1 systemd System and Service Manager
3 CHANGES WITH 244 in spe:
5 * systemd-udevd: removed the 30s timeout for killing stale workers on
6 exit. systemd-udevd now waits for workers to finish. The hard-coded
7 exit timeout of 30s was too short for some large installations, where
8 driver initialization could be prematurely interrupted during initrd
9 processing if the root file system had been mounted and init was
10 preparing to switch root. If udevd is run without systemd and workers
11 are hanging while udevd receives an exit signal, udevd will now exit
12 when udev.event_timeout is reached for the last hanging worker. With
13 systemd, the exit timeout can additionally be configured using
14 TimeoutStopSec= in systemd-udevd.service.
16 * Support for the cpuset cgroups v2 controller has been added.
17 Processes may be restricted to specific CPUs using the new
18 AllowedCPUs= setting, and to specific memory NUMA nodes using the new
19 AllowedMemoryNodes= setting.
21 * The signal used in restart jobs (as opposed to e.g. stop jobs) may
22 now be configured using a new RestartKillSignal= settting. This
23 allows units which signals to request termination to implement
24 different behaviour when stopping in preparation for a restart.
26 * "systemctl clean" may now be used also for socket, mount, and swap
29 * systemd will also read configuration options from the EFI variable
30 SystemdOptions. This may be used to configure systemd behaviour when
31 modifying the kernel command line is inconvenient, but configuration
32 on disk is read too late, for example for the options related to
33 cgroup hierarchy setup. 'bootctl system-options' may be used to
36 * systemd will now disable printk ratelimits in early boot. This should
37 allow us to capture more logs from the early boot phase where normal
38 storage is not available and the kernel ring buffer is used for
39 logging. Configuration on the kernel command line has higher priority
40 and overrides the systemd setting.
42 systemd programs which log to /dev/kmsg directly use internal
43 ratelimits to prevent runaway logging. (Normally this is only used
44 during early boot, so in practice this change has very little
47 * Unit files now support top level dropin directories of the form
48 <unit_type>.d/ (e.g. service.d/) that may be used to add configuration
49 that affects all corresponding unit files.
51 * The RuntimeMaxSec= setting is now supported by scopes, not just
52 .service units. This is particularly useful for PAM sessions which
53 create a scope unit for the user login. systemd.runtime_max_sec=
54 setting may used with the pam_systemd module to limit the duration
55 of the PAM session, for example for time-limited logins.
57 * udev now provides a program (fido_id) that identifies FIDO CTAP1
58 ("U2F")/CTAP2 security tokens based on the usage declared in their
59 report and descriptor and outputs suitable environment variables.
60 This replaces the externally maintained whitelists of all known
61 security tokens that were used previously.
63 * Automatically generated autosuspend udev rules for whitelisted
64 devices have been imported from the Chromium OS project. This should
65 improve power saving with many more devices.
67 * udev gained a new "CONST{key}=value" setting that allows matching
68 against system-wide constants without forking a helper binary.
69 Currently "arch" and "virt" keys are supported.
71 * udev now opens CDROMs in non-exclusive mode when querying their
72 capabilities. This should fix issues where other programs trying to
73 use the CDROM cannot gain access to it, but carries a risk of
74 interfering with programs writing to the disk, if they did not open
75 the device in exclusive mode as they should.
77 * systemd-networkd does not create a default route for IPv4 link local
78 addressing anymore. The creation of the route was unexpected and was
79 breaking routing in various cases, but people who rely on it being
80 created implicitly will need to adjust. Such a route may be requested
81 with DefaultRouteOnDevice=yes.
83 Similarly, systemd-networkd will not assign a link-local IPv6 address
84 when IPv6 link-local routing is not enabled.
86 * Receive and transmit buffers may now be configured on links with
87 the new RxBufferSize= and TxBufferSize= settings.
89 * systemd-networkd may now advertise additional IPv6 routes. A new
90 [IPv6RoutePrefix] section with Route= and LifetimeSec= options is
93 * systemd-networkd may now configure "next hop" routes using the
94 [NextHop] section and Gateway= and Id= settings.
96 * systemd-networkd will now retain DHCP config on restarts by default
97 (but this may be overridden using the KeepConfiguration= setting).
98 The default for SendRelease= has been changed to true.
100 * The DHCPv4 client now uses the OPTION_INFORMATION_REFRESH_TIME option
101 received from the server.
103 The client will use the received SIP server list if UseSIP=yes is
106 The client may be configured to request specific options from the
107 server using a new RequestOptions= setting.
109 A new IPServiceType= setting has been added to configure the "IP
110 service type" value used by the client.
112 * The DHCPv6 client learnt a new PrefixDelegationHint= option to
113 request prefix hints in the DHCPv6 solicitation.
115 * systemd-networkd and networkctl may now renew DHCP leases on demand.
116 networkctl has a new 'networkctl renew' verb.
118 * systemd-networkd may now reconfigure links on demand. networkctl
119 gained two new verbs: "reload" will reload the configuration, and
120 "reconfigure DEVICE…" will reconfigure one or more devices.
122 * .network files may now match on SSID and BSSID of a wireless network,
123 i.e. the access point name and hardware address using the new SSID=
124 and BSSID= options. networkctl will display the current SSID and
125 BSSID for wireless links.
127 .network files may also match on the wireless network type using the
128 new WLANInterfaceType= option.
130 * systemd-networkd now includes default configuration that enables
131 link-local addressing when connected to an ad-hoc wireless network.
133 * The DHCPv4 server may now be configured to emit SIP server list using
134 the new EmitSIP= and SIP= settings.
136 * systemd-networkd may configure the Traffic Control queueing
137 disciplines in the kernel using the new
138 [TrafficControlQueueingDiscipline] section and Parent=,
139 NetworkEmulatorDelaySec=, NetworkEmulatorDelayJitterSec=,
140 NetworkEmulatorPacketLimit=, NetworkEmulatorLossRate=,
141 NetworkEmulatorDuplicateRate= settings.
143 * systemd-tmpfiles gained a new w+ setting to append to files.
145 * systemd-analyze dump will now report when the memory configuration in
146 the kernel does not match what systemd has configured (usually,
147 because some external program has modified the kernel configuration
150 * journalctl --update-catalog now produces deterministic output (making
151 reproducible image builds easier).
153 * A new devicetree-overlay setting is now documented in the Boot Loader
156 * The default value of the WatchdogSec= setting used in systemd
157 services (the ones bundled with the project itself) may be set at
158 configuration time using the -Dservice-watchdog= setting. If set to
159 empty, the watchdogs will be disabled.
