator functionality was merged into
4979 systemd-gpt-auto-generator.
4981 * systemd-networkd now supports Group Policy for vxlan
4982 devices. It can be enabled via the new boolean configuration
4983 option called 'GroupPolicyExtension='.
4985 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
4986 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
4987 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
4989 — Berlin, 2015-07-31
4993 * The python-systemd code has been removed from the systemd repository.
4994 A new repository has been created which accommodates the code from
4995 now on, and we kindly ask distributions to create a separate package
4996 for this: https://github.com/systemd/python-systemd
4998 * The systemd daemon will now reload its main configuration
4999 (/etc/systemd/system.conf) on daemon-reload.
5001 * sd-dhcp now exposes vendor specific extensions via
5002 sd_dhcp_lease_get_vendor_specific().
5004 * systemd-networkd gained a number of new configuration options.
5006 - A new boolean configuration option for TAP devices called
5007 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
5008 device, thus allowing to send and receive GSO packets.
5010 - A new tunnel configuration option called 'CopyDSCP='.
5011 If enabled, the DSCP field of ip6 tunnels is copied into the
5012 decapsulated packet.
5014 - A set of boolean bridge configuration options were added.
5015 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
5016 and 'UnicastFlood=' are now parsed by networkd and applied to the
5017 respective bridge link device via the respective IFLA_BRPORT_*
5020 - A new string configuration option to override the hostname sent
5021 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
5022 is true, networkd will use the configured hostname instead of the
5023 system hostname when sending DHCP requests.
5025 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
5026 networkd will configure the IPv6 flow-label of the tunnel device
5027 according to RFC2460.
5029 - The 'macvtap' virtual network devices are now supported, similar to
5030 the already supported 'macvlan' devices.
5032 * systemd-resolved now implements RFC5452 to improve resilience against
5033 cache poisoning. Additionally, source port randomization is enabled
5034 by default to further protect against DNS spoofing attacks.
5036 * nss-mymachines now supports translating UIDs and GIDs of running
5037 containers with user-namespaces enabled. If a container 'foo'
5038 translates a host uid 'UID' to the container uid 'TUID', then
5039 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
5040 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
5041 mapped as 'vg-foo-TGID'.
5043 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
5044 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
5045 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
5046 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
5047 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
5048 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
5049 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
5050 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
5051 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
5052 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
5054 — Berlin, 2015-07-29
5058 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
5059 There are no known issues with current sysfs, and udev does not need
5060 or should be used to work around such bugs.
5062 * udev does no longer enable USB HID power management. Several reports
5063 indicate, that some devices cannot handle that setting.
5065 * The udev accelerometer helper was removed. The functionality
5066 is now fully included in iio-sensor-proxy. But this means,
5067 older iio-sensor-proxy versions will no longer provide
5068 accelerometer/orientation data with this systemd version.
5069 Please upgrade iio-sensor-proxy to version 1.0.
5071 * networkd gained a new configuration option IPv6PrivacyExtensions=
5072 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
5073 for Stateless Address") on selected networks.
5075 * For the sake of fewer build-time dependencies and less code in the
5076 main repository, the python bindings are about to be removed in the
5077 next release. A new repository has been created which accommodates
5078 the code from now on, and we kindly ask distributions to create a
5079 separate package for this. The removal will take place in v223.
5081 https://github.com/systemd/python-systemd
5083 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
5084 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
5085 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
5086 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
5087 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
5088 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
5089 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
5090 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
5091 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
5092 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
5094 — Berlin, 2015-07-07
5098 * The sd-bus.h and sd-event.h APIs have now been declared
5099 stable and have been added to the official interface of
5100 libsystemd.so. sd-bus implements an alternative D-Bus client
5101 library, that is relatively easy to use, very efficient and
5102 supports both classic D-Bus as well as kdbus as transport
5103 backend. sd-event is a generic event loop abstraction that
5104 is built around Linux epoll, but adds features such as event
5105 prioritization or efficient timer handling. Both APIs are good
5106 choices for C programs looking for a bus and/or event loop
5107 implementation that is minimal and does not have to be
5108 portable to other kernels.
5110 * kdbus support is no longer compile-time optional. It is now
5111 always built-in. However, it can still be disabled at
5112 runtime using the kdbus=0 kernel command line setting, and
5113 that setting may be changed to default to off, by specifying
5114 --disable-kdbus at build-time. Note though that the kernel
5115 command line setting has no effect if the kdbus.ko kernel
5116 module is not installed, in which case kdbus is (obviously)
5117 also disabled. We encourage all downstream distributions to
5118 begin testing kdbus by adding it to the kernel images in the
5119 development distributions, and leaving kdbus support in
5122 * The minimal required util-linux version has been bumped to
5125 * Support for chkconfig (--enable-chkconfig) was removed in
5126 favor of calling an abstraction tool
5127 /lib/systemd/systemd-sysv-install. This needs to be
5128 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
5129 in README for details.
5131 * If there's a systemd unit and a SysV init script for the
5132 same service name, and the user executes "systemctl enable"
5133 for it (or a related call), then this will now enable both
5134 (or execute the related operation on both), not just the
5137 * The libudev API documentation has been converted from gtkdoc
5140 * gudev has been removed from the systemd tree, it is now an
5143 * The systemd-cgtop tool learnt a new --raw switch to generate
5144 "raw" (machine parsable) output.
5146 * networkd's IPForwarding= .network file setting learnt the
5147 new setting "kernel", which ensures that networkd does not
5148 change the IP forwarding sysctl from the default kernel
5151 * The systemd-logind bus API now exposes a new boolean
5152 property "Docked" that reports whether logind considers the
5153 system "docked", i.e. connected to a docking station or not.
5155 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
5156 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
5157 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
5158 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
5159 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
5160 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
5161 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
5162 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
5163 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
5164 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
5165 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
5166 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
5167 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
5168 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
5169 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
5170 Fink, Zbigniew Jędrzejewski-Szmek
5172 — Berlin, 2015-06-19
5176 * The gudev library has been extracted into a separate repository
5177 available at: https://git.gnome.org/browse/libgudev/
5178 It is now managed as part of the Gnome project. Distributions
5179 are recommended to pass --disable-gudev to systemd and use
5180 gudev from the Gnome project instead. gudev is still included
5181 in systemd, for now. It will be removed soon, though. Please
5182 also see the announcement-thread on systemd-devel:
5183 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
5185 * systemd now exposes a CPUUsageNSec= property for each
5186 service unit on the bus, that contains the overall consumed
5187 CPU time of a service (the sum of what each process of the
5188 service consumed). This value is only available if
5189 CPUAccounting= is turned on for a service, and is then shown
5190 in the "systemctl status" output.
5192 * Support for configuring alternative mappings of the old SysV
5193 runlevels to systemd targets has been removed. They are now
5194 hardcoded in a way that runlevels 2, 3, 4 all map to
5195 multi-user.target and 5 to graphical.target (which
5196 previously was already the default behaviour).
5198 * The auto-mounter logic gained support for mount point
5199 expiry, using a new TimeoutIdleSec= setting in .automount
5200 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
5202 * The EFI System Partition (ESP) as mounted to /boot by
5203 systemd-efi-boot-generator will now be unmounted
5204 automatically after 2 minutes of not being used. This should
5205 minimize the risk of ESP corruptions.
5207 * New /etc/fstab options x-systemd.requires= and
5208 x-systemd.requires-mounts-for= are now supported to express
5209 additional dependencies for mounts. This is useful for
5210 journalling file systems that support external journal
5211 devices or overlay file systems that require underlying file
5212 systems to be mounted.
5214 * systemd does not support direct live-upgrades (via systemctl
5215 daemon-reexec) from versions older than v44 anymore. As no
5216 distribution we are aware of shipped such old versions in a
5217 stable release this should not be problematic.
5219 * When systemd forks off a new per-connection service instance
5220 it will now set the $REMOTE_ADDR environment variable to the
5221 remote IP address, and $REMOTE_PORT environment variable to
5222 the remote IP port. This behaviour is similar to the
5223 corresponding environment variables defined by CGI.
5225 * systemd-networkd gained support for uplink failure
5226 detection. The BindCarrier= option allows binding interface
5227 configuration dynamically to the link sense of other
5228 interfaces. This is useful to achieve behaviour like in
5231 * systemd-networkd gained support for configuring the DHCP
5232 client identifier to use when requesting leases.
5234 * systemd-networkd now has a per-network UseNTP= option to
5235 configure whether NTP server information acquired via DHCP
5236 is passed on to services like systemd-timesyncd.
5238 * systemd-networkd gained support for vti6 tunnels.
5240 * Note that systemd-networkd manages the sysctl variable
5241 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
5242 it is configured for since v219. The variable controls IP
5243 forwarding, and is a per-interface alternative to the global
5244 /proc/sys/net/ipv[46]/ip_forward. This setting is
5245 configurable in the IPForward= option, which defaults to
5246 "no". This means if networkd is used for an interface it is
5247 no longer sufficient to set the global sysctl option to turn
5248 on IP forwarding! Instead, the .network file option
5249 IPForward= needs to be turned on! Note that the
5250 implementation of this behaviour was broken in v219 and has
5253 * Many bonding and vxlan options are now configurable in
5256 * systemd-nspawn gained a new --property= setting to set unit
5257 properties for the container scope. This is useful for
5258 setting resource parameters (e.g. "CPUShares=500") on
5259 containers started from the command line.
5261 * systemd-nspawn gained a new --private-users= switch to make
5262 use of user namespacing available on recent Linux kernels.
5264 * systemd-nspawn may now be called as part of a shell pipeline
5265 in which case the pipes used for stdin and stdout are passed
5266 directly to the process invoked in the container, without
5267 indirection via a pseudo tty.
5269 * systemd-nspawn gained a new switch to control the UNIX
5270 signal to use when killing the init process of the container
5273 * systemd-nspawn gained a new --overlay= switch for mounting
5274 overlay file systems into the container using the new kernel
5277 * When a container image is imported via systemd-importd and
5278 the host file system is not btrfs, a loopback block device
5279 file is created in /var/lib/machines.raw with a btrfs file
5280 system inside. It is then mounted to /var/lib/machines to
5281 enable btrfs features for container management. The loopback
5282 file and btrfs file system is grown as needed when container
5283 images are imported via systemd-importd.
5285 * systemd-machined/systemd-importd gained support for btrfs
5286 quota, to enforce container disk space limits on disk. This
5287 is exposed in "machinectl set-limit".
5289 * systemd-importd now can import containers from local .tar,
5290 .raw and .qcow2 images, and export them to .tar and .raw. It
5291 can also import dkr v2 images now from the network (on top
5294 * systemd-importd gained support for verifying downloaded
5295 images with gpg2 (previously only gpg1 was supported).
5297 * systemd-machined, systemd-logind, systemd: most bus calls are
5298 now accessible to unprivileged processes via polkit. Also,
5299 systemd-logind will now allow users to kill their own sessions
5300 without further privileges or authorization.
5302 * systemd-shutdownd has been removed. This service was
5303 previously responsible for implementing scheduled shutdowns
5304 as exposed in /usr/bin/shutdown's time parameter. This
5305 functionality has now been moved into systemd-logind and is
5306 accessible via a bus interface.
5308 * "systemctl reboot" gained a new switch --firmware-setup that
5309 can be used to reboot into the EFI firmware setup, if that
5310 is available. systemd-logind now exposes an API on the bus
5311 to trigger such reboots, in case graphical desktop UIs want
5312 to cover this functionality.
5314 * "systemctl enable", "systemctl disable" and "systemctl mask"
5315 now support a new "--now" switch. If specified the units
5316 that are enabled will also be started, and the ones
5317 disabled/masked also stopped.
5319 * The Gummiboot EFI boot loader tool has been merged into
5320 systemd, and renamed to "systemd-boot". The bootctl tool has been
5321 updated to support systemd-boot.
5323 * An EFI kernel stub has been added that may be used to create
5324 kernel EFI binaries that contain not only the actual kernel,
5325 but also an initrd, boot splash, command line and OS release
5326 information. This combined binary can then be signed as a
5327 single image, so that the firmware can verify it all in one
5328 step. systemd-boot has special support for EFI binaries created
5329 like this and can extract OS release information from them
5330 and show them in the boot menu. This functionality is useful
5331 to implement cryptographically verified boot schemes.
5333 * Optional support has been added to systemd-fsck to pass
5334 fsck's progress report to an AF_UNIX socket in the file
5337 * udev will no longer create device symlinks for all block
5338 devices by default. A blacklist for excluding special block
5339 devices from this logic has been turned into a whitelist
5340 that requires picking block devices explicitly that require
5343 * A new (currently still internal) API sd-device.h has been
5344 added to libsystemd. This modernized API is supposed to
5345 replace libudev eventually. In fact, already much of libudev
5346 is now just a wrapper around sd-device.h.
5348 * A new hwdb database for storing metadata about pointing
5349 stick devices has been added.
5351 * systemd-tmpfiles gained support for setting file attributes
5352 similar to the "chattr" tool with new 'h' and 'H' lines.
5354 * systemd-journald will no longer unconditionally set the
5355 btrfs NOCOW flag on new journal files. This is instead done
5356 with tmpfiles snippet using the new 'h' line type. This
5357 allows easy disabling of this logic, by masking the
5358 journal-nocow.conf tmpfiles file.
5360 * systemd-journald will now translate audit message types to
5361 human readable identifiers when writing them to the
5362 journal. This should improve readability of audit messages.
5364 * The LUKS logic gained support for the offset= and skip=
5365 options in /etc/crypttab, as previously implemented by
5368 * /usr/lib/os-release gained a new optional field VARIANT= for
5369 distributions that support multiple variants (such as a
5370 desktop edition, a server edition, ...)
5372 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
5373 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
5374 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
5375 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
5376 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
5377 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
5378 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
5379 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
5380 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
5381 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
5382 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
5383 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
5384 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
5385 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
5386 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
5387 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
5388 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
5389 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
5390 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
5391 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
5392 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
5393 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
5394 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
5395 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
5396 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
5397 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
5398 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
5400 — Berlin, 2015-05-22
5404 * Introduce a new API "sd-hwdb.h" for querying the hardware
5405 metadata database. With this minimal interface one can query
5406 and enumerate the udev hwdb, decoupled from the old libudev
5407 library. libudev's interface for this is now only a wrapper
5408 around sd-hwdb. A new tool systemd-hwdb has been added to
5409 interface with and update the database.
5411 * When any of systemd's tools copies files (for example due to
5412 tmpfiles' C lines) a btrfs reflink will attempted first,
5413 before bytewise copying is done.
5415 * systemd-nspawn gained a new --ephemeral switch. When
5416 specified a btrfs snapshot is taken of the container's root
5417 directory, and immediately removed when the container
5418 terminates again. Thus, a container can be started whose
5419 changes never alter the container's root directory, and are
5420 lost on container termination. This switch can also be used
5421 for starting a container off the root file system of the
5422 host without affecting the host OS. This switch is only
5423 available on btrfs file systems.
5425 * systemd-nspawn gained a new --template= switch. It takes the
5426 path to a container tree to use as template for the tree
5427 specified via --directory=, should that directory be
5428 missing. This allows instantiating containers dynamically,
5429 on first run. This switch is only available on btrfs file
5432 * When a .mount unit refers to a mount point on which multiple
5433 mounts are stacked, and the .mount unit is stopped all of
5434 the stacked mount points will now be unmounted until no
5435 mount point remains.
5437 * systemd now has an explicit notion of supported and
5438 unsupported unit types. Jobs enqueued for unsupported unit
5439 types will now fail with an "unsupported" error code. More
5440 specifically .swap, .automount and .device units are not
5441 supported in containers, .busname units are not supported on
5442 non-kdbus systems. .swap and .automount are also not
5443 supported if their respective kernel compile time options
5446 * machinectl gained support for two new "copy-from" and
5447 "copy-to" commands for copying files from a running
5448 container to the host or vice versa.
5450 * machinectl gained support for a new "bind" command to bind
5451 mount host directories into local containers. This is
5452 currently only supported for nspawn containers.
5454 * networkd gained support for configuring bridge forwarding
5455 database entries (fdb) from .network files.
5457 * A new tiny daemon "systemd-importd" has been added that can
5458 download container images in tar, raw, qcow2 or dkr formats,
5459 and make them available locally in /var/lib/machines, so
5460 that they can run as nspawn containers. The daemon can GPG
5461 verify the downloads (not supported for dkr, since it has no
5462 provisions for verifying downloads). It will transparently
5463 decompress bz2, xz, gzip compressed downloads if necessary,
5464 and restore sparse files on disk. The daemon uses privilege
5465 separation to ensure the actual download logic runs with
5466 fewer privileges than the daemon itself. machinectl has
5467 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
5468 make the functionality of importd available to the
5469 user. With this in place the Fedora and Ubuntu "Cloud"
5470 images can be downloaded and booted as containers unmodified
5471 (the Fedora images lack the appropriate GPG signature files
5472 currently, so they cannot be verified, but this will change
5473 soon, hopefully). Note that downloading images is currently
5474 only fully supported on btrfs.
5476 * machinectl is now able to list container images found in
5477 /var/lib/machines, along with some metadata about sizes of
5478 disk and similar. If the directory is located on btrfs and
5479 quota is enabled, this includes quota display. A new command
5480 "image-status" has been added that shows additional
5481 information about images.
5483 * machinectl is now able to clone container images
5484 efficiently, if the underlying file system (btrfs) supports
5485 it, with the new "machinectl clone" command. It also
5486 gained commands for renaming and removing images, as well as
5487 marking them read-only or read-write (supported also on
5488 legacy file systems).
5490 * networkd gained support for collecting LLDP network
5491 announcements, from hardware that supports this. This is
5492 shown in networkctl output.
5494 * systemd-run gained support for a new -t (--pty) switch for
5495 invoking a binary on a pty whose input and output is
5496 connected to the invoking terminal. This allows executing
5497 processes as system services while interactively
5498 communicating with them via the terminal. Most interestingly
5499 this is supported across container boundaries. Invoking
5500 "systemd-run -t /bin/bash" is an alternative to running a
5501 full login session, the difference being that the former
5502 will not register a session, nor go through the PAM session
5505 * tmpfiles gained support for a new "v" line type for creating
5506 btrfs subvolumes. If the underlying file system is a legacy
5507 file system, this automatically degrades to creating a
5508 normal directory. Among others /var/lib/machines is now
5509 created like this at boot, should it be missing.
5511 * The directory /var/lib/containers/ has been deprecated and
5512 been replaced by /var/lib/machines. The term "machines" has
5513 been used in the systemd context as generic term for both
5514 VMs and containers, and hence appears more appropriate for
5515 this, as the directory can also contain raw images bootable
5518 * systemd-nspawn when invoked with -M but without --directory=
5519 or --image= is now capable of searching for the container
5520 root directory, subvolume or disk image automatically, in
5521 /var/lib/machines. systemd-nspawn@.service has been updated
5522 to make use of this, thus allowing it to be used for raw
5525 * A new machines.target unit has been introduced that is
5526 supposed to group all containers/VMs invoked as services on
5527 the system. systemd-nspawn@.service has been updated to
5528 integrate with that.
5530 * machinectl gained a new "start" command, for invoking a
5531 container as a service. "machinectl start foo" is mostly
5532 equivalent to "systemctl start systemd-nspawn@foo.service",
5533 but handles escaping in a nicer way.
5535 * systemd-nspawn will now mount most of the cgroupfs tree
5536 read-only into each container, with the exception of the
5537 container's own subtree in the name=systemd hierarchy.
5539 * journald now sets the special FS_NOCOW file flag for its
5540 journal files. This should improve performance on btrfs, by
5541 avoiding heavy fragmentation when journald's write-pattern
5542 is used on COW file systems. It degrades btrfs' data
5543 integrity guarantees for the files to the same levels as for
5544 ext3/ext4 however. This should be OK though as journald does
5545 its own data integrity checks and all its objects are
5546 checksummed on disk. Also, journald should handle btrfs disk
5547 full events a lot more gracefully now, by processing SIGBUS
5548 errors, and not relying on fallocate() anymore.
5550 * When journald detects that journal files it is writing to
5551 have been deleted it will immediately start new journal
5554 * systemd now provides a way to store file descriptors
5555 per-service in PID 1. This is useful for daemons to ensure
5556 that fds they require are not lost during a daemon
5557 restart. The fds are passed to the daemon on the next
5558 invocation in the same way socket activation fds are
5559 passed. This is now used by journald to ensure that the
5560 various sockets connected to all the system's stdout/stderr
5561 are not lost when journald is restarted. File descriptors
5562 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
5563 an extension to sd_notify(). Note that a limit is enforced
5564 on the number of fds a service can store in PID 1, and it
5565 defaults to 0, so that no fds may be stored, unless this is
5566 explicitly turned on.
5568 * The default TERM variable to use for units connected to a
5569 terminal, when no other value is explicitly is set is now
5570 vt220 rather than vt102. This should be fairly safe still,
5571 but allows PgUp/PgDn work.
5573 * The /etc/crypttab option header= as known from Debian is now
5576 * "loginctl user-status" and "loginctl session-status" will
5577 now show the last 10 lines of log messages of the
5578 user/session following the status output. Similar,
5579 "machinectl status" will show the last 10 log lines
5580 associated with a virtual machine or container
5581 service. (Note that this is usually not the log messages
5582 done in the VM/container itself, but simply what the
5583 container manager logs. For nspawn this includes all console
5586 * "loginctl session-status" without further argument will now
5587 show the status of the session of the caller. Similar,
5588 "lock-session", "unlock-session", "activate",
5589 "enable-linger", "disable-linger" may now be called without
5590 session/user parameter in which case they apply to the
5591 caller's session/user.
5593 * An X11 session scriptlet is now shipped that uploads
5594 $DISPLAY and $XAUTHORITY into the environment of the systemd
5595 --user daemon if a session begins. This should improve
5596 compatibility with X11 enabled applications run as systemd
5599 * Generators are now subject to masking via /etc and /run, the
5600 same way as unit files.
5602 * networkd .network files gained support for configuring
5603 per-link IPv4/IPv6 packet forwarding as well as IPv4
5604 masquerading. This is by default turned on for veth links to
5605 containers, as registered by systemd-nspawn. This means that
5606 nspawn containers run with --network-veth will now get
5607 automatic routed access to the host's networks without any
5608 further configuration or setup, as long as networkd runs on
5611 * systemd-nspawn gained the --port= (-p) switch to expose TCP
5612 or UDP posts of a container on the host. With this in place
5613 it is possible to run containers with private veth links
5614 (--network-veth), and have their functionality exposed on
5615 the host as if their services were running directly on the
5618 * systemd-nspawn's --network-veth switch now gained a short
5619 version "-n", since with the changes above it is now truly
5620 useful out-of-the-box. The systemd-nspawn@.service has been
5621 updated to make use of it too by default.