161 * libcryptsetup >= 2.0.1 is now required.
163 * systemd-resolved validates IP addresses in certificates now when GnuTLS
168 * This release enables unprivileged programs (i.e. requiring neither
169 setuid nor file capabilities) to send ICMP Echo (i.e. ping) requests
170 by turning on the "net.ipv4.ping_group_range" sysctl of the Linux
171 kernel for the whole UNIX group range, i.e. all processes. This
172 change should be reasonably safe, as the kernel support for it was
173 specifically implemented to allow safe access to ICMP Echo for
174 processes lacking any privileges. If this is not desirable, it can be
175 disabled again by setting the parameter to "1 0".
177 * Previously, filters defined with SystemCallFilter= would have the
178 effect that any calling of an offending system call would terminate
179 the calling thread. This behaviour never made much sense, since
180 killing individual threads of unsuspecting processes is likely to
181 create more problems than it solves. With this release the default
182 action changed from killing the thread to killing the whole
183 process. For this to work correctly both a kernel version (>= 4.14)
184 and a libseccomp version (>= 2.4.0) supporting this new seccomp
185 action is required. If an older kernel or libseccomp is used the old
186 behaviour continues to be used. This change does not affect any
187 services that have no system call filters defined, or that use
188 SystemCallErrorNumber= (and thus see EPERM or another error instead
189 of being killed when calling an offending system call). Note that
190 systemd documentation always claimed that the whole process is
191 killed. With this change behaviour is thus adjusted to match the
194 * On 64 bit systems, the "kernel.pid_max" sysctl is now bumped to
195 4194304 by default, i.e. the full 22bit range the kernel allows, up
196 from the old 16bit range. This should improve security and
197 robustness, as PID collisions are made less likely (though certainly
198 still possible). There are rumours this might create compatibility
199 problems, though at this moment no practical ones are known to
200 us. Downstream distributions are hence advised to undo this change in
201 their builds if they are concerned about maximum compatibility, but
202 for everybody else we recommend leaving the value bumped. Besides
203 improving security and robustness this should also simplify things as
204 the maximum number of allowed concurrent tasks was previously bounded
205 by both "kernel.pid_max" and "kernel.threads-max" and now effectively
206 only a single knob is left ("kernel.threads-max"). There have been
207 concerns that usability is affected by this change because larger PID
208 numbers are harder to type, but we believe the change from 5 digits
209 to 7 digits doesn't hamper usability.
211 * MemoryLow= and MemoryMin= gained hierarchy-aware counterparts,
212 DefaultMemoryLow= and DefaultMemoryMin=, which can be used to
213 hierarchically set default memory protection values for a particular
214 subtree of the unit hierarchy.
216 * Memory protection directives can now take a value of zero, allowing
217 explicit opting out of a default value propagated by an ancestor.
219 * systemd now defaults to the "unified" cgroup hierarchy setup during
220 build-time, i.e. -Ddefault-hierarchy=unified is now the build-time
221 default. Previously, -Ddefault-hierarchy=hybrid was the default. This
222 change reflects the fact that cgroupsv2 support has matured
223 substantially in both systemd and in the kernel, and is clearly the
224 way forward. Downstream production distributions might want to
225 continue to use -Ddefault-hierarchy=hybrid (or even =legacy) for
226 their builds as unfortunately the popular container managers have not
227 caught up with the kernel API changes.
229 * Man pages are not built by default anymore (html pages were already
230 disabled by default), to make development builds quicker. When
231 building systemd for a full installation with documentation, meson
232 should be called with -Dman=true and/or -Dhtml=true as appropriate.
233 The default was changed based on the assumption that quick one-off or
234 repeated development builds are much more common than full optimized
235 builds for installation, and people need to pass various other
236 options to when doing "proper" builds anyway, so the gain from making
237 development builds quicker is bigger than the one time disruption for
240 Two scripts are created in the *build* directory to generate and
241 preview man and html pages on demand, e.g.:
243 build/man/man systemctl
244 build/man/html systemd.index
246 * libidn2 is used by default if both libidn2 and libidn are installed.
247 Please use -Dlibidn=true if libidn is preferred.
249 * The D-Bus "wire format" of the CPUAffinity= attribute is changed on
250 big-endian machines. Before, bytes were written and read in native
251 machine order as exposed by the native libc __cpu_mask interface.
252 Now, little-endian order is always used (CPUs 0–7 are described by
253 bits 0–7 in byte 0, CPUs 8–15 are described by byte 1, and so on).
254 This change fixes D-Bus calls that cross endianness boundary.
256 The presentation format used for CPUAffinity= by "systemctl show" and
257 "systemd-analyze dump" is changed to present CPU indices instead of
258 the raw __cpu_mask bitmask. For example, CPUAffinity=0-1 would be
259 shown as CPUAffinity=03000000000000000000000000000… (on
260 little-endian) or CPUAffinity=00000000000000300000000000000… (on
261 64-bit big-endian), and is now shown as CPUAffinity=0-1, matching the
262 input format. The maximum integer that will be printed in the new
263 format is 8191 (four digits), while the old format always used a very
264 long number (with the length varying by architecture), so they can be
265 unambiguously distinguished.
267 * /usr/sbin/halt.local is no longer supported. Implementation in
268 distributions was inconsistent and it seems this functionality was
271 To replace this functionality, users should:
272 - either define a new unit and make it a dependency of final.target
273 (systemctl add-wants final.target my-halt-local.service)
274 - or move the shutdown script to /usr/lib/systemd/system-shutdown/
275 and ensure that it accepts "halt", "poweroff", "reboot", and
276 "kexec" as an argument, see the description in systemd-shutdown(8).
278 * When a [Match] section in .link or .network file is empty (contains
279 no match patterns), a warning will be emitted. Please add any "match
280 all" pattern instead, e.g. OriginalName=* or Name=* in case all
281 interfaces should really be matched.
283 * A new setting NUMAPolicy= may be used to set process memory
284 allocation policy. This setting can be specified in
285 /etc/systemd/system.conf and hence will set the default policy for
286 PID1. The default policy can be overridden on a per-service
287 basis. The related setting NUMAMask= is used to specify NUMA node
288 mask that should be associated with the selected policy.