5623 * systemd-nspawn will now maintain a per-image R/W lock, to
5624 ensure that the same image is not started more than once
5625 writable. (It's OK to run an image multiple times
5626 simultaneously in read-only mode.)
5628 * systemd-nspawn's --image= option is now capable of
5629 dissecting and booting MBR and GPT disk images that contain
5630 only a single active Linux partition. Previously it
5631 supported only GPT disk images with proper GPT type
5632 IDs. This allows running cloud images from major
5633 distributions directly with systemd-nspawn, without
5636 * In addition to collecting mouse dpi data in the udev
5637 hardware database, there's now support for collecting angle
5638 information for mouse scroll wheels. The database is
5639 supposed to guarantee similar scrolling behavior on mice
5640 that it knows about. There's also support for collecting
5641 information about Touchpad types.
5643 * udev's input_id built-in will now also collect touch screen
5644 dimension data and attach it to probed devices.
5646 * /etc/os-release gained support for a Distribution Privacy
5649 * networkd gained support for creating "ipvlan", "gretap",
5650 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
5652 * systemd-tmpfiles gained support for "a" lines for setting
5655 * systemd-nspawn will now mount /tmp in the container to
5656 tmpfs, automatically.
5658 * systemd now exposes the memory.usage_in_bytes cgroup
5659 attribute and shows it for each service in the "systemctl
5660 status" output, if available.
5662 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
5663 immediate reboot is triggered. This useful if shutdown is
5664 hung and is unable to complete, to expedite the
5665 operation. Note that this kind of reboot will still unmount
5666 all file systems, and hence should not result in fsck being
5669 * A .device unit for an optical block device will now be
5670 considered active only when a medium is in the drive. Also,
5671 mount units are now bound to their backing devices thus
5672 triggering automatic unmounting when devices become
5673 unavailable. With this in place systemd will now
5674 automatically unmount left-over mounts when a CD-ROM is
5675 ejected or an USB stick is yanked from the system.
5677 * networkd-wait-online now has support for waiting for
5678 specific interfaces only (with globbing), and for giving up
5679 after a configurable timeout.
5681 * networkd now exits when idle. It will be automatically
5682 restarted as soon as interfaces show up, are removed or
5683 change state. networkd will stay around as long as there is
5684 at least one DHCP state machine or similar around, that keep
5687 * networkd may now configure IPv6 link-local addressing in
5688 addition to IPv4 link-local addressing.
5690 * The IPv6 "token" for use in SLAAC may now be configured for
5691 each .network interface in networkd.
5693 * Routes configured with networkd may now be assigned a scope
5696 * networkd's [Match] sections now support globbing and lists
5697 of multiple space-separated matches per item.
5699 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
5700 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
5701 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
5702 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
5703 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
5704 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
5705 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
5706 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
5707 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
5708 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
5709 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
5710 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
5711 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
5712 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
5713 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
5714 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
5715 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
5716 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
5717 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
5718 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
5719 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
5720 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
5721 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
5722 Hoffmann, Zbigniew Jędrzejewski-Szmek
5724 — Berlin, 2015-02-16
5728 * When querying unit file enablement status (for example via
5729 "systemctl is-enabled"), a new state "indirect" is now known
5730 which indicates that a unit might not be enabled itself, but
5731 another unit listed in its Also= setting might be.
5733 * Similar to the various existing ConditionXYZ= settings for
5734 units, there are now matching AssertXYZ= settings. While
5735 failing conditions cause a unit to be skipped, but its job
5736 to succeed, failing assertions declared like this will cause
5737 a unit start operation and its job to fail.
5739 * hostnamed now knows a new chassis type "embedded".
5741 * systemctl gained a new "edit" command. When used on a unit
5742 file, this allows extending unit files with .d/ drop-in
5743 configuration snippets or editing the full file (after
5744 copying it from /usr/lib to /etc). This will invoke the
5745 user's editor (as configured with $EDITOR), and reload the
5746 modified configuration after editing.
5748 * "systemctl status" now shows the suggested enablement state
5749 for a unit, as declared in the (usually vendor-supplied)
5750 system preset files.
5752 * nss-myhostname will now resolve the single-label host name
5753 "gateway" to the locally configured default IP routing
5754 gateways, ordered by their metrics. This assigns a stable
5755 name to the used gateways, regardless which ones are
5756 currently configured. Note that the name will only be
5757 resolved after all other name sources (if nss-myhostname is
5758 configured properly) and should hence not negatively impact
5759 systems that use the single-label host name "gateway" in
5762 * systemd-inhibit now allows filtering by mode when listing
5765 * Scope and service units gained a new "Delegate" boolean
5766 property, which, when set, allows processes running inside the
5767 unit to further partition resources. This is primarily
5768 useful for systemd user instances as well as container
5771 * journald will now pick up audit messages directly from
5772 the kernel, and log them like any other log message. The
5773 audit fields are split up and fully indexed. This means that
5774 journalctl in many ways is now a (nicer!) alternative to
5775 ausearch, the traditional audit client. Note that this
5776 implements only a minimal audit client. If you want the
5777 special audit modes like reboot-on-log-overflow, please use
5778 the traditional auditd instead, which can be used in
5779 parallel to journald.
5781 * The ConditionSecurity= unit file option now understands the
5782 special string "audit" to check whether auditing is
5785 * journalctl gained two new commands --vacuum-size= and
5786 --vacuum-time= to delete old journal files until the
5787 remaining ones take up no more than the specified size on disk,
5788 or are not older than the specified time.
5790 * A new, native PPPoE library has been added to sd-network,
5791 systemd's library of light-weight networking protocols. This
5792 library will be used in a future version of networkd to
5793 enable PPPoE communication without an external pppd daemon.
5795 * The busctl tool now understands a new "capture" verb that
5796 works similar to "monitor", but writes a packet capture
5797 trace to STDOUT that can be redirected to a file which is
5798 compatible with libcap's capture file format. This can then
5799 be loaded in Wireshark and similar tools to inspect bus
5802 * The busctl tool now understands a new "tree" verb that shows
5803 the object trees of a specific service on the bus, or of all
5806 * The busctl tool now understands a new "introspect" verb that
5807 shows all interfaces and members of objects on the bus,
5808 including their signature and values. This is particularly
5809 useful to get more information about bus objects shown by
5810 the new "busctl tree" command.
5812 * The busctl tool now understands new verbs "call",
5813 "set-property" and "get-property" for invoking bus method
5814 calls, setting and getting bus object properties in a
5817 * busctl gained a new --augment-creds= argument that controls
5818 whether the tool shall augment credential information it
5819 gets from the bus with data from /proc, in a possibly
5822 * nspawn's --link-journal= switch gained two new values
5823 "try-guest" and "try-host" that work like "guest" and
5824 "host", but do not fail if the host has no persistent
5825 journalling enabled. -j is now equivalent to
5826 --link-journal=try-guest.
5828 * macvlan network devices created by nspawn will now have
5829 stable MAC addresses.
5831 * A new SmackProcessLabel= unit setting has been added, which
5832 controls the SMACK security label processes forked off by
5833 the respective unit shall use.
5835 * If compiled with --enable-xkbcommon, systemd-localed will
5836 verify x11 keymap settings by compiling the given keymap. It
5837 will spew out warnings if the compilation fails. This
5838 requires libxkbcommon to be installed.
5840 * When a coredump is collected, a larger number of metadata
5841 fields is now collected and included in the journal records
5842 created for it. More specifically, control group membership,
5843 environment variables, memory maps, working directory,
5844 chroot directory, /proc/$PID/status, and a list of open file
5845 descriptors is now stored in the log entry.
5847 * The udev hwdb now contains DPI information for mice. For
5850 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
5852 * All systemd programs that read standalone configuration
5853 files in /etc now also support a corresponding series of
5854 .conf.d configuration directories in /etc/, /run/,
5855 /usr/local/lib/, /usr/lib/, and (if configured with
5856 --enable-split-usr) /lib/. In particular, the following
5857 configuration files now have corresponding configuration
5858 directories: system.conf user.conf, logind.conf,
5859 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
5860 resolved.conf, timesyncd.conf, journal-remote.conf, and
5861 journal-upload.conf. Note that distributions should use the
5862 configuration directories in /usr/lib/; the directories in
5863 /etc/ are reserved for the system administrator.
5865 * systemd-rfkill will no longer take the rfkill device name
5866 into account when storing rfkill state on disk, as the name
5867 might be dynamically assigned and not stable. Instead, the
5868 ID_PATH udev variable combined with the rfkill type (wlan,
5869 bluetooth, ...) is used.
5871 * A new service systemd-machine-id-commit.service has been
5872 added. When used on systems where /etc is read-only during
5873 boot, and /etc/machine-id is not initialized (but an empty
5874 file), this service will copy the temporary machine ID
5875 created as replacement into /etc after the system is fully
5876 booted up. This is useful for systems that are freshly
5877 installed with a non-initialized machine ID, but should get
5878 a fixed machine ID for subsequent boots.
5880 * networkd's .netdev files now provide a large set of
5881 configuration parameters for VXLAN devices. Similarly, the
5882 bridge port cost parameter is now configurable in .network
5883 files. There's also new support for configuring IP source
5884 routing. networkd .link files gained support for a new
5885 OriginalName= match that is useful to match against the
5886 original interface name the kernel assigned. .network files
5887 may include MTU= and MACAddress= fields for altering the MTU
5888 and MAC address while being connected to a specific network
5891 * The LUKS logic gained supported for configuring
5892 UUID-specific key files. There's also new support for naming
5893 LUKS device from the kernel command line, using the new
5894 luks.name= argument.
5896 * Timer units may now be transiently created via the bus API
5897 (this was previously already available for scope and service
5898 units). In addition it is now possible to create multiple
5899 transient units at the same time with a single bus call. The
5900 "systemd-run" tool has been updated to make use of this for
5901 running commands on a specified time, in at(1)-style.
5903 * tmpfiles gained support for "t" lines, for assigning
5904 extended attributes to files. Among other uses this may be
5905 used to assign SMACK labels to files.
5907 Contributions from: Alin Rauta, Alison Chaiken, Andrej
5908 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
5909 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
5910 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
5911 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
5912 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
5913 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
5914 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
5915 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
5916 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
5917 Michael Biebl, Michael Chapman, Michael Marineau, Michal
5918 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
5919 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
5920 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
5921 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
5922 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
5923 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
5924 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
5926 — Berlin, 2014-12-10
5930 * journalctl gained the new options -t/--identifier= to match
5931 on the syslog identifier (aka "tag"), as well as --utc to
5932 show log timestamps in the UTC timezone. journalctl now also
5933 accepts -n/--lines=all to disable line capping in a pager.
5935 * journalctl gained a new switch, --flush, that synchronously
5936 flushes logs from /run/log/journal to /var/log/journal if
5937 persistent storage is enabled. systemd-journal-flush.service
5938 now waits until the operation is complete.
5940 * Services can notify the manager before they start a reload
5941 (by sending RELOADING=1) or shutdown (by sending
5942 STOPPING=1). This allows the manager to track and show the
5943 internal state of daemons and closes a race condition when
5944 the process is still running but has closed its D-Bus
5947 * Services with Type=oneshot do not have to have any ExecStart
5950 * User units are now loaded also from
5951 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
5952 /run/systemd/user directory that was already previously
5953 supported, but is under the control of the user.
5955 * Job timeouts (i.e. timeouts on the time a job that is
5956 queued stays in the run queue) can now optionally result in
5957 immediate reboot or power-off actions (JobTimeoutAction= and
5958 JobTimeoutRebootArgument=). This is useful on ".target"
5959 units, to limit the maximum time a target remains
5960 undispatched in the run queue, and to trigger an emergency
5961 operation in such a case. This is now used by default to
5962 turn off the system if boot-up (as defined by everything in
5963 basic.target) hangs and does not complete for at least
5964 15min. Also, if power-off or reboot hang for at least 30min
5965 an immediate power-off/reboot operation is triggered. This
5966 functionality is particularly useful to increase reliability
5967 on embedded devices, but also on laptops which might
5968 accidentally get powered on when carried in a backpack and
5969 whose boot stays stuck in a hard disk encryption passphrase
5972 * systemd-logind can be configured to also handle lid switch
5973 events even when the machine is docked or multiple displays
5974 are attached (HandleLidSwitchDocked= option).
5976 * A helper binary and a service have been added which can be
5977 used to resume from hibernation in the initramfs. A
5978 generator will parse the resume= option on the kernel
5979 command line to trigger resume.
5981 * A user console daemon systemd-consoled has been
5982 added. Currently, it is a preview, and will so far open a
5983 single terminal on each session of the user marked as
5984 Desktop=systemd-console.
5986 * Route metrics can be specified for DHCP routes added by
5989 * The SELinux context of socket-activated services can be set
5990 from the information provided by the networking stack
5991 (SELinuxContextFromNet= option).
5993 * Userspace firmware loading support has been removed and
5994 the minimum supported kernel version is thus bumped to 3.7.
5996 * Timeout for udev workers has been increased from 1 to 3
5997 minutes, but a warning will be printed after 1 minute to
5998 help diagnose kernel modules that take a long time to load.
6000 * Udev rules can now remove tags on devices with TAG-="foobar".
6002 * systemd's readahead implementation has been removed. In many
6003 circumstances it didn't give expected benefits even for
6004 rotational disk drives and was becoming less relevant in the
6005 age of SSDs. As none of the developers has been using
6006 rotating media anymore, and nobody stepped up to actively
6007 maintain this component of systemd it has now been removed.
6009 * Swap units can use Options= to specify discard options.
6010 Discard options specified for swaps in /etc/fstab are now
6013 * Docker containers are now detected as a separate type of
6016 * The Password Agent protocol gained support for queries where
6017 the user input is shown, useful e.g. for user names.
6018 systemd-ask-password gained a new --echo option to turn that
6021 * The default sysctl.d/ snippets will now set:
6023 net.core.default_qdisc = fq_codel
6025 This selects Fair Queuing Controlled Delay as the default
6026 queuing discipline for network interfaces. fq_codel helps
6027 fight the network bufferbloat problem. It is believed to be
6028 a good default with no tuning required for most workloads.
6029 Downstream distributions may override this choice. On 10Gbit
6030 servers that do not do forwarding, "fq" may perform better.
6031 Systems without a good clocksource should use "pfifo_fast".
6033 * If kdbus is enabled during build a new option BusPolicy= is
6034 available for service units, that allows locking all service
6035 processes into a stricter bus policy, in order to limit
6036 access to various bus services, or even hide most of them
6037 from the service's view entirely.
6039 * networkctl will now show the .network and .link file
6040 networkd has applied to a specific interface.
6042 * sd-login gained a new API call sd_session_get_desktop() to
6043 query which desktop environment has been selected for a
6046 * UNIX utmp support is now compile-time optional to support
6047 legacy-free systems.
6049 * systemctl gained two new commands "add-wants" and
6050 "add-requires" for pulling in units from specific targets
6053 * If the word "rescue" is specified on the kernel command line
6054 the system will now boot into rescue mode (aka
6055 rescue.target), which was previously available only by
6056 specifying "1" or "systemd.unit=rescue.target" on the kernel
6057 command line. This new kernel command line option nicely
6058 mirrors the already existing "emergency" kernel command line
6061 * New kernel command line options mount.usr=, mount.usrflags=,
6062 mount.usrfstype= have been added that match root=, rootflags=,
6063 rootfstype= but allow mounting a specific file system to
6066 * The $NOTIFY_SOCKET is now also passed to control processes of
6067 services, not only the main process.
6069 * This version reenables support for fsck's -l switch. This
6070 means at least version v2.25 of util-linux is required for
6071 operation, otherwise dead-locks on device nodes may
6072 occur. Again: you need to update util-linux to at least
6073 v2.25 when updating systemd to v217.
6075 * The "multi-seat-x" tool has been removed from systemd, as
6076 its functionality has been integrated into X servers 1.16,
6077 and the tool is hence redundant. It is recommended to update
6078 display managers invoking this tool to simply invoke X
6079 directly from now on, again.
6081 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
6082 message flag has been added for all of systemd's polkit
6083 authenticated method calls has been added. In particular this
6084 now allows optional interactive authorization via polkit for
6085 many of PID1's privileged operations such as unit file
6086 enabling and disabling.
6088 * "udevadm hwdb --update" learnt a new switch "--usr" for
6089 placing the rebuilt hardware database in /usr instead of
6090 /etc. When used only hardware database entries stored in
6091 /usr will be used, and any user database entries in /etc are
6092 ignored. This functionality is useful for vendors to ship a
6093 pre-built database on systems where local configuration is
6094 unnecessary or unlikely.
6096 * Calendar time specifications in .timer units now also
6097 understand the strings "semi-annually", "quarterly" and
6098 "minutely" as shortcuts (in addition to the preexisting
6099 "anually", "hourly", ...).
6101 * systemd-tmpfiles will now correctly create files in /dev
6102 at boot which are marked for creation only at boot. It is
6103 recommended to always create static device nodes with 'c!'
6104 and 'b!', so that they are created only at boot and not
6105 overwritten at runtime.
6107 * When the watchdog logic is used for a service (WatchdogSec=)
6108 and the watchdog timeout is hit the service will now be
6109 terminated with SIGABRT (instead of just SIGTERM), in order
6110 to make sure a proper coredump and backtrace is
6111 generated. This ensures that hanging services will result in
6112 similar coredump/backtrace behaviour as services that hit a
6115 Contributions from: Andreas Henriksson, Andrei Borzenkov,
6116 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
6117 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
6118 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
6119 Herrmann, David Sommerseth, David Strauss, Emil Renner
6120 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
6121 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
6122 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
6123 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
6124 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
6125 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
6126 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
6127 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
6128 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
6129 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
6130 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
6131 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
6132 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
6133 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
6134 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
6135 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
6138 — Berlin, 2014-10-28
6142 * timedated no longer reads NTP implementation unit names from
6143 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
6144 implementations should add a
6146 Conflicts=systemd-timesyncd.service
6148 to their unit files to take over and replace systemd's NTP
6149 default functionality.
6151 * systemd-sysusers gained a new line type "r" for configuring
6152 which UID/GID ranges to allocate system users/groups
6153 from. Lines of type "u" may now add an additional column
6154 that specifies the home directory for the system user to be
6155 created. Also, systemd-sysusers may now optionally read user
6156 information from STDIN instead of a file. This is useful for
6157 invoking it from RPM preinst scriptlets that need to create
6158 users before the first RPM file is installed since these
6159 files might need to be owned by them. A new
6160 %sysusers_create_inline RPM macro has been introduced to do
6161 just that. systemd-sysusers now updates the shadow files as
6162 well as the user/group databases, which should enhance
6163 compatibility with certain tools like grpck.
6165 * A number of bus APIs of PID 1 now optionally consult polkit to
6166 permit access for otherwise unprivileged clients under certain
6167 conditions. Note that this currently doesn't support
6168 interactive authentication yet, but this is expected to be
6169 added eventually, too.
6171 * /etc/machine-info now has new fields for configuring the
6172 deployment environment of the machine, as well as the
6173 location of the machine. hostnamectl has been updated with
6174 new command to update these fields.
6176 * systemd-timesyncd has been updated to automatically acquire
6177 NTP server information from systemd-networkd, which might
6178 have been discovered via DHCP.
6180 * systemd-resolved now includes a caching DNS stub resolver
6181 and a complete LLMNR name resolution implementation. A new
6182 NSS module "nss-resolve" has been added which can be used
6183 instead of glibc's own "nss-dns" to resolve hostnames via
6184 systemd-resolved. Hostnames, addresses and arbitrary RRs may
6185 be resolved via systemd-resolved D-Bus APIs. In contrast to
6186 the glibc internal resolver systemd-resolved is aware of
6187 multi-homed system, and keeps DNS server and caches separate
6188 and per-interface. Queries are sent simultaneously on all
6189 interfaces that have DNS servers configured, in order to
6190 properly handle VPNs and local LANs which might resolve
6191 separate sets of domain names. systemd-resolved may acquire
6192 DNS server information from systemd-networkd automatically,
6193 which in turn might have discovered them via DHCP. A tool
6194 "systemd-resolve-host" has been added that may be used to
6195 query the DNS logic in resolved. systemd-resolved implements
6196 IDNA and automatically uses IDNA or UTF-8 encoding depending
6197 on whether classic DNS or LLMNR is used as transport. In the
6198 next releases we intend to add a DNSSEC and mDNS/DNS-SD
6199 implementation to systemd-resolved.
6201 * A new NSS module nss-mymachines has been added, that
6202 automatically resolves the names of all local registered
6203 containers to their respective IP addresses.
6205 * A new client tool "networkctl" for systemd-networkd has been
6206 added. It currently is entirely passive and will query
6207 networking configuration from udev, rtnetlink and networkd,
6208 and present it to the user in a very friendly
6209 way. Eventually, we hope to extend it to become a full
6210 control utility for networkd.
6212 * .socket units gained a new DeferAcceptSec= setting that
6213 controls the kernels' TCP_DEFER_ACCEPT sockopt for
6214 TCP. Similarly, support for controlling TCP keep-alive
6215 settings has been added (KeepAliveTimeSec=,
6216 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
6217 turning off Nagle's algorithm on TCP has been added
6220 * logind learned a new session type "web", for use in projects
6221 like Cockpit which register web clients as PAM sessions.
6223 * timer units with at least one OnCalendar= setting will now
6224 be started only after time-sync.target has been
6225 reached. This way they will not elapse before the system
6226 clock has been corrected by a local NTP client or
6227 similar. This is particular useful on RTC-less embedded
6228 machines, that come up with an invalid system clock.
6230 * systemd-nspawn's --network-veth= switch should now result in
6231 stable MAC addresses for both the outer and the inner side
6234 * systemd-nspawn gained a new --volatile= switch for running
6235 container instances with /etc or /var unpopulated.
6237 * The kdbus client code has been updated to use the new Linux
6238 3.17 memfd subsystem instead of the old kdbus-specific one.
6240 * systemd-networkd's DHCP client and server now support
6241 FORCERENEW. There are also new configuration options to
6242 configure the vendor client identifier and broadcast mode
6245 * systemd will no longer inform the kernel about the current
6246 timezone, as this is necessarily incorrect and racy as the
6247 kernel has no understanding of DST and similar
6248 concepts. This hence means FAT timestamps will be always
6249 considered UTC, similar to what Android is already
6250 doing. Also, when the RTC is configured to the local time
6251 (rather than UTC) systemd will never synchronize back to it,
6252 as this might confuse Windows at a later boot.
6254 * systemd-analyze gained a new command "verify" for offline
6255 validation of unit files.
6257 * systemd-networkd gained support for a couple of additional
6258 settings for bonding networking setups. Also, the metric for
6259 statically configured routes may now be configured. For
6260 network interfaces where this is appropriate the peer IP
6261 address may now be configured.
6263 * systemd-networkd's DHCP client will no longer request
6264 broadcasting by default, as this tripped up some networks.
6265 For hardware where broadcast is required the feature should
6266 be switched back on using RequestBroadcast=yes.
6268 * systemd-networkd will now set up IPv4LL addresses (when
6269 enabled) even if DHCP is configured successfully.
6271 * udev will now default to respect network device names given
6272 by the kernel when the kernel indicates that these are
6273 predictable. This behavior can be tweaked by changing
6274 NamePolicy= in the relevant .link file.
6276 * A new library systemd-terminal has been added that
6277 implements full TTY stream parsing and rendering. This
6278 library is supposed to be used later on for implementing a
6279 full userspace VT subsystem, replacing the current kernel
6282 * A new tool systemd-journal-upload has been added to push
6283 journal data to a remote system running
6284 systemd-journal-remote.
6286 * journald will no longer forward all local data to another
6287 running syslog daemon. This change has been made because
6288 rsyslog (which appears to be the most commonly used syslog
6289 implementation these days) no longer makes use of this, and
6290 instead pulls the data out of the journal on its own. Since
6291 forwarding the messages to a non-existent syslog server is
6292 more expensive than we assumed we have now turned this
6293 off. If you run a syslog server that is not a recent rsyslog
6294 version, you have to turn this option on again
6295 (ForwardToSyslog= in journald.conf).
6297 * journald now optionally supports the LZ4 compressor for
6298 larger journal fields. This compressor should perform much
6299 better than XZ which was the previous default.
6301 * machinectl now shows the IP addresses of local containers,
6302 if it knows them, plus the interface name of the container.
6304 * A new tool "systemd-escape" has been added that makes it
6305 easy to escape strings to build unit names and similar.
6307 * sd_notify() messages may now include a new ERRNO= field
6308 which is parsed and collected by systemd and shown among the
6309 "systemctl status" output for a service.
6311 * A new component "systemd-firstboot" has been added that
6312 queries the most basic systemd information (timezone,
6313 hostname, root password) interactively on first
6314 boot. Alternatively it may also be used to provision these
6315 things offline on OS images installed into directories.
6317 * The default sysctl.d/ snippets will now set
6319 net.ipv4.conf.default.promote_secondaries=1
6321 This has the benefit of no flushing secondary IP addresses
6322 when primary addresses are removed.
6324 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
6325 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
6326 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
6327 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
6328 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
6329 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
6330 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
6331 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
6332 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
6333 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
6334 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
6335 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
6336 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
6337 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
6338 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
6340 — Berlin, 2014-08-19
6344 * A new tool systemd-sysusers has been added. This tool
6345 creates system users and groups in /etc/passwd and
6346 /etc/group, based on static declarative system user/group
6347 definitions in /usr/lib/sysusers.d/. This is useful to
6348 enable factory resets and volatile systems that boot up with
6349 an empty /etc directory, and thus need system users and
6350 groups created during early boot. systemd now also ships
6351 with two default sysusers.d/ files for the most basic
6352 users and groups systemd and the core operating system
6355 * A new tmpfiles snippet has been added that rebuilds the
6356 essential files in /etc on boot, should they be missing.
6358 * A directive for ensuring automatic clean-up of
6359 /var/cache/man/ has been removed from the default
6360 configuration. This line should now be shipped by the man
6361 implementation. The necessary change has been made to the
6362 man-db implementation. Note that you need to update your man
6363 implementation to one that ships this line, otherwise no
6364 automatic clean-up of /var/cache/man will take place.
6366 * A new condition ConditionNeedsUpdate= has been added that
6367 may conditionalize services to only run when /etc or /var
6368 are "older" than the vendor operating system resources in
6369 /usr. This is useful for reconstructing or updating /etc
6370 after an offline update of /usr or a factory reset, on the
6371 next reboot. Services that want to run once after such an
6372 update or reset should use this condition and order
6373 themselves before the new systemd-update-done.service, which
6374 will mark the two directories as fully updated. A number of
6375 service files have been added making use of this, to rebuild
6376 the udev hardware database, the journald message catalog and
6377 dynamic loader cache (ldconfig). The systemd-sysusers tool
6378 described above also makes use of this now. With this in
6379 place it is now possible to start up a minimal operating
6380 system with /etc empty cleanly. For more information on the
6381 concepts involved see this recent blog story:
6383 http://0pointer.de/blog/projects/stateless.html
6385 * A new system group "input" has been introduced, and all
6386 input device nodes get this group assigned. This is useful
6387 for system-level software to get access to input devices. It
6388 complements what is already done for "audio" and "video".
6390 * systemd-networkd learnt minimal DHCPv4 server support in
6391 addition to the existing DHCPv4 client support. It also
6392 learnt DHCPv6 client and IPv6 Router Solicitation client
6393 support. The DHCPv4 client gained support for static routes
6394 passed in from the server. Note that the [DHCPv4] section
6395 known in older systemd-networkd versions has been renamed to
6396 [DHCP] and is now also used by the DHCPv6 client. Existing
6397 .network files using settings of this section should be
6398 updated, though compatibility is maintained. Optionally, the
6399 client hostname may now be sent to the DHCP server.
6401 * networkd gained support for vxlan virtual networks as well
6402 as tun/tap and dummy devices.
6404 * networkd gained support for automatic allocation of address
6405 ranges for interfaces from a system-wide pool of
6406 addresses. This is useful for dynamically managing a large
6407 number of interfaces with a single network configuration
6408 file. In particular this is useful to easily assign
6409 appropriate IP addresses to the veth links of a large number
6410 of nspawn instances.
6412 * RPM macros for processing sysusers, sysctl and binfmt
6413 drop-in snippets at package installation time have been
6416 * The /etc/os-release file should now be placed in
6417 /usr/lib/os-release. The old location is automatically
6418 created as symlink. /usr/lib is the more appropriate
6419 location of this file, since it shall actually describe the
6420 vendor operating system shipped in /usr, and not the
6421 configuration stored in /etc.
6423 * .mount units gained a new boolean SloppyOptions= setting
6424 that maps to mount(8)'s -s option which enables permissive
6425 parsing of unknown mount options.
6427 * tmpfiles learnt a new "L+" directive which creates a symlink
6428 but (unlike "L") deletes a pre-existing file first, should
6429 it already exist and not already be the correct
6430 symlink. Similarly, "b+", "c+" and "p+" directives have been
6431 added as well, which create block and character devices, as
6432 well as fifos in the filesystem, possibly removing any
6433 pre-existing files of different types.
6435 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
6436 'argument' field (which so far specified the source to
6437 symlink/copy the files from) is now optional. If omitted the
6438 same file os copied from /usr/share/factory/ suffixed by the
6439 full destination path. This is useful for populating /etc
6440 with essential files, by copying them from vendor defaults
6441 shipped in /usr/share/factory/etc.
6443 * A new command "systemctl preset-all" has been added that
6444 applies the service preset settings to all installed unit
6445 files. A new switch --preset-mode= has been added that
6446 controls whether only enable or only disable operations
6449 * A new command "systemctl is-system-running" has been added
6450 that allows checking the overall state of the system, for
6451 example whether it is fully up and running.
6453 * When the system boots up with an empty /etc, the equivalent
6454 to "systemctl preset-all" is executed during early boot, to
6455 make sure all default services are enabled after a factory
6458 * systemd now contains a minimal preset file that enables the
6459 most basic services systemd ships by default.
6461 * Unit files' [Install] section gained a new DefaultInstance=
6462 field for defining the default instance to create if a
6463 template unit is enabled with no instance specified.
6465 * A new passive target cryptsetup-pre.target has been added
6466 that may be used by services that need to make they run and
6467 finish before the first LUKS cryptographic device is set up.
6469 * The /dev/loop-control and /dev/btrfs-control device nodes
6470 are now owned by the "disk" group by default, opening up
6471 access to this group.
6473 * systemd-coredump will now automatically generate a
6474 stack trace of all core dumps taking place on the system,
6475 based on elfutils' libdw library. This stack trace is logged
6478 * systemd-coredump may now optionally store coredumps directly
6479 on disk (in /var/lib/systemd/coredump, possibly compressed),
6480 instead of storing them unconditionally in the journal. This
6481 mode is the new default. A new configuration file
6482 /etc/systemd/coredump.conf has been added to configure this
6483 and other parameters of systemd-coredump.
6485 * coredumpctl gained a new "info" verb to show details about a
6486 specific coredump. A new switch "-1" has also been added
6487 that makes sure to only show information about the most
6488 recent entry instead of all entries. Also, as the tool is
6489 generally useful now the "systemd-" prefix of the binary
6490 name has been removed. Distributions that want to maintain
6491 compatibility with the old name should add a symlink from
6492 the old name to the new name.
6494 * journald's SplitMode= now defaults to "uid". This makes sure
6495 that unprivileged users can access their own coredumps with
6496 coredumpctl without restrictions.
6498 * New kernel command line options "systemd.wants=" (for
6499 pulling an additional unit during boot), "systemd.mask="
6500 (for masking a specific unit for the boot), and
6501 "systemd.debug-shell" (for enabling the debug shell on tty9)
6502 have been added. This is implemented in the new generator
6503 "systemd-debug-generator".
6505 * systemd-nspawn will now by default filter a couple of
6506 syscalls for containers, among them those required for
6507 kernel module loading, direct x86 IO port access, swap
6508 management, and kexec. Most importantly though
6509 open_by_handle_at() is now prohibited for containers,
6510 closing a hole similar to a recently discussed vulnerability
6511 in docker regarding access to files on file hierarchies the
6512 container should normally not have access to. Note that, for
6513 nspawn, we generally make no security claims anyway (and
6514 this is explicitly documented in the man page), so this is
6515 just a fix for one of the most obvious problems.
6517 * A new man page file-hierarchy(7) has been added that
6518 contains a minimized, modernized version of the file system
6519 layout systemd expects, similar in style to the FHS
6520 specification or hier(5). A new tool systemd-path(1) has
6521 been added to query many of these paths for the local
6524 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
6525 longer done. Since the directory now has a per-user size
6526 limit, and is cleaned on logout this appears unnecessary,
6527 in particular since this now brings the lifecycle of this
6528 directory closer in line with how IPC objects are handled.
6530 * systemd.pc now exports a number of additional directories,
6531 including $libdir (which is useful to identify the library
6532 path for the primary architecture of the system), and a
6533 couple of drop-in directories.
6535 * udev's predictable network interface names now use the dev_port
6536 sysfs attribute, introduced in linux 3.15 instead of dev_id to
6537 distinguish between ports of the same PCI function. dev_id should
6538 only be used for ports using the same HW address, hence the need
6541 * machined has been updated to export the OS version of a
6542 container (read from /etc/os-release and
6543 /usr/lib/os-release) on the bus. This is now shown in
6544 "machinectl status" for a machine.
6546 * A new service setting RestartForceExitStatus= has been
6547 added. If configured to a set of exit signals or process
6548 return values, the service will be restarted when the main
6549 daemon process exits with any of them, regardless of the
6552 * systemctl's -H switch for connecting to remote systemd
6553 machines has been extended so that it may be used to
6554 directly connect to a specific container on the
6555 host. "systemctl -H root@foobar:waldi" will now connect as
6556 user "root" to host "foobar", and then proceed directly to
6557 the container named "waldi". Note that currently you have to
6558 authenticate as user "root" for this to work, as entering
6559 containers is a privileged operation.
6561 Contributions from: Andreas Henriksson, Benjamin Steinwender,
6562 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
6563 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
6564 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
6565 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
6566 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
6567 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
6568 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
6569 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
6570 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
6571 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
6572 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
6574 — Berlin, 2014-07-03
6578 * As an experimental feature, udev now tries to lock the
6579 disk device node (flock(LOCK_SH|LOCK_NB)) while it
6580 executes events for the disk or any of its partitions.
6581 Applications like partitioning programs can lock the
6582 disk device node (flock(LOCK_EX)) and claim temporary
6583 device ownership that way; udev will entirely skip all event
6584 handling for this disk and its partitions. If the disk
6585 was opened for writing, the close will trigger a partition
6586 table rescan in udev's "watch" facility, and if needed
6587 synthesize "change" events for the disk and all its partitions.
6588 This is now unconditionally enabled, and if it turns out to
6589 cause major problems, we might turn it on only for specific
6590 devices, or might need to disable it entirely. Device Mapper
6591 devices are excluded from this logic.
6593 * We temporarily dropped the "-l" switch for fsck invocations,
6594 since they collide with the flock() logic above. util-linux
6595 upstream has been changed already to avoid this conflict,
6596 and we will re-add "-l" as soon as util-linux with this
6597 change has been released.
6599 * The dependency on libattr has been removed. Since a long
6600 time, the extended attribute calls have moved to glibc, and
6601 libattr is thus unnecessary.
6603 * Virtualization detection works without privileges now. This
6604 means the systemd-detect-virt binary no longer requires
6605 CAP_SYS_PTRACE file capabilities, and our daemons can run
6606 with fewer privileges.
6608 * systemd-networkd now runs under its own "systemd-network"
6609 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
6610 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
6611 loses the ability to write to files owned by root this way.
6613 * Similarly, systemd-resolved now runs under its own
6614 "systemd-resolve" user with no capabilities remaining.
6616 * Similarly, systemd-bus-proxyd now runs under its own
6617 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
6619 * systemd-networkd gained support for setting up "veth"
6620 virtual Ethernet devices for container connectivity, as well
6621 as GRE and VTI tunnels.
6623 * systemd-networkd will no longer automatically attempt to
6624 manually load kernel modules necessary for certain tunnel
6625 transports. Instead, it is assumed the kernel loads them
6626 automatically when required. This only works correctly on
6627 very new kernels. On older kernels, please consider adding
6628 the kernel modules to /etc/modules-load.d/ as a work-around.
6630 * The resolv.conf file systemd-resolved generates has been
6631 moved to /run/systemd/resolve/. If you have a symlink from
6632 /etc/resolv.conf, it might be necessary to correct it.
6634 * Two new service settings, ProtectHome= and ProtectSystem=,
6635 have been added. When enabled, they will make the user data
6636 (such as /home) inaccessible or read-only and the system
6637 (such as /usr) read-only, for specific services. This allows
6638 very light-weight per-service sandboxing to avoid
6639 modifications of user data or system files from
6640 services. These two new switches have been enabled for all
6641 of systemd's long-running services, where appropriate.
6643 * Socket units gained new SocketUser= and SocketGroup=
6644 settings to set the owner user and group of AF_UNIX sockets
6645 and FIFOs in the file system.
6647 * Socket units gained a new RemoveOnStop= setting. If enabled,
6648 all FIFOS and sockets in the file system will be removed
6649 when the specific socket unit is stopped.
6651 * Socket units gained a new Symlinks= setting. It takes a list
6652 of symlinks to create to file system sockets or FIFOs
6653 created by the specific Unix sockets. This is useful to
6654 manage symlinks to socket nodes with the same lifecycle as
6657 * The /dev/log socket and /dev/initctl FIFO have been moved to
6658 /run, and have been replaced by symlinks. This allows
6659 connecting to these facilities even if PrivateDevices=yes is
6660 used for a service (which makes /dev/log itself unavailable,
6661 but /run is left). This also has the benefit of ensuring
6662 that /dev only contains device nodes, directories and
6663 symlinks, and nothing else.
6665 * sd-daemon gained two new calls sd_pid_notify() and
6666 sd_pid_notifyf(). They are similar to sd_notify() and
6667 sd_notifyf(), but allow overriding of the source PID of
6668 notification messages if permissions permit this. This is
6669 useful to send notify messages on behalf of a different
6670 process (for example, the parent process). The
6671 systemd-notify tool has been updated to make use of this
6672 when sending messages (so that notification messages now
6673 originate from the shell script invoking systemd-notify and
6674 not the systemd-notify process itself. This should minimize
6675 a race where systemd fails to associate notification
6676 messages to services when the originating process already
6679 * A new "on-abnormal" setting for Restart= has been added. If
6680 set, it will result in automatic restarts on all "abnormal"
6681 reasons for a process to exit, which includes unclean
6682 signals, core dumps, timeouts and watchdog timeouts, but
6683 does not include clean and unclean exit codes or clean
6684 signals. Restart=on-abnormal is an alternative for
6685 Restart=on-failure for services that shall be able to
6686 terminate and avoid restarts on certain errors, by
6687 indicating so with an unclean exit code. Restart=on-failure
6688 or Restart=on-abnormal is now the recommended setting for
6689 all long-running services.
6691 * If the InaccessibleDirectories= service setting points to a
6692 mount point (or if there are any submounts contained within
6693 it), it is now attempted to completely unmount it, to make
6694 the file systems truly unavailable for the respective
6697 * The ReadOnlyDirectories= service setting and
6698 systemd-nspawn's --read-only parameter are now recursively
6699 applied to all submounts, too.
6701 * Mount units may now be created transiently via the bus APIs.
6703 * The support for SysV and LSB init scripts has been removed
6704 from the systemd daemon itself. Instead, it is now
6705 implemented as a generator that creates native systemd units
6706 from these scripts when needed. This enables us to remove a
6707 substantial amount of legacy code from PID 1, following the
6708 fact that many distributions only ship a very small number
6709 of LSB/SysV init scripts nowadays.
6711 * Privileged Xen (dom0) domains are not considered
6712 virtualization anymore by the virtualization detection
6713 logic. After all, they generally have unrestricted access to
6714 the hardware and usually are used to manage the unprivileged
6717 * systemd-tmpfiles gained a new "C" line type, for copying
6718 files or entire directories.
6720 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
6721 lines. So far, they have been non-globbing versions of the
6722 latter, and have thus been redundant. In future, it is
6723 recommended to only use "z". "m" has hence been removed
6724 from the documentation, even though it stays supported.
6726 * A tmpfiles snippet to recreate the most basic structure in
6727 /var has been added. This is enough to create the /var/run →
6728 /run symlink and create a couple of structural
6729 directories. This allows systems to boot up with an empty or
6730 volatile /var. Of course, while with this change, the core OS
6731 now is capable with dealing with a volatile /var, not all
6732 user services are ready for it. However, we hope that sooner
6733 or later, many service daemons will be changed upstream so
6734 that they are able to automatically create their necessary
6735 directories in /var at boot, should they be missing. This is
6736 the first step to allow state-less systems that only require
6737 the vendor image for /usr to boot.
6739 * systemd-nspawn has gained a new --tmpfs= switch to mount an
6740 empty tmpfs instance to a specific directory. This is
6741 particularly useful for making use of the automatic
6742 reconstruction of /var (see above), by passing --tmpfs=/var.
6744 * Access modes specified in tmpfiles snippets may now be
6745 prefixed with "~", which indicates that they shall be masked
6746 by whether the existing file or directory is currently
6747 writable, readable or executable at all. Also, if specified,
6748 the sgid/suid/sticky bits will be masked for all
6751 * A new passive target unit "network-pre.target" has been
6752 added which is useful for services that shall run before any
6753 network is configured, for example firewall scripts.
6755 * The "floppy" group that previously owned the /dev/fd*
6756 devices is no longer used. The "disk" group is now used
6757 instead. Distributions should probably deprecate usage of
6760 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
6761 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
6762 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
6763 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
6764 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
6765 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
6768 — Berlin, 2014-06-11
6772 * A new "systemd-timesyncd" daemon has been added for
6773 synchronizing the system clock across the network. It
6774 implements an SNTP client. In contrast to NTP
6775 implementations such as chrony or the NTP reference server,
6776 this only implements a client side, and does not bother with
6777 the full NTP complexity, focusing only on querying time from
6778 one remote server and synchronizing the local clock to
6779 it. Unless you intend to serve NTP to networked clients or
6780 want to connect to local hardware clocks, this simple NTP
6781 client should be more than appropriate for most
6782 installations. The daemon runs with minimal privileges, and
6783 has been hooked up with networkd to only operate when
6784 network connectivity is available. The daemon saves the
6785 current clock to disk every time a new NTP sync has been
6786 acquired, and uses this to possibly correct the system clock
6787 early at bootup, in order to accommodate for systems that
6788 lack an RTC such as the Raspberry Pi and embedded devices,
6789 and to make sure that time monotonically progresses on these
6790 systems, even if it is not always correct. To make use of
6791 this daemon, a new system user and group "systemd-timesync"
6792 needs to be created on installation of systemd.
6794 * The queue "seqnum" interface of libudev has been disabled, as
6795 it was generally incompatible with device namespacing as
6796 sequence numbers of devices go "missing" if the devices are
6797 part of a different namespace.
6799 * "systemctl list-timers" and "systemctl list-sockets" gained
6800 a --recursive switch for showing units of these types also
6801 for all local containers, similar in style to the already
6802 supported --recursive switch for "systemctl list-units".
6804 * A new RebootArgument= setting has been added for service
6805 units, which may be used to specify a kernel reboot argument
6806 to use when triggering reboots with StartLimitAction=.
6808 * A new FailureAction= setting has been added for service
6809 units which may be used to specify an operation to trigger
6810 when a service fails. This works similarly to
6811 StartLimitAction=, but unlike it, controls what is done
6812 immediately rather than only after several attempts to
6813 restart the service in question.
6815 * hostnamed got updated to also expose the kernel name,
6816 release, and version on the bus. This is useful for
6817 executing commands like hostnamectl with the -H switch.
6818 systemd-analyze makes use of this to properly display
6819 details when running non-locally.
6821 * The bootchart tool can now show cgroup information in the
6822 graphs it generates.
6824 * The CFS CPU quota cgroup attribute is now exposed for
6825 services. The new CPUQuota= switch has been added for this
6826 which takes a percentage value. Setting this will have the
6827 result that a service may never get more CPU time than the
6828 specified percentage, even if the machine is otherwise idle.
6830 * systemd-networkd learned IPIP and SIT tunnel support.