290 * PID 1 will now listen to Out-Of-Memory (OOM) events the kernel
291 generates when processes it manages are reaching their memory limits,
292 and will place their units in a special state, and optionally kill or
295 * The service manager will now expose bus properties for the IO
296 resources used by units. This information is also shown in "systemctl
297 status" now (for services that have IOAccounting=yes set). Moreover,
298 the IO accounting data is included in the resource log message
299 generated whenever a unit stops.
301 * Units may now configure an explicit time-out to wait for when killed
302 with SIGABRT, for example when a service watchdog is hit. Previously,
303 the regular TimeoutStopSec= time-out was applied in this case too —
304 now a separate time-out may be set using TimeoutAbortSec=.
306 * Services may now send a special WATCHDOG=trigger message with
307 sd_notify() to trigger an immediate "watchdog missed" event, and thus
308 trigger service termination. This is useful both for testing watchdog
309 handling, but also for defining error paths in services, that shall
310 be handled the same way as watchdog events.
312 * There are two new per-unit settings IPIngressFilterPath= and
313 IPEgressFilterPath= which allow configuration of a BPF program
314 (usually by specifying a path to a program uploaded to /sys/fs/bpf/)
315 to apply to the IP packet ingress/egress path of all processes of a
316 unit. This is useful to allow running systemd services with BPF
317 programs set up externally.
319 * systemctl gained a new "clean" verb for removing the state, cache,
320 runtime or logs directories of a service while it is terminated. The
321 new verb may also be used to remove the state maintained on disk for
322 timer units that have Persistent= configured.
324 * During the last phase of shutdown systemd will now automatically
325 increase the log level configured in the "kernel.printk" sysctl so
326 that any relevant loggable events happening during late shutdown are
327 made visible. Previously, loggable events happening so late during
328 shutdown were generally lost if the "kernel.printk" sysctl was set to
329 high thresholds, as regular logging daemons are terminated at that
330 time and thus nothing is written to disk.
332 * If processes terminated during the last phase of shutdown do not exit
333 quickly systemd will now show their names after a short time, to make
334 debugging easier. After a longer time-out they are forcibly killed,
337 * journalctl (and the other tools that display logs) will now highlight
338 warnings in yellow (previously, both LOG_NOTICE and LOG_WARNING where
339 shown in bright bold, now only LOG_NOTICE is). Moreover, audit logs
340 are now shown in blue color, to separate them visually from regular
341 logs. References to configuration files are now turned into clickable
342 links on terminals that support that.
344 * systemd-journald will now stop logging to /var/log/journal during
345 shutdown when /var/ is on a separate mount, so that it can be
346 unmounted safely during shutdown.
348 * systemd-resolved gained support for a new 'strict' DNS-over-TLS mode.
350 * systemd-resolved "Cache=" configuration option in resolved.conf has
351 been extended to also accept the 'no-negative' value. Previously,
352 only a boolean option was allowed (yes/no), having yes as the
353 default. If this option is set to 'no-negative', negative answers are
354 not cached while the old cache heuristics are used positive answers.
355 The default remains unchanged.
357 * The predictable naming scheme for network devices now supports
358 generating predictable names for "netdevsim" devices.
360 Moreover, the "en" prefix was dropped from the ID_NET_NAME_ONBOARD
363 Those two changes form a new net.naming-policy-scheme= entry.
364 Distributions which want to preserve naming stability may want to set
365 the -Ddefault-net-naming-scheme= configuration option.
367 * systemd-networkd now supports MACsec, nlmon, IPVTAP and Xfrm
370 * systemd-networkd's bridge FDB support now allows configuration of a
371 destination address for each entry (Destination=), as well as the
372 VXLAN VNI (VNI=), as well as an option to declare what an entry is
373 associated with (AssociatedWith=).
375 * systemd-networkd's DHCPv4 support now understands a new MaxAttempts=
376 option for configuring the maximum number of DHCP lease requests. It
377 also learnt a new BlackList= option for blacklisting DHCP servers (a
378 similar setting has also been added to the IPv6 RA client), as well
379 as a SendRelease= option for configuring whether to send a DHCP
380 RELEASE message when terminating.
382 * systemd-networkd's DHCPv4 and DHCPv6 stacks can now be configured
383 separately in the [DHCPv4] and [DHCPv6] sections.
385 * systemd-networkd's DHCP support will now optionally create an
386 implicit host route to the DNS server specified in the DHCP lease, in
387 addition to the routes listed explicitly in the lease. This should
388 ensure that in multi-homed systems DNS traffic leaves the systems on
389 the interface that acquired the DNS server information even if other
390 routes such as default routes exist. This behaviour may be turned on
391 with the new RoutesToDNS= option.
393 * systemd-networkd's VXLAN support gained a new option
394 GenericProtocolExtension= for enabling VXLAN Generic Protocol
395 Extension support, as well as IPDoNotFragment= for setting the IP
396 "Don't fragment" bit on outgoing packets. A similar option has been
397 added to the GENEVE support.
399 * In systemd-networkd's [Route] section you may now configure
400 FastOpenNoCookie= for configuring per-route TCP fast-open support, as
401 well as TTLPropagate= for configuring Label Switched Path (LSP) TTL
402 propagation. The Type= setting now supports local, broadcast,
403 anycast, multicast, any, xresolve routes, too.
405 * systemd-networkd's [Network] section learnt a new option
406 DefaultRouteOnDevice= for automatically configuring a default route
407 onto the network device.
409 * systemd-networkd's bridging support gained two new options ProxyARP=
410 and ProxyARPWifi= for configuring proxy ARP behaviour as well as
411 MulticastRouter= for configuring multicast routing behaviour. A new
412 option MulticastIGMPVersion= may be used to change bridge's multicast
413 Internet Group Management Protocol (IGMP) version.
415 * systemd-networkd's FooOverUDP support gained the ability to configure
416 local and peer IP addresses via Local= and Peer=. A new option
417 PeerPort= may be used to configure the peer's IP port.
419 * systemd-networkd's TUN support gained a new setting VnetHeader= for
420 tweaking Generic Segment Offload support.
422 * networkctl gained a new "delete" command for removing virtual network
423 devices, as well as a new "--stats" switch for showing device
426 * networkd.conf gained a new setting SpeedMeter= and
427 SpeedMeterIntervalSec=, to measure bitrate of network interfaces. The
428 measured speed may be shown by 'networkctl status'.
430 * "networkctl status" now displays MTU and queue lengths, and more
431 detailed information about VXLAN and bridge devices.
433 * systemd-networkd's .network and .link files gained a new Property=
434 setting in the [Match] section, to match against devices with
435 specific udev properties.