6832 * LSB init scripts exposing a dependency on $network will now
6833 get a dependency on network-online.target rather than simply
6834 network.target. This should bring LSB handling closer to
6835 what it was on SysV systems.
6837 * A new fsck.repair= kernel option has been added to control
6838 how fsck shall deal with unclean file systems at boot.
6840 * The (.ini) configuration file parser will now silently
6841 ignore sections whose name begins with "X-". This may be
6842 used to maintain application-specific extension sections in unit
6845 * machined gained a new API to query the IP addresses of
6846 registered containers. "machinectl status" has been updated
6847 to show these addresses in its output.
6849 * A new call sd_uid_get_display() has been added to the
6850 sd-login APIs for querying the "primary" session of a
6851 user. The "primary" session of the user is elected from the
6852 user's sessions and generally a graphical session is
6853 preferred over a text one.
6855 * A minimal systemd-resolved daemon has been added. It
6856 currently simply acts as a companion to systemd-networkd and
6857 manages resolv.conf based on per-interface DNS
6858 configuration, possibly supplied via DHCP. In the long run
6859 we hope to extend this into a local DNSSEC enabled DNS and
6862 * The systemd-networkd-wait-online tool is now enabled by
6863 default. It will delay network-online.target until a network
6864 connection has been configured. The tool primarily integrates
6865 with networkd, but will also make a best effort to make sense
6866 of network configuration performed in some other way.
6868 * Two new service options StartupCPUShares= and
6869 StartupBlockIOWeight= have been added that work similarly to
6870 CPUShares= and BlockIOWeight= however only apply during
6871 system startup. This is useful to prioritize certain services
6872 differently during bootup than during normal runtime.
6874 * hostnamed has been changed to prefer the statically
6875 configured hostname in /etc/hostname (unless set to
6876 'localhost' or empty) over any dynamic one supplied by
6877 dhcp. With this change, the rules for picking the hostname
6878 match more closely the rules of other configuration settings
6879 where the local administrator's configuration in /etc always
6880 overrides any other settings.
6882 Contributions from: Ali H. Caliskan, Alison Chaiken, Bas van
6883 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
6884 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
6885 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
6886 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
6887 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
6888 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
6889 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
6890 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
6891 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
6892 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
6893 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
6894 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
6895 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
6896 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
6897 Lindskog, WaLyong Cho, Will Woods, Zbigniew
6900 — Beijing, 2014-05-28
6904 * When restoring the screen brightness at boot, stay away from
6905 the darkest setting or from the lowest 5% of the available
6906 range, depending on which is the larger value of both. This
6907 should effectively protect the user from rebooting into a
6908 black screen, should the brightness have been set to minimum
6911 * sd-login gained a new sd_machine_get_class() call to
6912 determine the class ("vm" or "container") of a machine
6913 registered with machined.
6915 * sd-login gained new calls
6916 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
6917 to query the identity of the peer of a local AF_UNIX
6918 connection. They operate similarly to their sd_pid_get_xyz()
6921 * PID 1 will now maintain a system-wide system state engine
6922 with the states "starting", "running", "degraded",
6923 "maintenance", "stopping". These states are bound to system
6924 startup, normal runtime, runtime with at least one failed
6925 service, rescue/emergency mode and system shutdown. This
6926 state is shown in the "systemctl status" output when no unit
6927 name is passed. It is useful to determine system state, in
6928 particularly when doing so for many systems or containers at
6931 * A new command "list-machines" has been added to "systemctl"
6932 that lists all local OS containers and shows their system
6933 state (see above), if systemd runs inside of them.
6935 * systemctl gained a new "-r" switch to recursively enumerate
6936 units on all local containers, when used with the
6937 "list-unit" command (which is the default one that is
6938 executed when no parameters are specified).
6940 * The GPT automatic partition discovery logic will now honour
6941 two GPT partition flags: one may be set on a partition to
6942 cause it to be mounted read-only, and the other may be set
6943 on a partition to ignore it during automatic discovery.
6945 * Two new GPT type UUIDs have been added for automatic root
6946 partition discovery, for 32-bit and 64-bit ARM. This is not
6947 particularly useful for discovering the root directory on
6948 these architectures during bare-metal boots (since UEFI is
6949 not common there), but still very useful to allow booting of
6950 ARM disk images in nspawn with the -i option.
6952 * MAC addresses of interfaces created with nspawn's
6953 --network-interface= switch will now be generated from the
6954 machine name, and thus be stable between multiple invocations
6957 * logind will now automatically remove all IPC objects owned
6958 by a user if she or he fully logs out. This makes sure that
6959 users who are logged out cannot continue to consume IPC
6960 resources. This covers SysV memory, semaphores and message
6961 queues as well as POSIX shared memory and message
6962 queues. Traditionally, SysV and POSIX IPC had no lifecycle
6963 limits. With this functionality, that is corrected. This may
6964 be turned off by using the RemoveIPC= switch of logind.conf.
6966 * The systemd-machine-id-setup and tmpfiles tools gained a
6967 --root= switch to operate on a specific root directory,
6970 * journald can now forward logged messages to the TTYs of all
6971 logged in users ("wall"). This is the default for all
6972 emergency messages now.
6974 * A new tool systemd-journal-remote has been added to stream
6975 journal log messages across the network.
6977 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
6978 controller trees are mounted into it. Note that the
6979 directories mounted beneath it are not read-only. This is a
6980 security measure and is particularly useful because glibc
6981 actually includes a search logic to pick any tmpfs it can
6982 find to implement shm_open() if /dev/shm is not available
6983 (which it might very well be in namespaced setups).
6985 * machinectl gained a new "poweroff" command to cleanly power
6986 down a local OS container.
6988 * The PrivateDevices= unit file setting will now also drop the
6989 CAP_MKNOD capability from the capability bound set, and
6990 imply DevicePolicy=closed.
6992 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
6993 comprehensively on all long-running systemd services where
6994 this is appropriate.
6996 * systemd-udevd will now run in a disassociated mount
6997 namespace. To mount directories from udev rules, make sure to
6998 pull in mount units via SYSTEMD_WANTS properties.
7000 * The kdbus support gained support for uploading policy into
7001 the kernel. sd-bus gained support for creating "monitoring"
7002 connections that can eavesdrop into all bus communication
7003 for debugging purposes.
7005 * Timestamps may now be specified in seconds since the UNIX
7006 epoch Jan 1st, 1970 by specifying "@" followed by the value
7009 * Native tcpwrap support in systemd has been removed. tcpwrap
7010 is old code, not really maintained anymore and has serious
7011 shortcomings, and better options such as firewalls
7012 exist. For setups that require tcpwrap usage, please
7013 consider invoking your socket-activated service via tcpd,
7014 like on traditional inetd.
7016 * A new system.conf configuration option
7017 DefaultTimerAccuracySec= has been added that controls the
7018 default AccuracySec= setting of .timer units.
7020 * Timer units gained a new WakeSystem= switch. If enabled,
7021 timers configured this way will cause the system to resume
7022 from system suspend (if the system supports that, which most
7025 * Timer units gained a new Persistent= switch. If enabled,
7026 timers configured this way will save to disk when they have
7027 been last triggered. This information is then used on next
7028 reboot to possible execute overdue timer events, that
7029 could not take place because the system was powered off.
7030 This enables simple anacron-like behaviour for timer units.
7032 * systemctl's "list-timers" will now also list the time a
7033 timer unit was last triggered in addition to the next time
7034 it will be triggered.
7036 * systemd-networkd will now assign predictable IPv4LL
7037 addresses to its local interfaces.
7039 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
7040 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
7041 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
7042 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
7043 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
7044 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
7045 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
7046 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
7049 — Berlin, 2014-03-25
7053 * A new unit file setting RestrictAddressFamilies= has been
7054 added to restrict which socket address families unit
7055 processes gain access to. This takes address family names
7056 like "AF_INET" or "AF_UNIX", and is useful to minimize the
7057 attack surface of services via exotic protocol stacks. This
7058 is built on seccomp system call filters.
7060 * Two new unit file settings RuntimeDirectory= and
7061 RuntimeDirectoryMode= have been added that may be used to
7062 manage a per-daemon runtime directories below /run. This is
7063 an alternative for setting up directory permissions with
7064 tmpfiles snippets, and has the advantage that the runtime
7065 directory's lifetime is bound to the daemon runtime and that
7066 the daemon starts up with an empty directory each time. This
7067 is particularly useful when writing services that drop
7068 privileges using the User= or Group= setting.
7070 * The DeviceAllow= unit setting now supports globbing for
7071 matching against device group names.
7073 * The systemd configuration file system.conf gained new
7074 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
7075 DefaultMemoryAccounting= to globally turn on/off accounting
7076 for specific resources (cgroups) for all units. These
7077 settings may still be overridden individually in each unit
7080 * systemd-gpt-auto-generator is now able to discover /srv and
7081 root partitions in addition to /home and swap partitions. It
7082 also supports LUKS-encrypted partitions now. With this in
7083 place, automatic discovery of partitions to mount following
7084 the Discoverable Partitions Specification
7085 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
7086 is now a lot more complete. This allows booting without
7087 /etc/fstab and without root= on the kernel command line on
7088 systems prepared appropriately.
7090 * systemd-nspawn gained a new --image= switch which allows
7091 booting up disk images and Linux installations on any block
7092 device that follow the Discoverable Partitions Specification
7093 (see above). This means that installations made with
7094 appropriately updated installers may now be started and
7095 deployed using container managers, completely
7096 unmodified. (We hope that libvirt-lxc will add support for
7097 this feature soon, too.)
7099 * systemd-nspawn gained a new --network-macvlan= setting to
7100 set up a private macvlan interface for the
7101 container. Similarly, systemd-networkd gained a new
7102 Kind=macvlan setting in .netdev files.
7104 * systemd-networkd now supports configuring local addresses
7107 * A new tool systemd-network-wait-online has been added to
7108 synchronously wait for network connectivity using
7111 * The sd-bus.h bus API gained a new sd_bus_track object for
7112 tracking the lifecycle of bus peers. Note that sd-bus.h is
7113 still not a public API though (unless you specify
7114 --enable-kdbus on the configure command line, which however
7115 voids your warranty and you get no API stability guarantee).
7117 * The $XDG_RUNTIME_DIR runtime directories for each user are
7118 now individual tmpfs instances, which has the benefit of
7119 introducing separate pools for each user, with individual
7120 size limits, and thus making sure that unprivileged clients
7121 can no longer negatively impact the system or other users by
7122 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
7123 RuntimeDirectorySize= has been introduced that allows
7124 controlling the default size limit for all users. It
7125 defaults to 10% of the available physical memory. This is no
7126 replacement for quotas on tmpfs though (which the kernel
7127 still does not support), as /dev/shm and /tmp are still
7128 shared resources used by both the system and unprivileged
7131 * logind will now automatically turn off automatic suspending
7132 on laptop lid close when more than one display is
7133 connected. This was previously expected to be implemented
7134 individually in desktop environments (such as GNOME),
7135 however has been added to logind now, in order to fix a
7136 boot-time race where a desktop environment might not have
7137 been started yet and thus not been able to take an inhibitor
7138 lock at the time where logind already suspends the system
7139 due to a closed lid.
7141 * logind will now wait at least 30s after each system
7142 suspend/resume cycle, and 3min after system boot before
7143 suspending the system due to a closed laptop lid. This
7144 should give USB docking stations and similar enough time to
7145 be probed and configured after system resume and boot in
7146 order to then act as suspend blocker.
7148 * systemd-run gained a new --property= setting which allows
7149 initialization of resource control properties (and others)
7150 for the created scope or service unit. Example: "systemd-run
7151 --property=BlockIOWeight=10 updatedb" may be used to run
7152 updatedb at a low block IO scheduling weight.
7154 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
7155 now also work in --scope mode.
7157 * When systemd is compiled with kdbus support, basic support
7158 for enforced policies is now in place. (Note that enabling
7159 kdbus still voids your warranty and no API compatibility
7162 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
7163 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
7164 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
7165 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
7166 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
7167 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
7168 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
7169 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
7170 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
7171 Zbigniew Jędrzejewski-Szmek
7173 — Berlin, 2014-03-12
7177 * systemd will now relabel /dev after loading the SMACK policy
7178 according to SMACK rules.
7180 * A new unit file option AppArmorProfile= has been added to
7181 set the AppArmor profile for the processes of a unit.
7183 * A new condition check ConditionArchitecture= has been added
7184 to conditionalize units based on the system architecture, as
7185 reported by uname()'s "machine" field.
7187 * systemd-networkd now supports matching on the system
7188 virtualization, architecture, kernel command line, host name
7191 * logind is now a lot more aggressive when suspending the
7192 machine due to a closed laptop lid. Instead of acting only
7193 on the lid close action, it will continuously watch the lid
7194 status and act on it. This is useful for laptops where the
7195 power button is on the outside of the chassis so that it can
7196 be reached without opening the lid (such as the Lenovo
7197 Yoga). On those machines, logind will now immediately
7198 re-suspend the machine if the power button has been
7199 accidentally pressed while the laptop was suspended and in a
7200 backpack or similar.
7202 * logind will now watch SW_DOCK switches and inhibit reaction
7203 to the lid switch if it is pressed. This means that logind
7204 will not suspend the machine anymore if the lid is closed
7205 and the system is docked, if the laptop supports SW_DOCK
7206 notifications via the input layer. Note that ACPI docking
7207 stations do not generate this currently. Also note that this
7208 logic is usually not fully sufficient and Desktop
7209 Environments should take a lid switch inhibitor lock when an
7210 external display is connected, as systemd will not watch
7213 * nspawn will now make use of the devices cgroup controller by
7214 default, and only permit creation of and access to the usual
7215 API device nodes like /dev/null or /dev/random, as well as
7216 access to (but not creation of) the pty devices.
7218 * We will now ship a default .network file for
7219 systemd-networkd that automatically configures DHCP for
7220 network interfaces created by nspawn's --network-veth or
7221 --network-bridge= switches.
7223 * systemd will now understand the usual M, K, G, T suffixes
7224 according to SI conventions (i.e. to the base 1000) when
7225 referring to throughput and hardware metrics. It will stay
7226 with IEC conventions (i.e. to the base 1024) for software
7227 metrics, according to what is customary according to
7228 Wikipedia. We explicitly document which base applies for
7229 each configuration option.
7231 * The DeviceAllow= setting in unit files now supports a syntax
7232 to whitelist an entire group of devices node majors at once,
7233 based on the /proc/devices listing. For example, with the
7234 string "char-pts", it is now possible to whitelist all
7235 current and future pseudo-TTYs at once.
7237 * sd-event learned a new "post" event source. Event sources of
7238 this type are triggered by the dispatching of any event
7239 source of a type that is not "post". This is useful for
7240 implementing clean-up and check event sources that are
7241 triggered by other work being done in the program.
7243 * systemd-networkd is no longer statically enabled, but uses
7244 the usual [Install] sections so that it can be
7245 enabled/disabled using systemctl. It still is enabled by
7248 * When creating a veth interface pair with systemd-nspawn, the
7249 host side will now be prefixed with "vb-" if
7250 --network-bridge= is used, and with "ve-" if --network-veth
7251 is used. This way, it is easy to distinguish these cases on
7252 the host, for example to apply different configuration to
7253 them with systemd-networkd.
7255 * The compatibility libraries for libsystemd-journal.so,
7256 libsystem-id128.so, libsystemd-login.so and
7257 libsystemd-daemon.so do not make use of IFUNC
7258 anymore. Instead, we now build libsystemd.so multiple times
7259 under these alternative names. This means that the footprint
7260 is drastically increased, but given that these are
7261 transitional compatibility libraries, this should not matter
7262 much. This change has been made necessary to support the ARM
7263 platform for these compatibility libraries, as the ARM
7264 toolchain is not really at the same level as the toolchain
7265 for other architectures like x86 and does not support
7266 IFUNC. Please make sure to use --enable-compat-libs only
7267 during a transitional period!
7269 * The .include syntax has been deprecated and is not documented
7270 anymore. Drop-in files in .d directories should be used instead.
7272 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
7273 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
7274 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
7275 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
7276 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
7277 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
7278 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
7279 Zbigniew Jędrzejewski-Szmek
7281 — Berlin, 2014-02-24
7285 * A new component "systemd-networkd" has been added that can
7286 be used to configure local network interfaces statically or
7287 via DHCP. It is capable of bringing up bridges, VLANs, and
7288 bonding. Currently, no hook-ups for interactive network
7289 configuration are provided. Use this for your initrd,
7290 container, embedded, or server setup if you need a simple,
7291 yet powerful, network configuration solution. This
7292 configuration subsystem is quite nifty, as it allows wildcard
7293 hotplug matching in interfaces. For example, with a single
7294 configuration snippet, you can configure that all Ethernet
7295 interfaces showing up are automatically added to a bridge,
7296 or similar. It supports link-sensing and more.
7298 * A new tool "systemd-socket-proxyd" has been added which can
7299 act as a bidirectional proxy for TCP sockets. This is
7300 useful for adding socket activation support to services that
7301 do not actually support socket activation, including virtual
7302 machines and the like.
7304 * Add a new tool to save/restore rfkill state on
7307 * Save/restore state of keyboard backlights in addition to
7308 display backlights on shutdown/boot.
7310 * udev learned a new SECLABEL{} construct to label device
7311 nodes with a specific security label when they appear. For
7312 now, only SECLABEL{selinux} is supported, but the syntax is
7313 prepared for additional security frameworks.
7315 * udev gained a new scheme to configure link-level attributes
7316 from files in /etc/systemd/network/*.link. These files can
7317 match against MAC address, device path, driver name and type,
7318 and will apply attributes like the naming policy, link speed,
7319 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
7320 address assignment policy (randomized, ...).
7322 * The configuration of network interface naming rules for
7323 "permanent interface names" has changed: a new NamePolicy=
7324 setting in the [Link] section of .link files determines the
7325 priority of possible naming schemes (onboard, slot, MAC,
7326 path). The default value of this setting is determined by
7327 /usr/lib/net/links/99-default.link. Old
7328 80-net-name-slot.rules udev configuration file has been
7329 removed, so local configuration overriding this file should
7330 be adapted to override 99-default.link instead.
7332 * When the User= switch is used in a unit file, also
7333 initialize $SHELL= based on the user database entry.
7335 * systemd no longer depends on libdbus. All communication is
7336 now done with sd-bus, systemd's low-level bus library
7339 * kdbus support has been added to PID 1 itself. When kdbus is
7340 enabled, this causes PID 1 to set up the system bus and
7341 enable support for a new ".busname" unit type that
7342 encapsulates bus name activation on kdbus. It works a little
7343 bit like ".socket" units, except for bus names. A new
7344 generator has been added that converts classic dbus1 service
7345 activation files automatically into native systemd .busname
7348 * sd-bus: add a light-weight vtable implementation that allows
7349 defining objects on the bus with a simple static const
7350 vtable array of its methods, signals and properties.
7352 * systemd will not generate or install static dbus
7353 introspection data anymore to /usr/share/dbus-1/interfaces,
7354 as the precise format of these files is unclear, and
7355 nothing makes use of it.
7357 * A proxy daemon is now provided to proxy clients connecting
7358 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
7359 compatibility with classic D-Bus.
7361 * A bus driver implementation has been added that supports the
7362 classic D-Bus bus driver calls on kdbus, also for
7363 compatibility purposes.
7365 * A new API "sd-event.h" has been added that implements a
7366 minimal event loop API built around epoll. It provides a
7367 couple of features that direct epoll usage is lacking:
7368 prioritization of events, scales to large numbers of timer
7369 events, per-event timer slack (accuracy), system-wide
7370 coalescing of timer events, exit handlers, watchdog
7371 supervision support using systemd's sd_notify() API, child
7374 * A new API "sd-rntl.h" has been added that provides an API
7375 around the route netlink interface of the kernel, similar in
7376 style to "sd-bus.h".
7378 * A new API "sd-dhcp-client.h" has been added that provides a
7379 small DHCPv4 client-side implementation. This is used by
7382 * There is a new kernel command line option
7383 "systemd.restore_state=0|1". When set to "0", none of the
7384 systemd tools will restore saved runtime state to hardware
7385 devices. More specifically, the rfkill and backlight states
7388 * The FsckPassNo= compatibility option in mount/service units
7389 has been removed. The fstab generator will now add the
7390 necessary dependencies automatically, and does not require
7391 PID1's support for that anymore.
7393 * journalctl gained a new switch, --list-boots, that lists
7394 recent boots with their times and boot IDs.
7396 * The various tools like systemctl, loginctl, timedatectl,
7397 busctl, systemd-run, ... have gained a new switch "-M" to
7398 connect to a specific, local OS container (as direct
7399 connection, without requiring SSH). This works on any
7400 container that is registered with machined, such as those
7401 created by libvirt-lxc or nspawn.
7403 * systemd-run and systemd-analyze also gained support for "-H"
7404 to connect to remote hosts via SSH. This is particularly
7405 useful for systemd-run because it enables queuing of jobs
7406 onto remote systems.
7408 * machinectl gained a new command "login" to open a getty
7409 login in any local container. This works with any container
7410 that is registered with machined (such as those created by
7411 libvirt-lxc or nspawn), and which runs systemd inside.
7413 * machinectl gained a new "reboot" command that may be used to
7414 trigger a reboot on a specific container that is registered
7415 with machined. This works on any container that runs an init
7416 system of some kind.
7418 * systemctl gained a new "list-timers" command to print a nice
7419 listing of installed timer units with the times they elapse
7422 * Alternative reboot() parameters may now be specified on the
7423 "systemctl reboot" command line and are passed to the
7424 reboot() system call.
7426 * systemctl gained a new --job-mode= switch to configure the
7427 mode to queue a job with. This is a more generic version of
7428 --fail, --irreversible, and --ignore-dependencies, which are
7429 still available but not advertised anymore.
7431 * /etc/systemd/system.conf gained new settings to configure
7432 various default timeouts of units, as well as the default
7433 start limit interval and burst. These may still be overridden
7436 * PID1 will now export on the bus profile data of the security
7437 policy upload process (such as the SELinux policy upload to
7440 * journald: when forwarding logs to the console, include
7441 timestamps (following the setting in
7442 /sys/module/printk/parameters/time).
7444 * OnCalendar= in timer units now understands the special
7445 strings "yearly" and "annually". (Both are equivalent)
7447 * The accuracy of timer units is now configurable with the new
7448 AccuracySec= setting. It defaults to 1min.
7450 * A new dependency type JoinsNamespaceOf= has been added that
7451 allows running two services within the same /tmp and network
7452 namespace, if PrivateNetwork= or PrivateTmp= are used.
7454 * A new command "cat" has been added to systemctl. It outputs
7455 the original unit file of a unit, and concatenates the
7456 contents of additional "drop-in" unit file snippets, so that
7457 the full configuration is shown.
7459 * systemctl now supports globbing on the various "list-xyz"
7460 commands, like "list-units" or "list-sockets", as well as on
7461 those commands which take multiple unit names.
7463 * journalctl's --unit= switch gained support for globbing.
7465 * All systemd daemons now make use of the watchdog logic so
7466 that systemd automatically notices when they hang.
7468 * If the $container_ttys environment variable is set,
7469 getty-generator will automatically spawn a getty for each
7470 listed tty. This is useful for container managers to request
7471 login gettys to be spawned on as many ttys as needed.
7473 * %h, %s, %U specifier support is not available anymore when
7474 used in unit files for PID 1. This is because NSS calls are
7475 not safe from PID 1. They stay available for --user
7476 instances of systemd, and as special case for the root user.
7478 * loginctl gained a new "--no-legend" switch to turn off output
7481 * The "sd-login.h" API gained three new calls:
7482 sd_session_is_remote(), sd_session_get_remote_user(),
7483 sd_session_get_remote_host() to query information about
7486 * The udev hardware database now also carries vendor/product
7487 information of SDIO devices.
7489 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
7490 determine whether watchdog notifications are requested by
7493 * Socket-activated per-connection services now include a
7494 short description of the connection parameters in the
7497 * tmpfiles gained a new "--boot" option. When this is not used,
7498 only lines where the command character is not suffixed with
7499 "!" are executed. When this option is specified, those
7500 options are executed too. This partitions tmpfiles
7501 directives into those that can be safely executed at any
7502 time, and those which should be run only at boot (for
7503 example, a line that creates /run/nologin).
7505 * A new API "sd-resolve.h" has been added which provides a simple
7506 asynchronous wrapper around glibc NSS host name resolution
7507 calls, such as getaddrinfo(). In contrast to glibc's
7508 getaddrinfo_a(), it does not use signals. In contrast to most
7509 other asynchronous name resolution libraries, this one does
7510 not reimplement DNS, but reuses NSS, so that alternate
7511 host name resolution systems continue to work, such as mDNS,
7512 LDAP, etc. This API is based on libasyncns, but it has been
7513 cleaned up for inclusion in systemd.
7515 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
7516 "sd-daemon.h" are no longer found in individual libraries
7517 libsystemd-journal.so, libsystemd-login.so,
7518 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
7519 merged them into a single library, libsystemd.so, which
7520 provides all symbols. The reason for this is cyclic
7521 dependencies, as these libraries tend to use each other's
7522 symbols. So far, we have managed to workaround that by linking
7523 a copy of a good part of our code into each of these
7524 libraries again and again, which, however, makes certain
7525 things hard to do, like sharing static variables. Also, it
7526 substantially increases footprint. With this change, there
7527 is only one library for the basic APIs systemd
7528 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
7529 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
7530 library as well, however are subject to the --enable-kdbus
7531 switch (see below). Note that "sd-dhcp-client.h" is not part
7532 of this library (this is because it only consumes, never
7533 provides, services of/to other APIs). To make the transition
7534 easy from the separate libraries to the unified one, we
7535 provide the --enable-compat-libs compile-time switch which
7536 will generate stub libraries that are compatible with the
7537 old ones but redirect all calls to the new one.
7539 * All of the kdbus logic and the new APIs "sd-bus.h",
7540 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
7541 and "sd-utf8.h" are compile-time optional via the
7542 "--enable-kdbus" switch, and they are not compiled in by
7543 default. To make use of kdbus, you have to explicitly enable
7544 the switch. Note however, that neither the kernel nor the
7545 userspace API for all of this is considered stable yet. We
7546 want to maintain the freedom to still change the APIs for
7547 now. By specifying this build-time switch, you acknowledge
7548 that you are aware of the instability of the current
7551 * Also, note that while kdbus is pretty much complete,
7552 it lacks one thing: proper policy support. This means you
7553 can build a fully working system with all features; however,
7554 it will be highly insecure. Policy support will be added in
7555 one of the next releases, at the same time that we will
7556 declare the APIs stable.
7558 * When the kernel command line argument "kdbus" is specified,
7559 systemd will automatically load the kdbus.ko kernel module. At
7560 this stage of development, it is only useful for testing kdbus
7561 and should not be used in production. Note: if "--enable-kdbus"
7562 is specified, and the kdbus.ko kernel module is available, and
7563 "kdbus" is added to the kernel command line, the entire system
7564 runs with kdbus instead of dbus-daemon, with the above mentioned
7565 problem of missing the system policy enforcement. Also a future
7566 version of kdbus.ko or a newer systemd will not be compatible with
7567 each other, and will unlikely be able to boot the machine if only
7568 one of them is updated.
7570 * systemctl gained a new "import-environment" command which
7571 uploads the caller's environment (or parts thereof) into the
7572 service manager so that it is inherited by services started
7573 by the manager. This is useful to upload variables like
7574 $DISPLAY into the user service manager.
7576 * A new PrivateDevices= switch has been added to service units
7577 which allows running a service with a namespaced /dev
7578 directory that does not contain any device nodes for
7579 physical devices. More specifically, it only includes devices
7580 such as /dev/null, /dev/urandom, and /dev/zero which are API
7583 * logind has been extended to support behaviour like VT
7584 switching on seats that do not support a VT. This makes
7585 multi-session available on seats that are not the first seat
7586 (seat0), and on systems where kernel support for VTs has
7587 been disabled at compile-time.
7589 * If a process holds a delay lock for system sleep or shutdown
7590 and fails to release it in time, we will now log its
7591 identity. This makes it easier to identify processes that
7592 cause slow suspends or power-offs.
7594 * When parsing /etc/crypttab, support for a new key-slot=
7595 option as supported by Debian is added. It allows indicating
7596 which LUKS slot to use on disk, speeding up key loading.
7598 * The sd_journal_sendv() API call has been checked and
7599 officially declared to be async-signal-safe so that it may
7600 be invoked from signal handlers for logging purposes.
7602 * Boot-time status output is now enabled automatically after a
7603 short timeout if boot does not progress, in order to give
7604 the user an indication what she or he is waiting for.
7606 * The boot-time output has been improved to show how much time
7607 remains until jobs expire.
7609 * The KillMode= switch in service units gained a new possible
7610 value "mixed". If set, and the unit is shut down, then the
7611 initial SIGTERM signal is sent only to the main daemon
7612 process, while the following SIGKILL signal is sent to
7613 all remaining processes of the service.
7615 * When a scope unit is registered, a new property "Controller"
7616 may be set. If set to a valid bus name, systemd will send a
7617 RequestStop() signal to this name when it would like to shut
7618 down the scope. This may be used to hook manager logic into
7619 the shutdown logic of scope units. Also, scope units may now
7620 be put in a special "abandoned" state, in which case the
7621 manager process which created them takes no further
7622 responsibilities for it.
7624 * When reading unit files, systemd will now verify
7625 the access mode of these files, and warn about certain
7626 suspicious combinations. This has been added to make it
7627 easier to track down packaging bugs where unit files are
7628 marked executable or world-writable.
7630 * systemd-nspawn gained a new "--setenv=" switch to set
7631 container-wide environment variables. The similar option in
7632 systemd-activate was renamed from "--environment=" to
7633 "--setenv=" for consistency.
7635 * systemd-nspawn has been updated to create a new kdbus domain
7636 for each container that is invoked, thus allowing each
7637 container to have its own set of system and user buses,
7638 independent of the host.
7640 * systemd-nspawn gained a new --drop-capability= switch to run
7641 the container with less capabilities than the default. Both
7642 --drop-capability= and --capability= now take the special
7643 string "all" for dropping or keeping all capabilities.
7645 * systemd-nspawn gained new switches for executing containers
7646 with specific SELinux labels set.
7648 * systemd-nspawn gained a new --quiet switch to not generate
7649 any additional output but the container's own console
7652 * systemd-nspawn gained a new --share-system switch to run a
7653 container without PID namespacing enabled.
7655 * systemd-nspawn gained a new --register= switch to control
7656 whether the container is registered with systemd-machined or
7657 not. This is useful for containers that do not run full
7658 OS images, but only specific apps.
7660 * systemd-nspawn gained a new --keep-unit which may be used
7661 when invoked as the only program from a service unit, and
7662 results in registration of the unit service itself in
7663 systemd-machined, instead of a newly opened scope unit.
7665 * systemd-nspawn gained a new --network-interface= switch for
7666 moving arbitrary interfaces to the container. The new
7667 --network-veth switch creates a virtual Ethernet connection
7668 between host and container. The new --network-bridge=
7669 switch then allows assigning the host side of this virtual
7670 Ethernet connection to a bridge device.
7672 * systemd-nspawn gained a new --personality= switch for
7673 setting the kernel personality for the container. This is
7674 useful when running a 32-bit container on a 64-bit host. A
7675 similar option Personality= is now also available for service
7678 * logind will now also track a "Desktop" identifier for each
7679 session which encodes the desktop environment of it. This is
7680 useful for desktop environments that want to identify
7681 multiple running sessions of itself easily.
7683 * A new SELinuxContext= setting for service units has been
7684 added that allows setting a specific SELinux execution
7685 context for a service.
7687 * Most systemd client tools will now honour $SYSTEMD_LESS for
7688 settings of the "less" pager. By default, these tools will
7689 override $LESS to allow certain operations to work, such as
7690 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
7691 influence this logic.
7693 * systemd's "seccomp" hook-up has been changed to make use of
7694 the libseccomp library instead of using its own
7695 implementation. This has benefits for portability among
7698 * For usage together with SystemCallFilter=, a new
7699 SystemCallErrorNumber= setting has been introduced that
7700 allows configuration of a system error number to be returned
7701 on filtered system calls, instead of immediately killing the
7702 process. Also, SystemCallArchitectures= has been added to
7703 limit access to system calls of a particular architecture
7704 (in order to turn off support for unused secondary
7705 architectures). There is also a global
7706 SystemCallArchitectures= setting in system.conf now to turn
7707 off support for non-native system calls system-wide.
7709 * systemd requires a kernel with a working name_to_handle_at(),
7710 please see the kernel config requirements in the README file.
7712 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
7713 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
7714 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
7715 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
7716 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
7717 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
7718 Elia Pinto, Florian Weimer, George McCollister, Goffredo
7719 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
7720 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
7721 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
7722 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
7723 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
7724 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
7725 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
7726 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
7727 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
7728 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
7729 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
7730 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
7731 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
7732 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
7733 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
7734 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
7735 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
7737 — Berlin, 2014-02-20
7741 * logind has gained support for facilitating privileged input
7742 and drm device access for unprivileged clients. This work is
7743 useful to allow Wayland display servers (and similar
7744 programs, such as kmscon) to run under the user's ID and
7745 access input and drm devices which are normally
7746 protected. When this is used (and the kernel is new enough)
7747 logind will "mute" IO on the file descriptors passed to
7748 Wayland as long as it is in the background and "unmute" it
7749 if it returns into the foreground. This allows secure
7750 session switching without allowing background sessions to
7751 eavesdrop on input and display data. This also introduces
7752 session switching support if VT support is turned off in the
7753 kernel, and on seats that are not seat0.
7755 * A new kernel command line option luks.options= is understood
7756 now which allows specifying LUKS options for usage for LUKS
7757 encrypted partitions specified with luks.uuid=.
7759 * tmpfiles.d(5) snippets may now use specifier expansion in
7760 path names. More specifically %m, %b, %H, %v, are now
7761 replaced by the local machine id, boot id, hostname, and
7762 kernel version number.
7764 * A new tmpfiles.d(5) command "m" has been introduced which
7765 may be used to change the owner/group/access mode of a file
7766 or directory if it exists, but do nothing if it does not.
7768 * This release removes high-level support for the
7769 MemorySoftLimit= cgroup setting. The underlying kernel
7770 cgroup attribute memory.soft_limit= is currently badly
7771 designed and likely to be removed from the kernel API in its
7772 current form, hence we should not expose it for now.
7774 * The memory.use_hierarchy cgroup attribute is now enabled for
7775 all cgroups systemd creates in the memory cgroup
7776 hierarchy. This option is likely to be come the built-in
7777 default in the kernel anyway, and the non-hierarchical mode
7778 never made much sense in the intrinsically hierarchical
7781 * A new field _SYSTEMD_SLICE= is logged along with all journal
7782 messages containing the slice a message was generated
7783 from. This is useful to allow easy per-customer filtering of
7784 logs among other things.
7786 * systemd-journald will no longer adjust the group of journal
7787 files it creates to the "systemd-journal" group. Instead we
7788 rely on the journal directory to be owned by the
7789 "systemd-journal" group, and its setgid bit set, so that the
7790 kernel file system layer will automatically enforce that
7791 journal files inherit this group assignment. The reason for
7792 this change is that we cannot allow NSS look-ups from
7793 journald which would be necessary to resolve
7794 "systemd-journal" to a numeric GID, because this might
7795 create deadlocks if NSS involves synchronous queries to
7796 other daemons (such as nscd, or sssd) which in turn are
7797 logging clients of journald and might block on it, which
7798 would then dead lock. A tmpfiles.d(5) snippet included in
7799 systemd will make sure the setgid bit and group are
7800 properly set on the journal directory if it exists on every
7801 boot. However, we recommend adjusting it manually after
7802 upgrades too (or from RPM scriptlets), so that the change is
7803 not delayed until next reboot.
7805 * Backlight and random seed files in /var/lib/ have moved into
7806 the /var/lib/systemd/ directory, in order to centralize all
7807 systemd generated files in one directory.
7809 * Boot time performance measurements (as displayed by
7810 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
7811 performance information if that's available to determine how
7812 much time BIOS and boot loader initialization required. With
7813 a sufficiently new BIOS you hence no longer need to boot
7814 with Gummiboot to get access to such information.
7816 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
7817 Cristian Rodríguez, Dave Reisner, David Herrmann, David
7818 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
7819 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
7820 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
7821 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
7822 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
7824 — Berlin, 2013-10-02
7828 * The Restart= option for services now understands a new
7829 on-watchdog setting, which will restart the service
7830 automatically if the service stops sending out watchdog keep
7831 alive messages (as configured with WatchdogSec=).
7833 * The getty generator (which is responsible for bringing up a
7834 getty on configured serial consoles) will no longer only
7835 start a getty on the primary kernel console but on all
7836 others, too. This makes the order in which console= is
7837 specified on the kernel command line less important.
7839 * libsystemd-logind gained a new sd_session_get_vt() call to
7840 retrieve the VT number of a session.
7842 * If the option "tries=0" is set for an entry of /etc/crypttab
7843 its passphrase is queried indefinitely instead of any
7844 maximum number of tries.
7846 * If a service with a configure PID file terminates its PID
7847 file will now be removed automatically if it still exists
7848 afterwards. This should put an end to stale PID files.
7850 * systemd-run will now also take relative binary path names
7851 for execution and no longer insists on absolute paths.
7853 * InaccessibleDirectories= and ReadOnlyDirectories= now take
7854 paths that are optionally prefixed with "-" to indicate that
7855 it should not be considered a failure if they do not exist.
7857 * journalctl -o (and similar commands) now understands a new
7858 output mode "short-precise", it is similar to "short" but
7859 shows timestamps with usec accuracy.
7861 * The option "discard" (as known from Debian) is now
7862 synonymous to "allow-discards" in /etc/crypttab. In fact,
7863 "discard" is preferred now (since it is easier to remember
7866 * Some licensing clean-ups were made, so that more code is now
7867 LGPL-2.1 licensed than before.
7869 * A minimal tool to save/restore the display backlight
7870 brightness across reboots has been added. It will store the
7871 backlight setting as late as possible at shutdown, and
7872 restore it as early as possible during reboot.
7874 * A logic to automatically discover and enable home and swap
7875 partitions on GPT disks has been added. With this in place
7876 /etc/fstab becomes optional for many setups as systemd can
7877 discover certain partitions located on the root disk
7878 automatically. Home partitions are recognized under their
7879 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
7880 partitions are recognized under their GPT type ID
7881 0657fd6da4ab43c484e50933c84b4f4f.
7883 * systemd will no longer pass any environment from the kernel
7884 or initrd to system services. If you want to set an
7885 environment for all services, do so via the kernel command
7886 line systemd.setenv= assignment.
7888 * The systemd-sysctl tool no longer natively reads the file
7889 /etc/sysctl.conf. If desired, the file should be symlinked
7890 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
7891 legacy support by a symlink rather than built-in code, it
7892 also makes the otherwise hidden order of application of the
7893 different files visible. (Note that this partly reverts to a
7894 pre-198 application order of sysctl knobs!)
7896 * The "systemctl set-log-level" and "systemctl dump" commands
7897 have been moved to systemd-analyze.
7899 * systemd-run learned the new --remain-after-exit switch,
7900 which causes the scope unit not to be cleaned up
7901 automatically after the process terminated.
7903 * tmpfiles learned a new --exclude-prefix= switch to exclude
7904 certain paths from operation.
7906 * journald will now automatically flush all messages to disk
7907 as soon as a message at the log level CRIT, ALERT or EMERG
7910 Contributions from: Andrew Cook, Brandon Philips, Christian
7911 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
7912 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
7913 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
7914 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
7915 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
7916 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
7917 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
7918 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
7919 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
7920 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
7921 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
7922 William Giokas, Zbigniew Jędrzejewski-Szmek
7924 — Berlin, 2013-09-13
7928 * The documentation has been updated to cover the various new
7929 concepts introduced with 205.
7931 * Unit files now understand the new %v specifier which
7932 resolves to the kernel version string as returned by "uname
7935 * systemctl now supports filtering the unit list output by
7936 load state, active state and sub state, using the new
7939 * "systemctl status" will now show the results of the
7940 condition checks (like ConditionPathExists= and similar) of
7941 the last start attempts of the unit. They are also logged to
7944 * "journalctl -b" may now be used to look for boot output of a
7945 specific boot. Try "journalctl -b -1" for the previous boot,
7946 but the syntax is substantially more powerful.
7948 * "journalctl --show-cursor" has been added which prints the
7949 cursor string the last shown log line. This may then be used
7950 with the new "journalctl --after-cursor=" switch to continue
7951 browsing logs from that point on.
7953 * "journalctl --force" may now be used to force regeneration
7956 * Creation of "dead" device nodes has been moved from udev
7957 into kmod and tmpfiles. Previously, udev would read the kmod
7958 databases to pre-generate dead device nodes based on meta
7959 information contained in kernel modules, so that these would
7960 be auto-loaded on access rather then at boot. As this
7961 does not really have much to do with the exposing actual
7962 kernel devices to userspace this has always been slightly
7963 alien in the udev codebase. Following the new scheme kmod
7964 will now generate a runtime snippet for tmpfiles from the
7965 module meta information and it now is tmpfiles' job to the
7966 create the nodes. This also allows overriding access and
7967 other parameters for the nodes using the usual tmpfiles
7968 facilities. As side effect this allows us to remove the
7969 CAP_SYS_MKNOD capability bit from udevd entirely.
7971 * logind's device ACLs may now be applied to these "dead"
7972 devices nodes too, thus finally allowing managed access to
7973 devices such as /dev/snd/sequencer without loading the
7974 backing module right-away.
7976 * A new RPM macro has been added that may be used to apply
7977 tmpfiles configuration during package installation.
7979 * systemd-detect-virt and ConditionVirtualization= now can
7980 detect User-Mode-Linux machines (UML).
7982 * journald will now implicitly log the effective capabilities
7983 set of processes in the message metadata.
7985 * systemd-cryptsetup has gained support for TrueCrypt volumes.
7987 * The initrd interface has been simplified (more specifically,
7988 support for passing performance data via environment
7989 variables and fsck results via files in /run has been
7990 removed). These features were non-essential, and are
7991 nowadays available in a much nicer way by having systemd in
7992 the initrd serialize its state and have the hosts systemd
7993 deserialize it again.
7995 * The udev "keymap" data files and tools to apply keyboard
7996 specific mappings of scan to key codes, and force-release
7997 scan code lists have been entirely replaced by a udev
7998 "keyboard" builtin and a hwdb data file.
8000 * systemd will now honour the kernel's "quiet" command line
8001 argument also during late shutdown, resulting in a
8002 completely silent shutdown when used.
8004 * There's now an option to control the SO_REUSEPORT socket
8005 option in .socket units.
8007 * Instance units will now automatically get a per-template
8008 subslice of system.slice unless something else is explicitly
8009 configured. For example, instances of sshd@.service will now
8010 implicitly be placed in system-sshd.slice rather than
8011 system.slice as before.
8013 * Test coverage support may now be enabled at build time.
8015 Contributions from: Dave Reisner, Frederic Crozat, Harald
8016 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
8017 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
8018 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
8019 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
8020 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
8021 Giokas, Zbigniew Jędrzejewski-Szmek
8023 — Berlin, 2013-07-23
8027 * Two new unit types have been introduced:
8029 Scope units are very similar to service units, however, are
8030 created out of pre-existing processes — instead of PID 1
8031 forking off the processes. By using scope units it is
8032 possible for system services and applications to group their
8033 own child processes (worker processes) in a powerful way
8034 which then maybe used to organize them, or kill them
8035 together, or apply resource limits on them.
8037 Slice units may be used to partition system resources in an
8038 hierarchical fashion and then assign other units to them. By
8039 default there are now three slices: system.slice (for all
8040 system services), user.slice (for all user sessions),
8041 machine.slice (for VMs and containers).
8043 Slices and scopes have been introduced primarily in
8044 context of the work to move cgroup handling to a
8045 single-writer scheme, where only PID 1
8046 creates/removes/manages cgroups.
8048 * There's a new concept of "transient" units. In contrast to
8049 normal units these units are created via an API at runtime,
8050 not from configuration from disk. More specifically this
8051 means it is now possible to run arbitrary programs as
8052 independent services, with all execution parameters passed
8053 in via bus APIs rather than read from disk. Transient units
8054 make systemd substantially more dynamic then it ever was,
8055 and useful as a general batch manager.
8057 * logind has been updated to make use of scope and slice units
8058 for managing user sessions. As a user logs in he will get
8059 his own private slice unit, to which all sessions are added
8060 as scope units. We also added support for automatically
8061 adding an instance of user@.service for the user into the
8062 slice. Effectively logind will no longer create cgroup
8063 hierarchies on its own now, it will defer entirely to PID 1
8064 for this by means of scope, service and slice units. Since
8065 user sessions this way become entities managed by PID 1
8066 the output of "systemctl" is now a lot more comprehensive.