437 * systemd-networkd's tunnel support gained a new option
438 AssignToLoopback= for selecting whether to use the loopback device
439 "lo" as underlying device.
441 * systemd-networkd's MACAddress= setting in the [Neighbor] section has
442 been renamed to LinkLayerAddress=, and it now allows configuration of
445 * systemd-networkd's handling of the kernel's disable_ipv6 sysctl is
446 simplified: systemd-networkd will disable the sysctl (enable IPv6) if
447 IPv6 configuration (static or DHCPv6) was found for a given
448 interface. It will not touch the sysctl otherwise.
450 * The order of entries is $PATH used by the user manager instance was
451 changed to put bin/ entries before the corresponding sbin/ entries.
452 It is recommended to not rely on this order, and only ever have one
453 binary with a given name in the system paths under /usr.
455 * A new tool systemd-network-generator has been added that may generate
456 .network, .netdev and .link files from IP configuration specified on
457 the kernel command line in the format used by Dracut.
459 * The CriticalConnection= setting in .network files is now deprecated,
460 and replaced by a new KeepConfiguration= setting which allows more
461 detailed configuration of the IP configuration to keep in place.
463 * systemd-analyze gained a few new verbs:
465 - "systemd-analyze timestamp" parses and converts timestamps. This is
466 similar to the existing "systemd-analyze calendar" command which
467 does the same for recurring calendar events.
469 - "systemd-analyze timespan" parses and converts timespans (i.e.
470 durations as opposed to points in time).
472 - "systemd-analyze condition" will parse and test ConditionXYZ=
475 - "systemd-analyze exit-status" will parse and convert exit status
476 codes to their names and back.
478 - "systemd-analyze unit-files" will print a list of all unit
479 file paths and unit aliases.
481 * SuccessExitStatus=, RestartPreventExitStatus=, and
482 RestartForceExitStatus= now accept exit status names (e.g. "DATAERR"
483 is equivalent to "65"). Those exit status name mappings may be
484 displayed with the sytemd-analyze exit-status verb describe above.
486 * systemd-logind now exposes a per-session SetBrightness() bus call,
487 which may be used to securely change the brightness of a kernel
488 brightness device, if it belongs to the session's seat. By using this
489 call unprivileged clients can make changes to "backlight" and "leds"
490 devices securely with strict requirements on session membership.
491 Desktop environments may use this to generically make brightness
492 changes to such devices without shipping private SUID binaries or
493 udev rules for that purpose.
495 * "udevadm info" gained a --wait-for-initialization switch to wait for
496 a device to be initialized.
498 * systemd-hibernate-resume-generator will now look for resumeflags= on
499 the kernel command line, which is similar to rootflags= and may be
500 used to configure device timeout for the hibernation device.
502 * sd-event learnt a new API call sd_event_source_disable_unref() for
503 disabling and unref'ing an event source in a single function. A
504 related call sd_event_source_disable_unrefp() has been added for use
505 with gcc's cleanup extension.
507 * The sd-id128.h public API gained a new definition
508 SD_ID128_UUID_FORMAT_STR for formatting a 128bit ID in UUID format
511 * "busctl introspect" gained a new switch --xml-interface for dumping
512 XML introspection data unmodified.
514 * PID 1 may now show the unit name instead of the unit description
515 string in its status output during boot. This may be configured in
516 the StatusUnitFormat= setting in /etc/systemd/system.conf or the
517 kernel command line option systemd.status_unit_format=.
519 * PID 1 now understands a new option KExecWatchdogSec= in
520 /etc/systemd/system.conf to set a watchdog timeout for kexec reboots.
521 Previously watchdog functionality was only available for regular
522 reboots. The new setting defaults to off, because we don't know in
523 the general case if the watchdog will be reset after kexec (some
524 drivers do reset it, but not all), and the new userspace might not be
525 configured to handle the watchdog.
527 Moreover, the old ShutdownWatchdogSec= setting has been renamed to
528 RebootWatchdogSec= to more clearly communicate what it is about. The
529 old name is still accepted for compatibility.
531 * The systemd.debug_shell kernel command line option now optionally
532 takes a tty name to spawn the debug shell on, which allows a
533 different tty to be selected than the built-in default.
535 * Service units gained a new ExecCondition= setting which will run
536 before ExecStartPre= and either continue execution of the unit (for
537 clean exit codes), stop execution without marking the unit failed
538 (for exit codes 1 through 254), or stop execution and fail the unit
539 (for exit code 255 or abnormal termination).
541 * A new service systemd-pstore.service has been added that pulls data
542 from /sys/fs/pstore/ and saves it to /var/lib/pstore for later
545 * timedatectl gained new verbs for configuring per-interface NTP
546 service configuration for systemd-timesyncd.
548 * "localectl list-locales" won't list non-UTF-8 locales anymore. It's
549 2019. (You can set non-UTF-8 locales though, if you know their name.)
551 * If variable assignments in sysctl.d/ files are prefixed with "-" any
552 failures to apply them are now ignored.
554 * systemd-random-seed.service now optionally credits entropy when
555 applying the seed to the system. Set $SYSTEMD_RANDOM_SEED_CREDIT to
556 true for the service to enable this behaviour, but please consult the
557 documentation first, since this comes with a couple of caveats.
559 * systemd-random-seed.service is now a synchronization point for full
560 initialization of the kernel's entropy pool. Services that require
561 /dev/urandom to be correctly initialized should be ordered after this
564 * The systemd-boot boot loader has been updated to optionally maintain
565 a random seed file in the EFI System Partition (ESP). During the boot
566 phase, this random seed is read and updated with a new seed
567 cryptographically derived from it. Another derived seed is passed to
568 the OS. The latter seed is then credited to the kernel's entropy pool
569 very early during userspace initialization (from PID 1). This allows
570 systems to boot up with a fully initialized kernel entropy pool from
571 earliest boot on, and thus entirely removes all entropy pool
572 initialization delays from systems using systemd-boot. Special care
573 is taken to ensure different seeds are derived on system images
574 replicated to multiple systems. "bootctl status" will show whether
575 a seed was received from the boot loader.