8068 * A new mini-daemon "systemd-machined" has been added which
8069 may be used by virtualization managers to register local
8070 VMs/containers. nspawn has been updated accordingly, and
8071 libvirt will be updated shortly. machined will collect a bit
8072 of meta information about the VMs/containers, and assign
8073 them their own scope unit (see above). The collected
8074 meta-data is then made available via the "machinectl" tool,
8075 and exposed in "ps" and similar tools. machined/machinectl
8076 is compile-time optional.
8078 * As discussed earlier, the low-level cgroup configuration
8079 options ControlGroup=, ControlGroupModify=,
8080 ControlGroupPersistent=, ControlGroupAttribute= have been
8081 removed. Please use high-level attribute settings instead as
8082 well as slice units.
8084 * A new bus call SetUnitProperties() has been added to alter
8085 various runtime parameters of a unit. This is primarily
8086 useful to alter cgroup parameters dynamically in a nice way,
8087 but will be extended later on to make more properties
8088 modifiable at runtime. systemctl gained a new set-properties
8089 command that wraps this call.
8091 * A new tool "systemd-run" has been added which can be used to
8092 run arbitrary command lines as transient services or scopes,
8093 while configuring a number of settings via the command
8094 line. This tool is currently very basic, however already
8095 very useful. We plan to extend this tool to even allow
8096 queuing of execution jobs with time triggers from the
8097 command line, similar in fashion to "at".
8099 * nspawn will now inform the user explicitly that kernels with
8100 audit enabled break containers, and suggest the user to turn
8103 * Support for detecting the IMA and AppArmor security
8104 frameworks with ConditionSecurity= has been added.
8106 * journalctl gained a new "-k" switch for showing only kernel
8107 messages, mimicking dmesg output; in addition to "--user"
8108 and "--system" switches for showing only user's own logs
8111 * systemd-delta can now show information about drop-in
8112 snippets extending unit files.
8114 * libsystemd-bus has been substantially updated but is still
8115 not available as public API.
8117 * systemd will now look for the "debug" argument on the kernel
8118 command line and enable debug logging, similar to what
8119 "systemd.log_level=debug" already did before.
8121 * "systemctl set-default", "systemctl get-default" has been
8122 added to configure the default.target symlink, which
8123 controls what to boot into by default.
8125 * "systemctl set-log-level" has been added as a convenient
8126 way to raise and lower systemd logging threshold.
8128 * "systemd-analyze plot" will now show the time the various
8129 generators needed for execution, as well as information
8130 about the unit file loading.
8132 * libsystemd-journal gained a new sd_journal_open_files() call
8133 for opening specific journal files. journactl also gained a
8134 new switch to expose this new functionality. Previously we
8135 only supported opening all files from a directory, or all
8136 files from the system, as opening individual files only is
8137 racy due to journal file rotation.
8139 * systemd gained the new DefaultEnvironment= setting in
8140 /etc/systemd/system.conf to set environment variables for
8143 * If a privileged process logs a journal message with the
8144 OBJECT_PID= field set, then journald will automatically
8145 augment this with additional OBJECT_UID=, OBJECT_GID=,
8146 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
8147 system services want to log events about specific client
8148 processes. journactl/systemctl has been updated to make use
8149 of this information if all log messages regarding a specific
8152 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
8153 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
8154 Reisner, David Coppa, David King, David Strauss, Eelco
8155 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
8156 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
8157 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
8158 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
8159 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
8160 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
8161 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
8162 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
8163 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
8164 Łukasz Stelmach, 장동준
8168 * The Python bindings gained some minimal support for the APIs
8169 exposed by libsystemd-logind.
8171 * ConditionSecurity= gained support for detecting SMACK. Since
8172 this condition already supports SELinux and AppArmor we only
8173 miss IMA for this. Patches welcome!
8175 Contributions from: Karol Lewandowski, Lennart Poettering,
8176 Zbigniew Jędrzejewski-Szmek
8180 * systemd-nspawn will now create /etc/resolv.conf if
8181 necessary, before bind-mounting the host's file onto it.
8183 * systemd-nspawn will now store meta information about a
8184 container on the container's cgroup as extended attribute
8185 fields, including the root directory.
8187 * The cgroup hierarchy has been reworked in many ways. All
8188 objects any of the components systemd creates in the cgroup
8189 tree are now suffixed. More specifically, user sessions are
8190 now placed in cgroups suffixed with ".session", users in
8191 cgroups suffixed with ".user", and nspawn containers in
8192 cgroups suffixed with ".nspawn". Furthermore, all cgroup
8193 names are now escaped in a simple scheme to avoid collision
8194 of userspace object names with kernel filenames. This work
8195 is preparation for making these objects relocatable in the
8196 cgroup tree, in order to allow easy resource partitioning of
8197 these objects without causing naming conflicts.
8199 * systemctl list-dependencies gained the new switches
8200 --plain, --reverse, --after and --before.
8202 * systemd-inhibit now shows the process name of processes that
8203 have taken an inhibitor lock.
8205 * nss-myhostname will now also resolve "localhost"
8206 implicitly. This makes /etc/hosts an optional file and
8207 nicely handles that on IPv6 ::1 maps to both "localhost" and
8210 * libsystemd-logind.so gained a new call
8211 sd_get_machine_names() to enumerate running containers and
8212 VMs (currently only supported by very new libvirt and
8213 nspawn). sd_login_monitor can now be used to watch
8214 VMs/containers coming and going.
8216 * .include is not allowed recursively anymore, and only in
8217 unit files. Usually it is better to use drop-in snippets in
8218 .d/*.conf anyway, as introduced with systemd 198.
8220 * systemd-analyze gained a new "critical-chain" command that
8221 determines the slowest chain of units run during system
8222 boot-up. It is very useful for tracking down where
8223 optimizing boot time is the most beneficial.
8225 * systemd will no longer allow manipulating service paths in
8226 the name=systemd:/system cgroup tree using ControlGroup= in
8227 units. (But is still fine with it in all other dirs.)
8229 * There's a new systemd-nspawn@.service service file that may
8230 be used to easily run nspawn containers as system
8231 services. With the container's root directory in
8232 /var/lib/container/foobar it is now sufficient to run
8233 "systemctl start systemd-nspawn@foobar.service" to boot it.
8235 * systemd-cgls gained a new parameter "--machine" to list only
8236 the processes within a certain container.
8238 * ConditionSecurity= now can check for "apparmor". We still
8239 are lacking checks for SMACK and IMA for this condition
8240 check though. Patches welcome!
8242 * A new configuration file /etc/systemd/sleep.conf has been
8243 added that may be used to configure which kernel operation
8244 systemd is supposed to execute when "suspend", "hibernate"
8245 or "hybrid-sleep" is requested. This makes the new kernel
8246 "freeze" state accessible to the user.
8248 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
8249 the passed argument if applicable.
8251 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
8252 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
8253 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
8254 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
8255 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
8256 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
8257 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
8262 * The output of 'systemctl list-jobs' got some polishing. The
8263 '--type=' argument may now be passed more than once. A new
8264 command 'systemctl list-sockets' has been added which shows
8265 a list of kernel sockets systemd is listening on with the
8266 socket units they belong to, plus the units these socket
8269 * The experimental libsystemd-bus library got substantial
8270 updates to work in conjunction with the (also experimental)
8271 kdbus kernel project. It works well enough to exchange
8272 messages with some sophistication. Note that kdbus is not
8273 ready yet, and the library is mostly an elaborate test case
8274 for now, and not installable.
8276 * systemd gained a new unit 'systemd-static-nodes.service'
8277 that generates static device nodes earlier during boot, and
8278 can run in conjunction with udev.
8280 * libsystemd-login gained a new call sd_pid_get_user_unit()
8281 to retrieve the user systemd unit a process is running
8282 in. This is useful for systems where systemd is used as
8285 * systemd-nspawn now places all containers in the new /machine
8286 top-level cgroup directory in the name=systemd
8287 hierarchy. libvirt will soon do the same, so that we get a
8288 uniform separation of /system, /user and /machine for system
8289 services, user processes and containers/virtual
8290 machines. This new cgroup hierarchy is also useful to stick
8291 stable names to specific container instances, which can be
8292 recognized later this way (this name may be controlled
8293 via systemd-nspawn's new -M switch). libsystemd-login also
8294 gained a new call sd_pid_get_machine_name() to retrieve the
8295 name of the container/VM a specific process belongs to.
8297 * bootchart can now store its data in the journal.
8299 * libsystemd-journal gained a new call
8300 sd_journal_add_conjunction() for AND expressions to the
8301 matching logic. This can be used to express more complex
8302 logical expressions.
8304 * journactl can now take multiple --unit= and --user-unit=
8307 * The cryptsetup logic now understands the "luks.key=" kernel
8308 command line switch for specifying a file to read the
8309 decryption key from. Also, if a configured key file is not
8310 found the tool will now automatically fall back to prompting
8313 * Python systemd.journal module was updated to wrap recently
8314 added functions from libsystemd-journal. The interface was
8315 changed to bring the low level interface in s.j._Reader
8316 closer to the C API, and the high level interface in
8317 s.j.Reader was updated to wrap and convert all data about
8320 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
8321 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
8322 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
8323 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
8324 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
8325 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
8329 * journalctl --update-catalog now understands a new --root=
8330 option to operate on catalogs found in a different root
8333 * During shutdown after systemd has terminated all running
8334 services a final killing loop kills all remaining left-over
8335 processes. We will now print the name of these processes
8336 when we send SIGKILL to them, since this usually indicates a
8339 * If /etc/crypttab refers to password files stored on
8340 configured mount points automatic dependencies will now be
8341 generated to ensure the specific mount is established first
8342 before the key file is attempted to be read.
8344 * 'systemctl status' will now show information about the
8345 network sockets a socket unit is listening on.
8347 * 'systemctl status' will also shown information about any
8348 drop-in configuration file for units. (Drop-In configuration
8349 files in this context are files such as
8350 /etc/systemd/systemd/foobar.service.d/*.conf)
8352 * systemd-cgtop now optionally shows summed up CPU times of
8353 cgroups. Press '%' while running cgtop to switch between
8354 percentage and absolute mode. This is useful to determine
8355 which cgroups use up the most CPU time over the entire
8356 runtime of the system. systemd-cgtop has also been updated
8357 to be 'pipeable' for processing with further shell tools.
8359 * 'hostnamectl set-hostname' will now allow setting of FQDN
8362 * The formatting and parsing of time span values has been
8363 changed. The parser now understands fractional expressions
8364 such as "5.5h". The formatter will now output fractional
8365 expressions for all time spans under 1min, i.e. "5.123456s"
8366 rather than "5s 123ms 456us". For time spans under 1s
8367 millisecond values are shown, for those under 1ms
8368 microsecond values are shown. This should greatly improve
8369 all time-related output of systemd.
8371 * libsystemd-login and libsystemd-journal gained new
8372 functions for querying the poll() events mask and poll()
8373 timeout value for integration into arbitrary event
8376 * localectl gained the ability to list available X11 keymaps
8377 (models, layouts, variants, options).
8379 * 'systemd-analyze dot' gained the ability to filter for
8380 specific units via shell-style globs, to create smaller,
8381 more useful graphs. I.e. it is now possible to create simple
8382 graphs of all the dependencies between only target units, or
8383 of all units that Avahi has dependencies with.
8385 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
8386 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
8387 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
8388 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
8389 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
8390 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
8391 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
8395 * The boot-time readahead implementation for rotating media
8396 will now read the read-ahead data in multiple passes which
8397 consist of all read requests made in equidistant time
8398 intervals. This means instead of strictly reading read-ahead
8399 data in its physical order on disk we now try to find a
8400 middle ground between physical and access time order.
8402 * /etc/os-release files gained a new BUILD_ID= field for usage
8403 on operating systems that provide continuous builds of OS
8406 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
8407 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
8408 William Douglas, Zbigniew Jędrzejewski-Szmek
8412 * systemd-python gained an API exposing libsystemd-daemon.
8414 * The SMACK setup logic gained support for uploading CIPSO
8417 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
8418 ReadOnlyDirectories= and InaccessibleDirectories= has
8419 changed. The private /tmp and /var/tmp directories are now
8420 shared by all processes of a service (which means
8421 ExecStartPre= may now leave data in /tmp that ExecStart= of
8422 the same service can still access). When a service is
8423 stopped its temporary directories are immediately deleted
8424 (normal clean-up with tmpfiles is still done in addition to
8427 * By default, systemd will now set a couple of sysctl
8428 variables in the kernel: the safe sysrq options are turned
8429 on, IP route verification is turned on, and source routing
8430 disabled. The recently added hardlink and softlink
8431 protection of the kernel is turned on. These settings should
8432 be reasonably safe, and good defaults for all new systems.
8434 * The predictable network naming logic may now be turned off
8435 with a new kernel command line switch: net.ifnames=0.
8437 * A new libsystemd-bus module has been added that implements a
8438 pretty complete D-Bus client library. For details see:
8440 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
8442 * journald will now explicitly flush the journal files to disk
8443 at the latest 5min after each write. The file will then also
8444 be marked offline until the next write. This should increase
8445 reliability in case of a crash. The synchronization delay
8446 can be configured via SyncIntervalSec= in journald.conf.
8448 * There's a new remote-fs-setup.target unit that can be used
8449 to pull in specific services when at least one remote file
8450 system is to be mounted.
8452 * There are new targets timers.target and paths.target as
8453 canonical targets to pull user timer and path units in
8454 from. This complements sockets.target with a similar
8455 purpose for socket units.
8457 * libudev gained a new call udev_device_set_attribute_value()
8458 to set sysfs attributes of a device.
8460 * The udev daemon now sets the default number of worker
8461 processes executed in parallel based on the number of available
8462 CPUs instead of the amount of available RAM. This is supposed
8463 to provide a more reliable default and limit a too aggressive
8464 parallelism for setups with 1000s of devices connected.
8466 Contributions from: Auke Kok, Colin Walters, Cristian
8467 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
8468 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
8469 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
8470 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
8471 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
8472 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
8473 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
8474 Zbigniew Jędrzejewski-Szmek
8478 * Configuration of unit files may now be extended via drop-in
8479 files without having to edit/override the unit files
8480 themselves. More specifically, if the administrator wants to
8481 change one value for a service file foobar.service he can
8482 now do so by dropping in a configuration snippet into
8483 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
8484 will load all these snippets and apply them on top of the
8485 main unit configuration file, possibly extending or
8486 overriding its settings. Using these drop-in snippets is
8487 generally nicer than the two earlier options for changing
8488 unit files locally: copying the files from
8489 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
8490 them there; or creating a new file in /etc/systemd/system/
8491 that incorporates the original one via ".include". Drop-in
8492 snippets into these .d/ directories can be placed in any
8493 directory systemd looks for units in, and the usual
8494 overriding semantics between /usr/lib, /etc and /run apply
8497 * Most unit file settings which take lists of items can now be
8498 reset by assigning the empty string to them. For example,
8499 normally, settings such as Environment=FOO=BAR append a new
8500 environment variable assignment to the environment block,
8501 each time they are used. By assigning Environment= the empty
8502 string the environment block can be reset to empty. This is
8503 particularly useful with the .d/*.conf drop-in snippets
8504 mentioned above, since this adds the ability to reset list
8505 settings from vendor unit files via these drop-ins.
8507 * systemctl gained a new "list-dependencies" command for
8508 listing the dependencies of a unit recursively.
8510 * Inhibitors are now honored and listed by "systemctl
8511 suspend", "systemctl poweroff" (and similar) too, not only
8512 GNOME. These commands will also list active sessions by
8515 * Resource limits (as exposed by the various control group
8516 controllers) can now be controlled dynamically at runtime
8517 for all units. More specifically, you can now use a command
8518 like "systemctl set-cgroup-attr foobar.service cpu.shares
8519 2000" to alter the CPU shares a specific service gets. These
8520 settings are stored persistently on disk, and thus allow the
8521 administrator to easily adjust the resource usage of
8522 services with a few simple commands. This dynamic resource
8523 management logic is also available to other programs via the
8524 bus. Almost any kernel cgroup attribute and controller is
8527 * systemd-vconsole-setup will now copy all font settings to
8528 all allocated VTs, where it previously applied them only to
8531 * libsystemd-login gained the new sd_session_get_tty() API
8534 * This release drops support for a few legacy or
8535 distribution-specific LSB facility names when parsing init
8536 scripts: $x-display-manager, $mail-transfer-agent,
8537 $mail-transport-agent, $mail-transfer-agent, $smtp,
8538 $null. Also, the mail-transfer-agent.target unit backing
8539 this has been removed. Distributions which want to retain
8540 compatibility with this should carry the burden for
8541 supporting this themselves and patch support for these back
8542 in, if they really need to. Also, the facilities $syslog and
8543 $local_fs are now ignored, since systemd does not support
8544 early-boot LSB init scripts anymore, and these facilities
8545 are implied anyway for normal services. syslog.target has
8548 * There are new bus calls on PID1's Manager object for
8549 cancelling jobs, and removing snapshot units. Previously,
8550 both calls were only available on the Job and Snapshot
8553 * systemd-journal-gatewayd gained SSL support.
8555 * The various "environment" files, such as /etc/locale.conf
8556 now support continuation lines with a backslash ("\") as
8557 last character in the line, similarly in style (but different)
8558 to how this is supported in shells.
8560 * For normal user processes the _SYSTEMD_USER_UNIT= field is
8561 now implicitly appended to every log entry logged. systemctl
8562 has been updated to filter by this field when operating on a
8563 user systemd instance.
8565 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
8566 CAP_AUDIT_CONTROL capabilities to the capabilities set for
8567 the container. This makes it easier to boot unmodified
8568 Fedora systems in a container, which however still requires
8569 audit=0 to be passed on the kernel command line. Auditing in
8570 kernel and userspace is unfortunately still too broken in
8571 context of containers, hence we recommend compiling it out
8572 of the kernel or using audit=0. Hopefully this will be fixed
8573 one day for good in the kernel.
8575 * nspawn gained the new --bind= and --bind-ro= parameters to
8576 bind mount specific directories from the host into the
8579 * nspawn will now mount its own devpts file system instance
8580 into the container, in order not to leak pty devices from
8581 the host into the container.
8583 * systemd will now read the firmware boot time performance
8584 information from the EFI variables, if the used boot loader
8585 supports this, and takes it into account for boot performance
8586 analysis via "systemd-analyze". This is currently supported
8587 only in conjunction with Gummiboot, but could be supported
8588 by other boot loaders too. For details see:
8590 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
8592 * A new generator has been added that automatically mounts the
8593 EFI System Partition (ESP) to /boot, if that directory
8594 exists, is empty, and no other file system has been
8595 configured to be mounted there.
8597 * logind will now send out PrepareForSleep(false) out
8598 unconditionally, after coming back from suspend. This may be
8599 used by applications as asynchronous notification for
8600 system resume events.
8602 * "systemctl unlock-sessions" has been added, that allows
8603 unlocking the screens of all user sessions at once, similar
8604 to how "systemctl lock-sessions" already locked all users
8605 sessions. This is backed by a new D-Bus call UnlockSessions().
8607 * "loginctl seat-status" will now show the master device of a
8608 seat. (i.e. the device of a seat that needs to be around for
8609 the seat to be considered available, usually the graphics
8612 * tmpfiles gained a new "X" line type, that allows
8613 configuration of files and directories (with wildcards) that
8614 shall be excluded from automatic cleanup ("aging").
8616 * udev default rules set the device node permissions now only
8617 at "add" events, and do not change them any longer with a
8618 later "change" event.
8620 * The log messages for lid events and power/sleep keypresses
8621 now carry a message ID.
8623 * We now have a substantially larger unit test suite, but this
8624 continues to be work in progress.
8626 * udevadm hwdb gained a new --root= parameter to change the
8627 root directory to operate relative to.
8629 * logind will now issue a background sync() request to the kernel
8630 early at shutdown, so that dirty buffers are flushed to disk early
8631 instead of at the last moment, in order to optimize shutdown
8634 * A new bootctl tool has been added that is an interface for
8635 certain boot loader operations. This is currently a preview
8636 and is likely to be extended into a small mechanism daemon
8637 like timedated, localed, hostnamed, and can be used by
8638 graphical UIs to enumerate available boot options, and
8639 request boot into firmware operations.
8641 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
8642 the rest of the package. It also has been updated to work
8643 correctly in initrds.
8645 * polkit previously has been runtime optional, and is now also
8646 compile time optional via a configure switch.
8648 * systemd-analyze has been reimplemented in C. Also "systemctl
8649 dot" has moved into systemd-analyze.
8651 * "systemctl status" with no further parameters will now print
8652 the status of all active or failed units.
8654 * Operations such as "systemctl start" can now be executed
8655 with a new mode "--irreversible" which may be used to queue
8656 operations that cannot accidentally be reversed by a later
8657 job queuing. This is by default used to make shutdown
8658 requests more robust.
8660 * The Python API of systemd now gained a new module for
8661 reading journal files.
8663 * A new tool kernel-install has been added that can install
8664 kernel images according to the Boot Loader Specification:
8666 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
8668 * Boot time console output has been improved to provide
8669 animated boot time output for hanging jobs.
8671 * A new tool systemd-activate has been added which can be used
8672 to test socket activation with, directly from the command
8673 line. This should make it much easier to test and debug
8674 socket activation in daemons.
8676 * journalctl gained a new "--reverse" (or -r) option to show
8677 journal output in reverse order (i.e. newest line first).
8679 * journalctl gained a new "--pager-end" (or -e) option to jump
8680 to immediately jump to the end of the journal in the
8681 pager. This is only supported in conjunction with "less".
8683 * journalctl gained a new "--user-unit=" option, that works
8684 similarly to "--unit=" but filters for user units rather than
8687 * A number of unit files to ease adoption of systemd in
8688 initrds has been added. This moves some minimal logic from
8689 the various initrd implementations into systemd proper.
8691 * The journal files are now owned by a new group
8692 "systemd-journal", which exists specifically to allow access
8693 to the journal, and nothing else. Previously, we used the
8694 "adm" group for that, which however possibly covers more
8695 than just journal/log file access. This new group is now
8696 already used by systemd-journal-gatewayd to ensure this
8697 daemon gets access to the journal files and as little else
8698 as possible. Note that "make install" will also set FS ACLs
8699 up for /var/log/journal to give "adm" and "wheel" read
8700 access to it, in addition to "systemd-journal" which owns
8701 the journal files. We recommend that packaging scripts also
8702 add read access to "adm" + "wheel" to /var/log/journal, and
8703 all existing/future journal files. To normal users and
8704 administrators little changes, however packagers need to
8705 ensure to create the "systemd-journal" system group at
8706 package installation time.
8708 * The systemd-journal-gatewayd now runs as unprivileged user
8709 systemd-journal-gateway:systemd-journal-gateway. Packaging
8710 scripts need to create these system user/group at
8713 * timedated now exposes a new boolean property CanNTP that
8714 indicates whether a local NTP service is available or not.
8716 * systemd-detect-virt will now also detect xen PVs
8718 * The pstore file system is now mounted by default, if it is
8721 * In addition to the SELinux and IMA policies we will now also
8722 load SMACK policies at early boot.
8724 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
8725 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
8726 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
8727 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
8728 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
8729 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
8730 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
8731 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
8732 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
8733 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
8734 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
8735 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
8736 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
8737 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
8741 * Timer units now support calendar time events in addition to
8742 monotonic time events. That means you can now trigger a unit
8743 based on a calendar time specification such as "Thu,Fri
8744 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
8745 or fifth day of any month of the year 2013, given that it is
8746 a thursday or friday. This brings timer event support
8747 considerably closer to cron's capabilities. For details on
8748 the supported calendar time specification language see
8751 * udev now supports a number of different naming policies for
8752 network interfaces for predictable names, and a combination
8753 of these policies is now the default. Please see this wiki
8754 document for details:
8756 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
8758 * Auke Kok's bootchart implementation has been added to the
8759 systemd tree. It is an optional component that can graph the
8760 boot in quite some detail. It is one of the best bootchart
8761 implementations around and minimal in its code and
8764 * nss-myhostname has been integrated into the systemd source
8765 tree. nss-myhostname guarantees that the local hostname
8766 always stays resolvable via NSS. It has been a weak
8767 requirement of systemd-hostnamed since a long time, and
8768 since its code is actually trivial we decided to just
8769 include it in systemd's source tree. It can be turned off
8770 with a configure switch.
8772 * The read-ahead logic is now capable of properly detecting
8773 whether a btrfs file system is on SSD or rotating media, in
8774 order to optimize the read-ahead scheme. Previously, it was
8775 only capable of detecting this on traditional file systems
8778 * In udev, additional device properties are now read from the
8779 IAB in addition to the OUI database. Also, Bluetooth company
8780 identities are attached to the devices as well.
8782 * In service files %U may be used as specifier that is
8783 replaced by the configured user name of the service.
8785 * nspawn may now be invoked without a controlling TTY. This
8786 makes it suitable for invocation as its own service. This
8787 may be used to set up a simple containerized server system
8788 using only core OS tools.
8790 * systemd and nspawn can now accept socket file descriptors
8791 when they are started for socket activation. This enables
8792 implementation of socket activated nspawn
8793 containers. i.e. think about autospawning an entire OS image
8794 when the first SSH or HTTP connection is received. We expect
8795 that similar functionality will also be added to libvirt-lxc
8798 * journalctl will now suppress ANSI color codes when
8799 presenting log data.
8801 * systemctl will no longer show control group information for
8802 a unit if the control group is empty anyway.
8804 * logind can now automatically suspend/hibernate/shutdown the
8807 * /etc/machine-info and hostnamed now also expose the chassis
8808 type of the system. This can be used to determine whether
8809 the local system is a laptop, desktop, handset or
8810 tablet. This information may either be configured by the
8811 user/vendor or is automatically determined from ACPI and DMI
8812 information if possible.
8814 * A number of polkit actions are now bound together with "imply"
8815 rules. This should simplify creating UIs because many actions
8816 will now authenticate similar ones as well.
8818 * Unit files learnt a new condition ConditionACPower= which
8819 may be used to conditionalize a unit depending on whether an
8820 AC power source is connected or not, of whether the system
8821 is running on battery power.
8823 * systemctl gained a new "is-failed" verb that may be used in
8824 shell scripts and suchlike to check whether a specific unit
8825 is in the "failed" state.
8827 * The EnvironmentFile= setting in unit files now supports file
8828 globbing, and can hence be used to easily read a number of
8829 environment files at once.
8831 * systemd will no longer detect and recognize specific
8832 distributions. All distribution-specific #ifdeffery has been
8833 removed, systemd is now fully generic and
8834 distribution-agnostic. Effectively, not too much is lost as
8835 a lot of the code is still accessible via explicit configure
8836 switches. However, support for some distribution specific
8837 legacy configuration file formats has been dropped. We
8838 recommend distributions to simply adopt the configuration
8839 files everybody else uses now and convert the old
8840 configuration from packaging scripts. Most distributions
8841 already did that. If that's not possible or desirable,
8842 distributions are welcome to forward port the specific
8843 pieces of code locally from the git history.
8845 * When logging a message about a unit systemd will now always
8846 log the unit name in the message meta data.
8848 * localectl will now also discover system locale data that is
8849 not stored in locale archives, but directly unpacked.
8851 * logind will no longer unconditionally use framebuffer
8852 devices as seat masters, i.e. as devices that are required
8853 to be existing before a seat is considered preset. Instead,
8854 it will now look for all devices that are tagged as
8855 "seat-master" in udev. By default, framebuffer devices will
8856 be marked as such, but depending on local systems, other
8857 devices might be marked as well. This may be used to
8858 integrate graphics cards using closed source drivers (such
8859 as NVidia ones) more nicely into logind. Note however, that
8860 we recommend using the open source NVidia drivers instead,
8861 and no udev rules for the closed-source drivers will be
8862 shipped from us upstream.
8864 Contributions from: Adam Williamson, Alessandro Crismani, Auke
8865 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
8866 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
8867 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
8868 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
8869 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
8870 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
8871 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
8872 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
8873 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
8874 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
8875 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
8880 * udev gained support for loading additional device properties
8881 from an indexed database that is keyed by vendor/product IDs
8882 and similar device identifiers. For the beginning this
8883 "hwdb" is populated with data from the well-known PCI and
8884 USB database, but also includes PNP, ACPI and OID data. In
8885 the longer run this indexed database shall grow into
8886 becoming the one central database for non-essential
8887 userspace device metadata. Previously, data from the PCI/USB
8888 database was only attached to select devices, since the
8889 lookup was a relatively expensive operation due to O(n) time
8890 complexity (with n being the number of entries in the
8891 database). Since this is now O(1), we decided to add in this
8892 data for all devices where this is available, by
8893 default. Note that the indexed database needs to be rebuilt
8894 when new data files are installed. To achieve this you need
8895 to update your packaging scripts to invoke "udevadm hwdb
8896 --update" after installation of hwdb data files. For
8897 RPM-based distributions we introduced the new
8898 %udev_hwdb_update macro for this purpose.
8900 * The Journal gained support for the "Message Catalog", an
8901 indexed database to link up additional information with
8902 journal entries. For further details please check:
8904 https://www.freedesktop.org/wiki/Software/systemd/catalog
8906 The indexed message catalog database also needs to be
8907 rebuilt after installation of message catalog files. Use
8908 "journalctl --update-catalog" for this. For RPM-based
8909 distributions we introduced the %journal_catalog_update
8910 macro for this purpose.
8912 * The Python Journal bindings gained support for the standard
8913 Python logging framework.
8915 * The Journal API gained new functions for checking whether
8916 the underlying file system of a journal file is capable of
8917 properly reporting file change notifications, or whether
8918 applications that want to reflect journal changes "live"
8919 need to recheck journal files continuously in appropriate
8922 * It is now possible to set the "age" field for tmpfiles
8923 entries to 0, indicating that files matching this entry
8924 shall always be removed when the directories are cleaned up.
8926 * coredumpctl gained a new "gdb" verb which invokes gdb
8927 right-away on the selected coredump.
8929 * There's now support for "hybrid sleep" on kernels that
8930 support this, in addition to "suspend" and "hibernate". Use
8931 "systemctl hybrid-sleep" to make use of this.
8933 * logind's HandleSuspendKey= setting (and related settings)
8934 now gained support for a new "lock" setting to simply
8935 request the screen lock on all local sessions, instead of
8936 actually executing a suspend or hibernation.
8938 * systemd will now mount the EFI variables file system by
8941 * Socket units now gained support for configuration of the
8942 SMACK security label.
8944 * timedatectl will now output the time of the last and next
8945 daylight saving change.
8947 * We dropped support for various legacy and distro-specific
8948 concepts, such as insserv, early-boot SysV services
8949 (i.e. those for non-standard runlevels such as 'b' or 'S')
8950 or ArchLinux /etc/rc.conf support. We recommend the
8951 distributions who still need support this to either continue
8952 to maintain the necessary patches downstream, or find a
8953 different solution. (Talk to us if you have questions!)
8955 * Various systemd components will now bypass polkit checks for
8956 root and otherwise handle properly if polkit is not found to
8957 be around. This should fix most issues for polkit-less
8958 systems. Quite frankly this should have been this way since
8959 day one. It is absolutely our intention to make systemd work
8960 fine on polkit-less systems, and we consider it a bug if
8961 something does not work as it should if polkit is not around.
8963 * For embedded systems it is now possible to build udev and
8964 systemd without blkid and/or kmod support.
8966 * "systemctl switch-root" is now capable of switching root
8967 more than once. I.e. in addition to transitions from the
8968 initrd to the host OS it is now possible to transition to
8969 further OS images from the host. This is useful to implement
8970 offline updating tools.
8972 * Various other additions have been made to the RPM macros
8973 shipped with systemd. Use %udev_rules_update() after
8974 installing new udev rules files. %_udevhwdbdir,
8975 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
8976 %_sysctldir are now available which resolve to the right
8977 directories for packages to place various data files in.
8979 * journalctl gained the new --full switch (in addition to
8980 --all, to disable ellipsation for long messages.
8982 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
8983 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
8984 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
8985 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
8986 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
8987 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
8988 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
8989 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
8990 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
8994 * journalctl gained new --since= and --until= switches to
8995 filter by time. It also now supports nice filtering for
8996 units via --unit=/-u.
8998 * Type=oneshot services may use ExecReload= and do the
9001 * The journal daemon now supports time-based rotation and
9002 vacuuming, in addition to the usual disk-space based
9005 * The journal will now index the available field values for
9006 each field name. This enables clients to show pretty drop
9007 downs of available match values when filtering. The bash
9008 completion of journalctl has been updated
9009 accordingly. journalctl gained a new switch -F to list all
9010 values a certain field takes in the journal database.
9012 * More service events are now written as structured messages
9013 to the journal, and made recognizable via message IDs.
9015 * The timedated, localed and hostnamed mini-services which
9016 previously only provided support for changing time, locale
9017 and hostname settings from graphical DEs such as GNOME now
9018 also have a minimal (but very useful) text-based client
9019 utility each. This is probably the nicest way to changing
9020 these settings from the command line now, especially since
9021 it lists available options and is fully integrated with bash
9024 * There's now a new tool "systemd-coredumpctl" to list and
9025 extract coredumps from the journal.
9027 * We now install a README each in /var/log/ and
9028 /etc/rc.d/init.d explaining where the system logs and init
9029 scripts went. This hopefully should help folks who go to
9030 that dirs and look into the otherwise now empty void and
9031 scratch their heads.
9033 * When user-services are invoked (by systemd --user) the
9034 $MANAGERPID env var is set to the PID of systemd.
9036 * SIGRTMIN+24 when sent to a --user instance will now result
9037 in immediate termination of systemd.
9039 * gatewayd received numerous feature additions such as a
9040 "follow" mode, for live syncing and filtering.
9042 * browse.html now allows filtering and showing detailed
9043 information on specific entries. Keyboard navigation and
9044 mouse screen support has been added.
9046 * gatewayd/journalctl now supports HTML5/JSON
9047 Server-Sent-Events as output.
9049 * The SysV init script compatibility logic will now
9050 heuristically determine whether a script supports the
9051 "reload" verb, and only then make this available as
9054 * "systemctl status --follow" has been removed, use "journalctl
9057 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
9058 have been removed since they are hardly useful to be
9061 * And I'd like to take the opportunity to specifically mention
9062 Zbigniew for his great contributions. Zbigniew, you rock!
9064 Contributions from: Andrew Eikum, Christian Hesse, Colin
9065 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
9066 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
9067 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
9068 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
9069 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
9070 Jędrzejewski-Szmek, Сковорода Никита Андреевич
9074 * If /etc/vconsole.conf is non-existent or empty we will no
9075 longer load any console font or key map at boot by
9076 default. Instead the kernel defaults will be left
9077 intact. This is definitely the right thing to do, as no
9078 configuration should mean no configuration, and hard-coding
9079 font names that are different on all archs is probably a bad
9080 idea. Also, the kernel default key map and font should be
9081 good enough for most cases anyway, and mostly identical to
9082 the userspace fonts/key maps we previously overloaded them
9083 with. If distributions want to continue to default to a
9084 non-kernel font or key map they should ship a default
9085 /etc/vconsole.conf with the appropriate contents.
9087 Contributions from: Colin Walters, Daniel J Walsh, Dave
9088 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
9089 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
9093 * journalctl gained a new --cursor= switch to show entries
9094 starting from the specified location in the journal.
9096 * We now enforce a size limit on journal entry fields exported
9097 with "-o json" in journalctl. Fields larger than 4K will be
9098 assigned null. This can be turned off with --all.
9100 * An (optional) journal gateway daemon is now available as
9101 "systemd-journal-gatewayd.service". This service provides
9102 access to the journal via HTTP and JSON. This functionality
9103 will be used to implement live log synchronization in both
9104 pull and push modes, but has various other users too, such
9105 as easy log access for debugging of embedded devices. Right
9106 now it is already useful to retrieve the journal via HTTP:
9108 # systemctl start systemd-journal-gatewayd.service
9109 # wget http://localhost:19531/entries
9111 This will download the journal contents in a
9112 /var/log/messages compatible format. The same as JSON:
9114 # curl -H"Accept: application/json" http://localhost:19531/entries
9116 This service is also accessible via a web browser where a
9117 single static HTML5 app is served that uses the JSON logic
9118 to enable the user to do some basic browsing of the
9119 journal. This will be extended later on. Here's an example
9120 screenshot of this app in its current state:
9122 http://0pointer.de/public/journal-gatewayd
9124 Contributions from: Kay Sievers, Lennart Poettering, Robert
9125 Milasan, Tom Gundersen
9129 * The bash completion logic is now available for journalctl
9132 * We do not mount the "cpuset" controller anymore together with
9133 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
9134 started if no parameters are assigned to it. "cpuset" hence
9135 broke code that assumed it could create "cpu" groups and
9138 * journalctl -f will now subscribe to terminal size changes,
9139 and line break accordingly.
9141 Contributions from: Dave Reisner, Kay Sievers, Lennart
9142 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
9146 * nspawn will now create a symlink /etc/localtime in the
9147 container environment, copying the host's timezone
9148 setting. Previously this has been done via a bind mount, but
9149 since symlinks cannot be bind mounted this has now been
9150 changed to create/update the appropriate symlink.
9152 * journalctl -n's line number argument is now optional, and
9153 will default to 10 if omitted.
9155 * journald will now log the maximum size the journal files may
9156 take up on disk. This is particularly useful if the default
9157 built-in logic of determining this parameter from the file
9158 system size is used. Use "systemctl status
9159 systemd-journald.service" to see this information.
9161 * The multi-seat X wrapper tool has been stripped down. As X
9162 is now capable of enumerating graphics devices via udev in a
9163 seat-aware way the wrapper is not strictly necessary
9164 anymore. A stripped down temporary stop-gap is still shipped
9165 until the upstream display managers have been updated to
9166 fully support the new X logic. Expect this wrapper to be
9167 removed entirely in one of the next releases.
9169 * HandleSleepKey= in logind.conf has been split up into
9170 HandleSuspendKey= and HandleHibernateKey=. The old setting
9171 is not available anymore. X11 and the kernel are
9172 distinguishing between these keys and we should too. This
9173 also means the inhibition lock for these keys has been split
9176 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
9177 Poettering, Lukas Nykryn, Václav Pavlín
9181 * Whenever a unit changes state we will now log this to the
9182 journal and show along the unit's own log output in
9185 * ConditionPathIsMountPoint= can now properly detect bind
9186 mount points too. (Previously, a bind mount of one file
9187 system to another place in the same file system could not be
9188 detected as mount, since they shared struct stat's st_dev
9191 * We will now mount the cgroup controllers cpu, cpuacct,
9192 cpuset and the controllers net_cls, net_prio together by
9195 * nspawn containers will now have a virtualized boot
9196 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
9197 over with a randomized ID at container initialization). This
9198 has the effect of making "journalctl -b" do the right thing
9201 * The JSON output journal serialization has been updated not
9202 to generate "endless" list objects anymore, but rather one
9203 JSON object per line. This is more in line how most JSON
9204 parsers expect JSON objects. The new output mode
9205 "json-pretty" has been added to provide similar output, but
9206 neatly aligned for readability by humans.
9208 * We dropped all explicit sync() invocations in the shutdown
9209 code. The kernel does this implicitly anyway in the kernel
9210 reboot() syscall. halt(8)'s -n option is now a compatibility
9213 * We now support virtualized reboot() in containers, as
9214 supported by newer kernels. We will fall back to exit() if
9215 CAP_SYS_REBOOT is not available to the container. Also,
9216 nspawn makes use of this now and will actually reboot the
9217 container if the containerized OS asks for that.
9219 * journalctl will only show local log output by default
9220 now. Use --merge (-m) to show remote log output, too.
9222 * libsystemd-journal gained the new sd_journal_get_usage()
9223 call to determine the current disk usage of all journal
9224 files. This is exposed in the new "journalctl --disk-usage"
9227 * journald gained a new configuration setting SplitMode= in
9228 journald.conf which may be used to control how user journals
9229 are split off. See journald.conf(5) for details.
9231 * A new condition type ConditionFileNotEmpty= has been added.
9233 * tmpfiles' "w" lines now support file globbing, to write
9234 multiple files at once.
9236 * We added Python bindings for the journal submission
9237 APIs. More Python APIs for a number of selected APIs will
9238 likely follow. Note that we intend to add native bindings
9239 only for the Python language, as we consider it common
9240 enough to deserve bindings shipped within systemd. There are
9241 various projects outside of systemd that provide bindings
9242 for languages such as PHP or Lua.
9244 * Many conditions will now resolve specifiers such as %i. In
9245 addition, PathChanged= and related directives of .path units
9246 now support specifiers as well.
9248 * There's now a new RPM macro definition for the system preset
9251 * journald will now warn if it ca not forward a message to the
9252 syslog daemon because its socket is full.
9254 * timedated will no longer write or process /etc/timezone,
9255 except on Debian. As we do not support late mounted /usr
9256 anymore /etc/localtime always being a symlink is now safe,
9257 and hence the information in /etc/timezone is not necessary
9260 * logind will now always reserve one VT for a text getty (VT6
9261 by default). Previously if more than 6 X sessions where
9262 started they took up all the VTs with auto-spawned gettys,
9263 so that no text gettys were available anymore.
9265 * udev will now automatically inform the btrfs kernel logic
9266 about btrfs RAID components showing up. This should make
9267 simple hotplug based btrfs RAID assembly work.
9269 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
9270 (but not for its children which will stay at the kernel
9271 default). This should allow setups with a lot more listening
9274 * systemd will now always pass the configured timezone to the
9275 kernel at boot. timedated will do the same when the timezone
9278 * logind's inhibition logic has been updated. By default,
9279 logind will now handle the lid switch, the power and sleep
9280 keys all the time, even in graphical sessions. If DEs want
9281 to handle these events on their own they should take the new
9282 handle-power-key, handle-sleep-key and handle-lid-switch
9283 inhibitors during their runtime. A simple way to achieve
9284 that is to invoke the DE wrapped in an invocation of:
9286 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
9288 * Access to unit operations is now checked via SELinux taking
9289 the unit file label and client process label into account.
9291 * systemd will now notify the administrator in the journal
9292 when he over-mounts a non-empty directory.
9294 * There are new specifiers that are resolved in unit files,
9295 for the host name (%H), the machine ID (%m) and the boot ID
9298 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
9299 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
9300 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
9301 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
9302 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
9303 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
9304 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
9308 * Support for reading structured kernel messages from
9309 /dev/kmsg has now been added and is enabled by default.
9311 * Support for reading kernel messages from /proc/kmsg has now
9312 been removed. If you want kernel messages in the journal
9313 make sure to run a recent kernel (>= 3.5) that supports
9314 reading structured messages from /dev/kmsg (see
9315 above). /proc/kmsg is now exclusive property of classic
9316 syslog daemons again.
9318 * The libudev API gained the new
9319 udev_device_new_from_device_id() call.
9321 * The logic for file system namespace (ReadOnlyDirectory=,
9322 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
9323 require pivot_root() anymore. This means fewer temporary
9324 directories are created below /tmp for this feature.
9326 * nspawn containers will now see and receive all submounts
9327 made on the host OS below the root file system of the
9330 * Forward Secure Sealing is now supported for Journal files,
9331 which provide cryptographical sealing of journal files so
9332 that attackers cannot alter log history anymore without this
9333 being detectable. Lennart will soon post a blog story about
9334 this explaining it in more detail.
9336 * There are two new service settings RestartPreventExitStatus=
9337 and SuccessExitStatus= which allow configuration of exit
9338 status (exit code or signal) which will be excepted from the
9339 restart logic, resp. consider successful.
9341 * journalctl gained the new --verify switch that can be used
9342 to check the integrity of the structure of journal files and
9343 (if Forward Secure Sealing is enabled) the contents of
9346 * nspawn containers will now be run with /dev/stdin, /dev/fd/
9347 and similar symlinks pre-created. This makes running shells
9348 as container init process a lot more fun.
9350 * The fstab support can now handle PARTUUID= and PARTLABEL=
9353 * A new ConditionHost= condition has been added to match
9354 against the hostname (with globs) and machine ID. This is
9355 useful for clusters where a single OS image is used to
9356 provision a large number of hosts which shall run slightly
9357 different sets of services.
9359 * Services which hit the restart limit will now be placed in a
9362 Contributions from: Bertram Poettering, Dave Reisner, Huang
9363 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
9364 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
9368 * When running in --user mode systemd will now become a
9369 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
9370 tree a lot more organized.
9372 * A new PartOf= unit dependency type has been introduced that
9373 may be used to group services in a natural way.