577 * bootctl gained two new verbs:
579 - "bootctl random-seed" will generate the file in ESP and an EFI
580 variable to allow a random seed to be passed to the OS as described
583 - "bootctl is-installed" checks whether systemd-boot is currently
586 * bootctl will warn if it detects that boot entries are misconfigured
587 (for example if the kernel image was removed without purging the
590 * A new document has been added describing systemd's use and support
591 for the kernel's entropy pool subsystem:
593 https://systemd.io/RANDOM_SEEDS
595 * When the system is hibernated the swap device to write the
596 hibernation image to is now automatically picked from all available
597 swap devices, preferring the swap device with the highest configured
598 priority over all others, and picking the device with the most free
599 space if there are multiple devices with the highest priority.
601 * /etc/crypttab support has learnt a new keyfile-timeout= per-device
602 option that permits selecting the timout how long to wait for a
603 device with an encryption key before asking for the password.
605 * IOWeight= has learnt to properly set the IO weight when using the
606 BFQ scheduler officially found in kernels 5.0+.
608 * A new mailing list has been created for reporting of security issues:
609 systemd-security@redhat.com. For mode details, see
610 https://systemd.io/CONTRIBUTING#security-vulnerability-reports.
612 Contributions from: Aaron Barany, Adrian Bunk, Alan Jenkins, Albrecht
613 Lohofener, Andrej Valek, Anita Zhang, Arian van Putten, Balint Reczey,
614 Bastien Nocera, Ben Boeckel, Benjamin Robin, camoz, Chen Qi, Chris
615 Chiu, Chris Down, Christian Kellner, Clinton Roy, Connor Reeder, Daniel
616 Black, Daniele Medri, Dan Streetman, Dave Reisner, Dave Ross, David
617 Art, David Tardon, Debarshi Ray, Dimitri John Ledkov, Dominick Grift,
618 Donald Buczek, Douglas Christman, Eric DeVolder, EtherGraf, Evgeny
619 Vereshchagin, Feldwor, Felix Riemann, Florian Dollinger, Francesco
620 Pennica, Franck Bui, Frantisek Sumsal, Franz Pletz, frederik, Hans
621 de Goede, Iago López Galeiras, Insun Pyo, Ivan Shapovalov, Iwan Timmer,
622 Jack, Jakob Unterwurzacher, Jan Chren, Jan Klötzke, Jan Losinski, Jan
623 Pokorný, Jan Synacek, Jan-Michael Brummer, Jeka Pats, Jeremy Soller,
624 Jérémy Rosen, Jiri Pirko, Joe Lin, Joerg Behrmann, Joe Richey, Jóhann
625 B. Guðmundsson, Johannes Christ, Johannes Schmitz, Jonathan Rouleau,
626 Jorge Niedbalski, Kai Krakow, Kai Lüke, Karel Zak, Kashyap Chamarthy,
627 Krayushkin Konstantin, Lennart Poettering, Lubomir Rintel, Luca
628 Boccassi, Luís Ferreira, Marc-André Lureau, Markus Felten, Martin Pitt,
629 Matthew Leeds, Mattias Jernberg, Michael Biebl, Michael Olbrich,
630 Michael Prokop, Michael Stapelberg, Michael Zhivich, Michal Koutný,
631 Michal Sekletar, Mike Gilbert, Milan Broz, Miroslav Lichvar, mpe85,
632 Mr-Foo, Network Silence, Oliver Harley, pan93412, Paul Menzel, pEJipE,
633 Peter A. Bigot, Philip Withnall, Piotr Drąg, Rafael Fontenelle, Roberto
634 Santalla, Ronan Pigott, root, RussianNeuroMancer, Sebastian Jennen,
635 shinygold, Shreyas Behera, Simon Schricker, Susant Sahani, Thadeu Lima
636 de Souza Cascardo, Theo Ouzhinski, Thiebaud Weksteen, Thomas Haller,
637 Thomas Weißschuh, Tomas Mraz, Tommi Rantala, Topi Miettinen, VD-Lycos,
638 ven, Wieland Hoffmann, William A. Kennington III, William Wold, Xi
639 Ruoyao, Yuri Chornoivan, Yu Watanabe, Zach Smith, Zbigniew
640 Jędrzejewski-Szmek, Zhang Xianwei
642 – Camerino, 2019-09-03
646 * In .link files, MACAddressPolicy=persistent (the default) is changed
647 to cover more devices. For devices like bridges, tun, tap, bond, and
648 similar interfaces that do not have other identifying information,
649 the interface name is used as the basis for persistent seed for MAC
650 and IPv4LL addresses. The way that devices that were handled
651 previously is not changed, and this change is about covering more
652 devices then previously by the "persistent" policy.
654 MACAddressPolicy=random may be used to force randomized MACs and
655 IPv4LL addresses for a device if desired.
657 Hint: the log output from udev (at debug level) was enhanced to
658 clarify what policy is followed and which attributes are used.
659 `SYSTEMD_LOG_LEVEL=debug udevadm test-builtin net_setup_link /sys/class/net/<name>`
660 may be used to view this.
662 Hint: if a bridge interface is created without any slaves, and gains
663 a slave later, then now the bridge does not inherit slave's MAC.
664 To inherit slave's MAC, for example, create the following file:
666 # /etc/systemd/network/98-bridge-inherit-mac.link
671 MACAddressPolicy=none
674 * The .device units generated by systemd-fstab-generator and other
675 generators do not automatically pull in the corresponding .mount unit
676 as a Wants= dependency. This means that simply plugging in the device
677 will not cause the mount unit to be started automatically. But please
678 note that the mount unit may be started for other reasons, in
679 particular if it is part of local-fs.target, and any unit which
680 (transitively) depends on local-fs.target is started.
682 * networkctl list/status/lldp now accept globbing wildcards for network
683 interface names to match against all existing interfaces.
685 * The $PIDFILE environment variable is set to point the absolute path
686 configured with PIDFile= for processes of that service.
688 * The fallback DNS server list was augmented with Cloudflare public DNS
689 servers. Use `-Ddns-servers=` to set a different fallback.
691 * A new special target usb-gadget.target will be started automatically
692 when a USB Device Controller is detected (which means that the system
693 is a USB peripheral).
695 * A new unit setting CPUQuotaPeriodSec= assigns the time period
696 relatively to which the CPU time quota specified by CPUQuota= is
699 * A new unit setting ProtectHostname= may be used to prevent services
700 from modifying hostname information (even if they otherwise would
701 have privileges to do so).
703 * A new unit setting NetworkNamespacePath= may be used to specify a
704 namespace for service or socket units through a path referring to a
705 Linux network namespace pseudo-file.
707 * The PrivateNetwork= setting and JoinsNamespaceOf= dependencies now
708 have an effect on .socket units: when used the listening socket is
709 created within the configured network namespace instead of the host
712 * ExecStart= command lines in unit files may now be prefixed with ':'
713 in which case environment variable substitution is
714 disabled. (Supported for the other ExecXYZ= settings, too.)
716 * .timer units gained two new boolean settings OnClockChange= and
717 OnTimezoneChange= which may be used to also trigger a unit when the
718 system clock is changed or the local timezone is
719 modified. systemd-run has been updated to make these options easily
720 accessible from the command line for transient timers.
722 * Two new conditions for units have been added: ConditionMemory= may be
723 used to conditionalize a unit based on installed system
724 RAM. ConditionCPUs= may be used to conditionalize a unit based on
727 * The @default system call filter group understood by SystemCallFilter=
728 has been updated to include the new rseq() system call introduced in
731 * A new time-set.target has been added that indicates that the system
732 time has been set from a local source (possibly imprecise). The
733 existing time-sync.target is stronger and indicates that the time has
734 been synchronized with a precise external source. Services where
735 approximate time is sufficient should use the new target.
737 * "systemctl start" (and related commands) learnt a new
738 --show-transaction option. If specified brief information about all
739 jobs queued because of the requested operation is shown.
741 * systemd-networkd recognizes a new operation state 'enslaved', used
742 (instead of 'degraded' or 'carrier') for interfaces which form a
743 bridge, bond, or similar, and an new 'degraded-carrier' operational
744 state used for the bond or bridge master interface when one of the
745 enslaved devices is not operational.
747 * .network files learnt the new IgnoreCarrierLoss= option for leaving
748 networks configured even if the carrier is lost.
750 * The RequiredForOnline= setting in .network files may now specify a
751 minimum operational state required for the interface to be considered
752 "online" by systemd-networkd-wait-online. Related to this
753 systemd-networkd-wait-online gained a new option --operational-state=
754 to configure the same, and its --interface= option was updated to
755 optionally also take an operational state specific for an interface.
757 * systemd-networkd-wait-online gained a new setting --any for waiting
758 for only one of the requested interfaces instead of all of them.
760 * systemd-networkd now implements L2TP tunnels.
762 * Two new .network settings UseAutonomousPrefix= and UseOnLinkPrefix=
763 may be used to cause autonomous and onlink prefixes received in IPv6
764 Router Advertisements to be ignored.
766 * New MulticastFlood=, NeighborSuppression=, and Learning= .network
767 file settings may be used to tweak bridge behaviour.
769 * The new TripleSampling= option in .network files may be used to
770 configure CAN triple sampling.
772 * A new .netdev settings PrivateKeyFile= and PresharedKeyFile= may be
773 used to point to private or preshared key for a WireGuard interface.
775 * /etc/crypttab now supports the same-cpu-crypt and
776 submit-from-crypt-cpus options to tweak encryption work scheduling
779 * systemd-tmpfiles will now take a BSD file lock before operating on a
780 contents of directory. This may be used to temporarily exclude
781 directories from aging by taking the same lock (useful for example
782 when extracting a tarball into /tmp or /var/tmp as a privileged user,
783 which might create files with really old timestamps, which
784 nevertheless should not be deleted). For further details, see:
786 https://systemd.io/TEMPORARY_DIRECTORIES
788 * systemd-tmpfiles' h line type gained support for the
789 FS_PROJINHERIT_FL ('P') file attribute (introduced in kernel 4.5),
790 controlling project quota inheritance.
792 * sd-boot and bootctl now implement support for an Extended Boot Loader
793 (XBOOTLDR) partition, that is intended to be mounted to /boot, in
794 addition to the ESP partition mounted to /efi or /boot/efi.
795 Configuration file fragments, kernels, initrds and other EFI images
796 to boot will be loaded from both the ESP and XBOOTLDR partitions.
797 The XBOOTLDR partition was previously described by the Boot Loader
798 Specification, but implementation was missing in sd-boot. Support for
799 this concept allows using the sd-boot boot loader in more
800 conservative scenarios where the boot loader itself is placed in the
801 ESP but the kernels to boot (and their metadata) in a separate
804 * A system may now be booted with systemd.volatile=overlay on the
805 kernel command line, which causes the root file system to be set up
806 an overlayfs mount combining the root-only root directory with a
807 writable tmpfs. In this setup, the underlying root device is not
808 modified, and any changes are lost at reboot.
810 * Similar, systemd-nspawn can now boot containers with a volatile
811 overlayfs root with the new --volatile=overlay switch.
813 * systemd-nspawn can now consume OCI runtime bundles using a new
814 --oci-bundle= option. This implementation is fully usable, with most
815 features in the specification implemented, but since this a lot of
816 new code and functionality, this feature should most likely not
817 be used in production yet.
819 * systemd-nspawn now supports various options described by the OCI
820 runtime specification on the command-line and in .nspawn files:
821 --inaccessible=/Inaccessible= may be used to mask parts of the file
822 system tree, --console=/--pipe may be used to configure how standard
823 input, output, and error are set up.
825 * busctl learned the `emit` verb to generate D-Bus signals.
827 * systemd-analyze cat-config may be used to gather and display
828 configuration spread over multiple files, for example system and user
829 presets, tmpfiles.d, sysusers.d, udev rules, etc.
831 * systemd-analyze calendar now takes an optional new parameter
832 --iterations= which may be used to show a maximum number of iterations
833 the specified expression will elapse next.
835 * The sd-bus C API gained support for naming method parameters in the
838 * systemd-logind gained D-Bus APIs to specify the "reboot parameter"
839 the reboot() system call expects.
841 * journalctl learnt a new --cursor-file= option that points to a file
842 from which a cursor should be loaded in the beginning and to which
843 the updated cursor should be stored at the end.
845 * ACRN hypervisor and Windows Subsystem for Linux (WSL) are now
846 detected by systemd-detect-virt (and may also be used in
847 ConditionVirtualization=).
849 * The behaviour of systemd-logind may now be modified with environment
850 variables $SYSTEMD_REBOOT_TO_FIRMWARE_SETUP,
851 $SYSTEMD_REBOOT_TO_BOOT_LOADER_MENU, and
852 $SYSTEMD_REBOOT_TO_BOOT_LOADER_ENTRY. They cause logind to either
853 skip the relevant operation completely (when set to false), or to
854 create a flag file in /run/systemd (when set to true), instead of
855 actually commencing the real operation when requested. The presence
856 of /run/systemd/reboot-to-firmware-setup,
857 /run/systemd/reboot-to-boot-loader-menu, and
858 /run/systemd/reboot-to-boot-loader-entry, may be used by alternative
859 boot loader implementations to replace some steps logind performs
860 during reboot with their own operations.
862 * systemctl can be used to request a reboot into the boot loader menu
863 or a specific boot loader entry with the new --boot-load-menu= and
864 --boot-loader-entry= options to a reboot command. (This requires a
865 boot loader that supports this, for example sd-boot.)
867 * kernel-install will no longer unconditionally create the output
868 directory (e.g. /efi/<machine-id>/<kernel-version>) for boot loader
869 snippets, but will do only if the machine-specific parent directory
870 (i.e. /efi/<machine-id>/) already exists. bootctl has been modified
871 to create this parent directory during sd-boot installation.
873 This makes it easier to use kernel-install with plugins which support
874 a different layout of the bootloader partitions (for example grub2).
876 * During package installation (with `ninja install`), we would create
877 symlinks for getty@tty1.service, systemd-networkd.service,
878 systemd-networkd.socket, systemd-resolved.service,
879 remote-cryptsetup.target, remote-fs.target,
880 systemd-networkd-wait-online.service, and systemd-timesyncd.service
881 in /etc, as if `systemctl enable` was called for those units, to make
882 the system usable immediately after installation. Now this is not
883 done anymore, and instead calling `systemctl preset-all` is
884 recommended after the first installation of systemd.
886 * A new boolean sandboxing option RestrictSUIDSGID= has been added that
887 is built on seccomp. When turned on creation of SUID/SGID files is
890 * The NoNewPrivileges= and the new RestrictSUIDSGID= options are now
891 implied if DynamicUser= is turned on for a service. This hardens
892 these services, so that they neither can benefit from nor create
893 SUID/SGID executables. This is a minor compatibility breakage, given
894 that when DynamicUser= was first introduced SUID/SGID behaviour was
895 unaffected. However, the security benefit of these two options is
896 substantial, and the setting is still relatively new, hence we opted
897 to make it mandatory for services with dynamic users.
899 Contributions from: Adam Jackson, Alexander Tsoy, Andrey Yashkin,
900 Andrzej Pietrasiewicz, Anita Zhang, Balint Reczey, Beniamino Galvani,
901 Ben Iofel, Benjamin Berg, Benjamin Dahlhoff, Chris, Chris Morin,
902 Christopher Wong, Claudius Ellsel, Clemens Gruber, dana, Daniel Black,
903 Davide Cavalca, David Michael, David Rheinsberg, emersion, Evgeny
904 Vereshchagin, Filipe Brandenburger, Franck Bui, Frantisek Sumsal,
905 Giacinto Cifelli, Hans de Goede, Hugo Kindel, Ignat Korchagin, Insun
906 Pyo, Jan Engelhardt, Jonas Dorel, Jonathan Lebon, Jonathon Kowalski,
907 Jörg Sommer, Jörg Thalheim, Jussi Pakkanen, Kai-Heng Feng, Lennart
908 Poettering, Lubomir Rintel, Luís Ferreira, Martin Pitt, Matthias
909 Klumpp, Michael Biebl, Michael Niewöhner, Michael Olbrich, Michal
910 Sekletar, Mike Lothian, Paul Menzel, Piotr Drąg, Riccardo Schirone,
911 Robin Elvedi, Roman Kulikov, Ronald Tschalär, Ross Burton, Ryan
912 Gonzalez, Sebastian Krzyszkowiak, Stephane Chazelas, StKob, Susant
913 Sahani, Sylvain Plantefève, Szabolcs Fruhwald, Taro Yamada, Theo
914 Ouzhinski, Thomas Haller, Tobias Jungel, Tom Yan, Tony Asleson, Topi
915 Miettinen, unixsysadmin, Van Laser, Vesa Jääskeläinen, Yu, Li-Yu,
916 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
922 * The default locale can now be configured at compile time. Otherwise,
923 a suitable default will be selected automatically (one of C.UTF-8,
926 * The version string shown by systemd and other tools now includes the
927 git commit hash when built from git. An override may be specified
928 during compilation, which is intended to be used by distributions to
929 include the package release information.
931 * systemd-cat can now filter standard input and standard error streams
932 for different syslog priorities using the new --stderr-priority=
935 * systemd-journald and systemd-journal-remote reject entries which
936 contain too many fields (CVE-2018-16865) and set limits on the
937 process' command line length (CVE-2018-16864).
939 * $DBUS_SESSION_BUS_ADDRESS environment variable is set by pam_systemd
942 * A new network device NamePolicy "keep" is implemented for link files,
943 and used by default in 99-default.link (the fallback configuration
944 provided by systemd). With this policy, if the network device name
945 was already set by userspace, the device will not be renamed again.
946 This matches the naming scheme that was implemented before
947 systemd-240. If naming-scheme < 240 is specified, the "keep" policy
948 is also enabled by default, even if not specified. Effectively, this
949 means that if naming-scheme >= 240 is specified, network devices will
950 be renamed according to the configuration, even if they have been
951 renamed already, if "keep" is not specified as the naming policy in
952 the .link file. The 99-default.link file provided by systemd includes
953 "keep" for backwards compatibility, but it is recommended for user
954 installed .link files to *not* include it.
956 The "kernel" policy, which keeps kernel names declared to be
957 "persistent", now works again as documented.
959 * kernel-install script now optionally takes the paths to one or more
960 initrd files, and passes them to all plugins.
962 * The mincore() system call has been dropped from the @system-service
963 system call filter group, as it is pretty exotic and may potentially
964 used for side-channel attacks.
966 * -fPIE is dropped from compiler and linker options. Please specify
967 -Db_pie=true option to meson to build position-independent
968 executables. Note that the meson option is supported since meson-0.49.
970 * The fs.protected_regular and fs.protected_fifos sysctls, which were
971 added in Linux 4.19 to make some data spoofing attacks harder, are
972 now enabled by default. While this will hopefully improve the
973 security of most installations, it is technically a backwards
974 incompatible change; to disable these sysctls again, place the
975 following lines in /etc/sysctl.d/60-protected.conf or a similar file:
977 fs.protected_regular = 0
978 fs.protected_fifos = 0
980 Note that the similar hardlink and symlink protection has been
981 enabled since v199, and may be disabled likewise.
983 * The files read from the EnvironmentFile= setting in unit files now
984 parse backslashes inside quotes literally, matching the behaviour of
987 * udevadm trigger, udevadm control, udevadm settle and udevadm monitor
988 now automatically become NOPs when run in a chroot() environment.
990 * The tmpfiles.d/ "C" line type will now copy directory trees not only
991 when the destination is so far missing, but also if it already exists
992 as a directory and is empty. This is useful to cater for systems
993 where directory trees are put together from multiple separate mount
994 points but otherwise empty.
996 * A new function sd_bus_close_unref() (and the associated
997 sd_bus_close_unrefp()) has been added to libsystemd, that combines
998 sd_bus_close() and sd_bus_unref() in one.
1000 * udevadm control learnt a new option for --ping for testing whether a
1001 systemd-udevd instance is running and reacting.
1003 * udevadm trigger learnt a new option for --wait-daemon for waiting
1004 systemd-udevd daemon to be initialized.
1006 Contributions from: Aaron Plattner, Alberts Muktupāvels, Alex Mayer,
1007 Ayman Bagabas, Beniamino Galvani, Burt P, Chris Down, Chris Lamb, Chris
1008 Morin, Christian Hesse, Claudius Ellsel, dana, Daniel Axtens, Daniele
1009 Medri, Dave Reisner, David Santamaría Rogado, Diego Canuhe, Dimitri
1010 John Ledkov, Evgeny Vereshchagin, Fabrice Fontaine, Filipe
1011 Brandenburger, Franck Bui, Frantisek Sumsal, govwin, Hans de Goede,
1012 James Hilliard, Jan Engelhardt, Jani Uusitalo, Jan Janssen, Jan
1013 Synacek, Jonathan McDowell, Jonathan Roemer, Jonathon Kowalski, Joost
1014 Heitbrink, Jörg Thalheim, Lance, Lennart Poettering, Louis Taylor,
1015 Lucas Werkmeister, Mantas Mikulėnas, Marc-Antoine Perennou,
1016 marvelousblack, Michael Biebl, Michael Sloan, Michal Sekletar, Mike
1017 Auty, Mike Gilbert, Mikhail Kasimov, Neil Brown, Niklas Hambüchen,
1018 Patrick Williams, Paul Seyfert, Peter Hutterer, Philip Withnall, Roger
1019 James, Ronnie P. Thomas, Ryan Gonzalez, Sam Morris, Stephan Edel,
1020 Stephan Gerhold, Susant Sahani, Taro Yamada, Thomas Haller, Topi
1021 Miettinen, YiFei Zhu, YmrDtnJu, YunQiang Su, Yu Watanabe, Zbigniew
1022 Jędrzejewski-Szmek, zsergeant77, Дамјан Георгиевски
1024 — Berlin, 2019-02-14
1028 * NoNewPrivileges=yes has been set for all long-running services
1029 implemented by systemd. Previously, this was problematic due to
1030 SELinux (as this would also prohibit the transition from PID1's label
1031 to the service's label). This restriction has since been lifted, but
1032 an SELinux policy update is required.
1033 (See e.g. https://github.com/fedora-selinux/selinux-policy/pull/234.)
1035 * DynamicUser=yes is dropped from systemd-networkd.service,
1036 systemd-resolved.service and systemd-timesyncd.service, which was
1037 enabled in v239 for systemd-networkd.service and systemd-resolved.service,
1038 and since v236 for systemd-timesyncd.service. The users and groups
1039 systemd-network, systemd-resolve and systemd-timesync are created
1040 by systemd-sysusers again. Distributors or system administrators
1041 may need to create these users and groups if they not exist (or need
1042 to re-enable DynamicUser= for those units) while upgrading systemd.
1043 Also, the clock file for systemd-timesyncd may need to move from
1044 /var/lib/private/systemd/timesync/clock to /var/lib/systemd/timesync/clock.
1046 * When unit files are loaded from disk, previously systemd would
1047 sometimes (depending on the unit loading order) load units from the
1048 target path of symlinks in .wants/ or .requires/ directories of other
1049 units. This meant that unit could be loaded from different paths
1050 depending on whether the unit was requested explicitly or as a
1051 dependency of another unit, not honouring the priority of directories
1052 in search path. It also meant that it was possible to successfully
1053 load and start units which are not found in the unit search path, as
1054 long as they were requested as a dependency and linked to from
1055 .wants/ or .requires/. The target paths of those symlinks are not
1056 used for loading units anymore and the unit file must be found in
1059 * A new service type has been added: Type=exec. It's very similar to
1060 Type=simple but ensures the service manager will wait for both fork()
1061 and execve() of the main service binary to complete before proceeding
1062 with follow-up units. This is primarily useful so that the manager
1063 propagates any errors in the preparation phase of service execution
1064 back to the job that requested the unit to be started. For example,
1065 consider a service that has ExecStart= set to a file system binary
1066 that doesn't exist. With Type=simple starting the unit would be
1067 considered instantly successful, as only fork() has to complete
1068 successfully and the manager does not wait for execve(), and hence
1069 its failure is seen "too late". With the new Type=exec service type
1070 starting the unit will fail, as the manager will wait for the
1071 execve() and notice its failure, which is then propagated back to the
1074 NOTE: with the next release 241 of systemd we intend to change the
1075 systemd-run tool to default to Type=exec for transient services
1076 started by it. This should be mostly safe, but in specific corner
1077 cases might result in problems, as the systemd-run tool will then
1078 block on NSS calls (such as user name look-ups due to User=) done
1079 between the fork() and execve(), which under specific circumstances
1080 might cause problems. It is recommended to specify "-p Type=simple"
1081 explicitly in the few cases where this applies. For regular,
1082 non-transient services (i.e. those defined with unit files on disk)
1083 we will continue to default to Type=simple.
1085 * The Linux kernel's current default RLIMIT_NOFILE resource limit for
1086 userspace processes is set to 1024 (soft) and 4096
1087 (hard). Previously, systemd passed this on unmodified to all
1088 processes it forked off. With this systemd release the hard limit
1089 systemd passes on is increased to 512K, overriding the kernel's
1090 defaults and substantially increasing the number of simultaneous file
1091 descriptors unprivileged userspace processes can allocate. Note that
1092 the soft limit remains at 1024 for compatibility reasons: the
1093 traditional UNIX select() call cannot deal with file descriptors >=
1094 1024 and increasing the soft limit globally might thus result in
1095 programs unexpectedly allocating a high file descriptor and thus
1096 failing abnormally when attempting to use it with