9375 * "systemctl enable" may now be used to enable instances of
9378 * journalctl now prints error log levels in red, and
9379 warning/notice log levels in bright white. It also supports
9380 filtering by log level now.
9382 * cgtop gained a new -n switch (similar to top), to configure
9383 the maximum number of iterations to run for. It also gained
9384 -b, to run in batch mode (accepting no input).
9386 * The suffix ".service" may now be omitted on most systemctl
9387 command lines involving service unit names.
9389 * There's a new bus call in logind to lock all sessions, as
9390 well as a loginctl verb for it "lock-sessions".
9392 * libsystemd-logind.so gained a new call sd_journal_perror()
9393 that works similar to libc perror() but logs to the journal
9394 and encodes structured information about the error number.
9396 * /etc/crypttab entries now understand the new keyfile-size=
9399 * shutdown(8) now can send a (configurable) wall message when
9400 a shutdown is cancelled.
9402 * The mount propagation mode for the root file system will now
9403 default to "shared", which is useful to make containers work
9404 nicely out-of-the-box so that they receive new mounts from
9405 the host. This can be undone locally by running "mount
9406 --make-rprivate /" if needed.
9408 * The prefdm.service file has been removed. Distributions
9409 should maintain this unit downstream if they intend to keep
9410 it around. However, we recommend writing normal unit files
9411 for display managers instead.
9413 * Since systemd is a crucial part of the OS we will now
9414 default to a number of compiler switches that improve
9415 security (hardening) such as read-only relocations, stack
9416 protection, and suchlike.
9418 * The TimeoutSec= setting for services is now split into
9419 TimeoutStartSec= and TimeoutStopSec= to allow configuration
9420 of individual time outs for the start and the stop phase of
9423 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
9424 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
9425 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
9426 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
9427 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
9428 Gundersen, Zbigniew Jędrzejewski-Szmek
9432 * The journal and id128 C APIs are now fully documented as man
9435 * Extra safety checks have been added when transitioning from
9436 the initial RAM disk to the main system to avoid accidental
9439 * /etc/crypttab entries now understand the new keyfile-offset=
9442 * systemctl -t can now be used to filter by unit load state.
9444 * The journal C API gained the new sd_journal_wait() call to
9445 make writing synchronous journal clients easier.
9447 * journalctl gained the new -D switch to show journals from a
9450 * journalctl now displays a special marker between log
9451 messages of two different boots.
9453 * The journal is now explicitly flushed to /var via a service
9454 systemd-journal-flush.service, rather than implicitly simply
9455 by seeing /var/log/journal to be writable.
9457 * journalctl (and the journal C APIs) can now match for much
9458 more complex expressions, with alternatives and
9461 * When transitioning from the initial RAM disk to the main
9462 system we will now kill all processes in a killing spree to
9463 ensure no processes stay around by accident.
9465 * Three new specifiers may be used in unit files: %u, %h, %s
9466 resolve to the user name, user home directory resp. user
9467 shell. This is useful for running systemd user instances.
9469 * We now automatically rotate journal files if their data
9470 object hash table gets a fill level > 75%. We also size the
9471 hash table based on the configured maximum file size. This
9472 together should lower hash collisions drastically and thus
9473 speed things up a bit.
9475 * journalctl gained the new "--header" switch to introspect
9476 header data of journal files.
9478 * A new setting SystemCallFilters= has been added to services
9479 which may be used to apply blacklists or whitelists to
9480 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
9482 * nspawn gained a new --link-journal= switch (and quicker: -j)
9483 to link the container journal with the host. This makes it
9484 very easy to centralize log viewing on the host for all
9485 guests while still keeping the journal files separated.
9487 * Many bugfixes and optimizations
9489 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
9490 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
9491 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
9496 * Several tools now understand kernel command line arguments,
9497 which are only read when run in an initial RAM disk. They
9498 usually follow closely their normal counterparts, but are
9501 * There's a new tool to analyze the readahead files that are
9502 automatically generated at boot. Use:
9504 /usr/lib/systemd/systemd-readahead analyze /.readahead
9506 * We now provide an early debug shell on tty9 if this enabled. Use:
9508 systemctl enable debug-shell.service
9510 * All plymouth related units have been moved into the Plymouth
9511 package. Please make sure to upgrade your Plymouth version
9514 * systemd-tmpfiles now supports getting passed the basename of
9515 a configuration file only, in which case it will look for it
9516 in all appropriate directories automatically.
9518 * udevadm info now takes a /dev or /sys path as argument, and
9519 does the right thing. Example:
9521 udevadm info /dev/sda
9522 udevadm info /sys/class/block/sda
9524 * systemctl now prints a warning if a unit is stopped but a
9525 unit that might trigger it continues to run. Example: a
9526 service is stopped but the socket that activates it is left
9529 * "systemctl status" will now mention if the log output was
9530 shortened due to rotation since a service has been started.
9532 * The journal API now exposes functions to determine the
9533 "cutoff" times due to rotation.
9535 * journald now understands SIGUSR1 and SIGUSR2 for triggering
9536 immediately flushing of runtime logs to /var if possible,
9537 resp. for triggering immediate rotation of the journal
9540 * It is now considered an error if a service is attempted to
9541 be stopped that is not loaded.
9543 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
9545 * systemd-analyze now supports Python 3
9547 * tmpfiles now supports cleaning up directories via aging
9548 where the first level dirs are always kept around but
9549 directories beneath it automatically aged. This is enabled
9550 by prefixing the age field with '~'.
9552 * Seat objects now expose CanGraphical, CanTTY properties
9553 which is required to deal with very fast bootups where the
9554 display manager might be running before the graphics drivers
9555 completed initialization.
9557 * Seat objects now expose a State property.
9559 * We now include RPM macros for service enabling/disabling
9560 based on the preset logic. We recommend RPM based
9561 distributions to make use of these macros if possible. This
9562 makes it simpler to reuse RPM spec files across
9565 * We now make sure that the collected systemd unit name is
9566 always valid when services log to the journal via
9569 * There's a new man page kernel-command-line(7) detailing all
9570 command line options we understand.
9572 * The fstab generator may now be disabled at boot by passing
9573 fstab=0 on the kernel command line.
9575 * A new kernel command line option modules-load= is now understood
9576 to load a specific kernel module statically, early at boot.
9578 * Unit names specified on the systemctl command line are now
9579 automatically escaped as needed. Also, if file system or
9580 device paths are specified they are automatically turned
9581 into the appropriate mount or device unit names. Example:
9583 systemctl status /home
9584 systemctl status /dev/sda
9586 * The SysVConsole= configuration option has been removed from
9587 system.conf parsing.
9589 * The SysV search path is no longer exported on the D-Bus
9592 * The Names= option has been removed from unit file parsing.
9594 * There's a new man page bootup(7) detailing the boot process.
9596 * Every unit and every generator we ship with systemd now
9597 comes with full documentation. The self-explanatory boot is
9600 * A couple of services gained "systemd-" prefixes in their
9601 name if they wrap systemd code, rather than only external
9602 code. Among them fsck@.service which is now
9603 systemd-fsck@.service.
9605 * The HaveWatchdog property has been removed from the D-Bus
9608 * systemd.confirm_spawn= on the kernel command line should now
9611 * There's a new man page crypttab(5) which details all options
9612 we actually understand.
9614 * systemd-nspawn gained a new --capability= switch to pass
9615 additional capabilities to the container.
9617 * timedated will now read known NTP implementation unit names
9618 from /usr/lib/systemd/ntp-units.d/*.list,
9619 systemd-timedated-ntp.target has been removed.
9621 * journalctl gained a new switch "-b" that lists log data of
9622 the current boot only.
9624 * The notify socket is in the abstract namespace again, in
9625 order to support daemons which chroot() at start-up.
9627 * There is a new Storage= configuration option for journald
9628 which allows configuration of where log data should go. This
9629 also provides a way to disable journal logging entirely, so
9630 that data collected is only forwarded to the console, the
9631 kernel log buffer or another syslog implementation.
9633 * Many bugfixes and optimizations
9635 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
9636 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
9637 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
9638 Shawn Landden, Tom Gundersen
9642 * "systemctl help <unit>" now shows the man page if one is
9645 * Several new man pages have been added.
9647 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
9648 MaxLevelConsole= can now be specified in
9649 journald.conf. These options allow reducing the amount of
9650 data stored on disk or forwarded by the log level.
9652 * TimerSlackNSec= can now be specified in system.conf for
9653 PID1. This allows system-wide power savings.
9655 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
9656 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
9661 * logind is now capable of (optionally) handling power and
9662 sleep keys as well as the lid switch.
9664 * journalctl now understands the syntax "journalctl
9665 /usr/bin/avahi-daemon" to get all log output of a specific
9668 * CapabilityBoundingSet= in system.conf now also influences
9669 the capability bound set of usermode helpers of the kernel.
9671 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
9672 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
9673 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
9674 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
9678 * Note that we skipped 139 releases here in order to set the
9679 new version to something that is greater than both udev's
9680 and systemd's most recent version number.
9682 * udev: all udev sources are merged into the systemd source tree now.
9683 All future udev development will happen in the systemd tree. It
9684 is still fully supported to use the udev daemon and tools without
9685 systemd running, like in initramfs or other init systems. Building
9686 udev though, will require the *build* of the systemd tree, but
9687 udev can be properly *run* without systemd.
9689 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
9690 should be used to create dead device nodes as workarounds for broken
9693 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
9694 no longer supported. udev_monitor_new_from_netlink() needs to be
9695 used to subscribe to events.
9697 * udev: when udevd is started by systemd, processes which are left
9698 behind by forking them off of udev rules, are unconditionally cleaned
9699 up and killed now after the event handling has finished. Services or
9700 daemons must be started as systemd services. Services can be
9701 pulled-in by udev to get started, but they can no longer be directly
9702 forked by udev rules.
9704 * udev: the daemon binary is called systemd-udevd now and installed
9705 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
9706 to adapt to that, create symlink, or rename the binary after building
9709 * libudev no longer provides these symbols:
9710 udev_monitor_from_socket()
9711 udev_queue_get_failed_list_entry()
9712 udev_get_{dev,sys,run}_path()
9713 The versions number was bumped and symbol versioning introduced.
9715 * systemd-loginctl and systemd-journalctl have been renamed
9716 to loginctl and journalctl to match systemctl.
9718 * The config files: /etc/systemd/systemd-logind.conf and
9719 /etc/systemd/systemd-journald.conf have been renamed to
9720 logind.conf and journald.conf. Package updates should rename
9721 the files to the new names on upgrade.
9723 * For almost all files the license is now LGPL2.1+, changed
9724 from the previous GPL2.0+. Exceptions are some minor stuff
9725 of udev (which will be changed to LGPL2.1 eventually, too),
9726 and the MIT licensed sd-daemon.[ch] library that is suitable
9727 to be used as drop-in files.
9729 * systemd and logind now handle system sleep states, in
9730 particular suspending and hibernating.
9732 * logind now implements a sleep/shutdown/idle inhibiting logic
9733 suitable for a variety of uses. Soonishly Lennart will blog
9734 about this in more detail.
9736 * var-run.mount and var-lock.mount are no longer provided
9737 (which previously bind mounted these directories to their new
9738 places). Distributions which have not converted these
9739 directories to symlinks should consider stealing these files
9740 from git history and add them downstream.
9742 * We introduced the Documentation= field for units and added
9743 this to all our shipped units. This is useful to make it
9744 easier to explore the boot and the purpose of the various
9747 * All smaller setup units (such as
9748 systemd-vconsole-setup.service) now detect properly if they
9749 are run in a container and are skipped when
9750 appropriate. This guarantees an entirely noise-free boot in
9751 Linux container environments such as systemd-nspawn.
9753 * A framework for implementing offline system updates is now
9754 integrated, for details see:
9755 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
9757 * A new service type Type=idle is available now which helps us
9758 avoiding ugly interleaving of getty output and boot status
9761 * There's now a system-wide CapabilityBoundingSet= option to
9762 globally reduce the set of capabilities for the
9763 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
9764 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
9765 even CAP_NET_ADMIN system-wide for secure systems.
9767 * There are now system-wide DefaultLimitXXX= options to
9768 globally change the defaults of the various resource limits
9769 for all units started by PID 1.
9771 * Harald Hoyer's systemd test suite has been integrated into
9772 systemd which allows easy testing of systemd builds in qemu
9773 and nspawn. (This is really awesome! Ask us for details!)
9775 * The fstab parser is now implemented as generator, not inside
9778 * systemctl will now warn you if .mount units generated from
9779 /etc/fstab are out of date due to changes in fstab that
9780 have not been read by systemd yet.
9782 * systemd is now suitable for usage in initrds. Dracut has
9783 already been updated to make use of this. With this in place
9784 initrds get a slight bit faster but primarily are much
9785 easier to introspect and debug since "systemctl status" in
9786 the host system can be used to introspect initrd services,
9787 and the journal from the initrd is kept around too.
9789 * systemd-delta has been added, a tool to explore differences
9790 between user/admin configuration and vendor defaults.
9792 * PrivateTmp= now affects both /tmp and /var/tmp.
9794 * Boot time status messages are now much prettier and feature
9795 proper english language. Booting up systemd has never been
9798 * Read-ahead pack files now include the inode number of all
9799 files to pre-cache. When the inode changes the pre-caching
9800 is not attempted. This should be nicer to deal with updated
9801 packages which might result in changes of read-ahead
9804 * We now temporaritly lower the kernel's read_ahead_kb variable
9805 when collecting read-ahead data to ensure the kernel's
9806 built-in read-ahead does not add noise to our measurements
9807 of necessary blocks to pre-cache.
9809 * There's now RequiresMountsFor= to add automatic dependencies
9810 for all mounts necessary for a specific file system path.
9812 * MountAuto= and SwapAuto= have been removed from
9813 system.conf. Mounting file systems at boot has to take place
9816 * nspawn now learned a new switch --uuid= to set the machine
9817 ID on the command line.
9819 * nspawn now learned the -b switch to automatically search
9822 * vt102 is now the default TERM for serial TTYs, upgraded from
9825 * systemd-logind now works on VT-less systems.
9827 * The build tree has been reorganized. The individual
9828 components now have directories of their own.
9830 * A new condition type ConditionPathIsReadWrite= is now available.
9832 * nspawn learned the new -C switch to create cgroups for the
9833 container in other hierarchies.
9835 * We now have support for hardware watchdogs, configurable in
9838 * The scheduled shutdown logic now has a public API.
9840 * We now mount /tmp as tmpfs by default, but this can be
9841 masked and /etc/fstab can override it.
9843 * Since udisks does not make use of /media anymore we are not
9844 mounting a tmpfs on it anymore.
9846 * journalctl gained a new --local switch to only interleave
9847 locally generated journal files.
9849 * We can now load the IMA policy at boot automatically.
9851 * The GTK tools have been split off into a systemd-ui.
9853 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
9854 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
9855 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
9856 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
9857 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
9858 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
9859 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
9860 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
9861 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
9866 * This is mostly a bugfix release
9868 * Support optional initialization of the machine ID from the
9869 KVM or container configured UUID.
9871 * Support immediate reboots with "systemctl reboot -ff"
9873 * Show /etc/os-release data in systemd-analyze output
9875 * Many bugfixes for the journal, including endianness fixes and
9876 ensuring that disk space enforcement works
9878 * sd-login.h is C++ compatible again
9880 * Extend the /etc/os-release format on request of the Debian
9883 * We now refuse non-UTF8 strings used in various configuration
9884 and unit files. This is done to ensure we do not pass invalid
9885 data over D-Bus or expose it elsewhere.
9887 * Register Mimo USB Screens as suitable for automatic seat
9890 * Read SELinux client context from journal clients in a race
9893 * Reorder configuration file lookup order. /etc now always
9894 overrides /run in order to allow the administrator to always
9895 and unconditionally override vendor-supplied or
9896 automatically generated data.
9898 * The various user visible bits of the journal now have man
9899 pages. We still lack man pages for the journal API calls
9902 * We now ship all man pages in HTML format again in the
9905 Contributions from: Dave Reisner, Dirk Eibach, Frederic
9906 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
9907 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
9912 * This is mostly a bugfix release
9914 * systems lacking /etc/os-release are no longer supported.
9916 * Various functionality updates to libsystemd-login.so
9918 * Track class of PAM logins to distinguish greeters from
9921 Contributions from: Kay Sievers, Lennart Poettering, Michael
9926 * This is an important bugfix release for v41.
9928 * Building man pages is now optional which should be useful
9929 for those building systemd from git but unwilling to install
9932 * Watchdog support for supervising services is now usable. In
9933 a future release support for hardware watchdogs
9934 (i.e. /dev/watchdog) will be added building on this.
9936 * Service start rate limiting is now configurable and can be
9937 turned off per service. When a start rate limit is hit a
9938 reboot can automatically be triggered.
9940 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
9942 Contributions from: Benjamin Franzke, Bill Nottingham,
9943 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
9944 Schmidt, Michał Górny, Piotr Drąg
9948 * The systemd binary is installed /usr/lib/systemd/systemd now;
9949 An existing /sbin/init symlink needs to be adapted with the
9952 * The code that loads kernel modules has been ported to invoke
9953 libkmod directly, instead of modprobe. This means we do not
9954 support systems with module-init-tools anymore.
9956 * Watchdog support is now already useful, but still not
9959 * A new kernel command line option systemd.setenv= is
9960 understood to set system wide environment variables
9961 dynamically at boot.
9963 * We now limit the set of capabilities of systemd-journald.
9965 * We now set SIGPIPE to ignore by default, since it only is
9966 useful in shell pipelines, and has little use in general
9967 code. This can be disabled with IgnoreSIPIPE=no in unit
9970 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
9971 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
9976 * This is mostly a bugfix release
9978 * We now expose the reason why a service failed in the
9979 "Result" D-Bus property.
9981 * Rudimentary service watchdog support (will be completed over
9982 the next few releases.)
9984 * When systemd forks off in order execute some service we will
9985 now immediately changes its argv[0] to reflect which process
9986 it will execute. This is useful to minimize the time window
9987 with a generic argv[0], which makes bootcharts more useful
9989 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
9990 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
9991 Mike Kazantsev, Ray Strode
9995 * This is mostly a test release, but incorporates many
9998 * New systemd-cgtop tool to show control groups by their
10001 * Linking against libacl for ACLs is optional again. If
10002 disabled, support tracking device access for active logins
10003 goes becomes unavailable, and so does access to the user
10004 journals by the respective users.
10006 * If a group "adm" exists, journal files are automatically
10007 owned by them, thus allow members of this group full access
10008 to the system journal as well as all user journals.
10010 * The journal now stores the SELinux context of the logging
10011 client for all entries.
10013 * Add C++ inclusion guards to all public headers
10015 * New output mode "cat" in the journal to print only text
10016 messages, without any meta data like date or time.
10018 * Include tiny X server wrapper as a temporary stop-gap to
10019 teach XOrg udev display enumeration. This is used by display
10020 managers such as gdm, and will go away as soon as XOrg
10021 learned native udev hotplugging for display devices.
10023 * Add new systemd-cat tool for executing arbitrary programs
10024 with STDERR/STDOUT connected to the journal. Can also act as
10025 BSD logger replacement, and does so by default.
10027 * Optionally store all locally generated coredumps in the
10028 journal along with meta data.
10030 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
10031 writing short strings to files (for usage for /sys), and for
10032 creating symlinks, character and block device nodes.
10034 * New unit file option ControlGroupPersistent= to make cgroups
10035 persistent, following the mechanisms outlined in
10036 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
10038 * Support multiple local RTCs in a sane way
10040 * No longer monopolize IO when replaying readahead data on
10041 rotating disks, since we might starve non-file-system IO to
10042 death, since fanotify() will not see accesses done by blkid,
10045 * Do not show kernel threads in systemd-cgls anymore, unless
10046 requested with new -k switch.
10048 Contributions from: Dan Horák, Kay Sievers, Lennart
10049 Poettering, Michal Schmidt
10053 * This is mostly a test release, but incorporates many
10056 * The git repository moved to:
10057 git://anongit.freedesktop.org/systemd/systemd
10058 ssh://git.freedesktop.org/git/systemd/systemd
10060 * First release with the journal
10061 http://0pointer.de/blog/projects/the-journal.html
10063 * The journal replaces both systemd-kmsg-syslogd and
10064 systemd-stdout-bridge.
10066 * New sd_pid_get_unit() API call in libsystemd-logind
10068 * Many systemadm clean-ups
10070 * Introduce remote-fs-pre.target which is ordered before all
10071 remote mounts and may be used to start services before all
10074 * Added Mageia support
10076 * Add bash completion for systemd-loginctl
10078 * Actively monitor PID file creation for daemons which exit in
10079 the parent process before having finished writing the PID
10080 file in the daemon process. Daemons which do this need to be
10081 fixed (i.e. PID file creation must have finished before the
10082 parent exits), but we now react a bit more gracefully to them.
10084 * Add colourful boot output, mimicking the well-known output
10085 of existing distributions.
10087 * New option PassCredentials= for socket units, for
10088 compatibility with a recent kernel ABI breakage.
10090 * /etc/rc.local is now hooked in via a generator binary, and
10091 thus will no longer act as synchronization point during
10094 * systemctl list-unit-files now supports --root=.
10096 * systemd-tmpfiles now understands two new commands: z, Z for
10097 relabelling files according to the SELinux database. This is
10098 useful to apply SELinux labels to specific files in /sys,
10099 among other things.
10101 * Output of SysV services is now forwarded to both the console
10102 and the journal by default, not only just the console.
10104 * New man pages for all APIs from libsystemd-login.
10106 * The build tree got reorganized and the build system is a
10107 lot more modular allowing embedded setups to specifically
10108 select the components of systemd they are interested in.
10110 * Support for Linux systems lacking the kernel VT subsystem is
10113 * configure's --with-rootdir= got renamed to
10114 --with-rootprefix= to follow the naming used by udev and
10117 * Unless specified otherwise we will now install to /usr instead
10118 of /usr/local by default.
10120 * Processes with '@' in argv[0][0] are now excluded from the
10121 final shut-down killing spree, following the logic explained
10123 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
10125 * All processes remaining in a service cgroup when we enter
10126 the START or START_PRE states are now killed with
10127 SIGKILL. That means it is no longer possible to spawn
10128 background processes from ExecStart= lines (which was never
10129 supported anyway, and bad style).
10131 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
10132 reloading of units together.
10134 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
10135 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
10136 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
10137 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
10138 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek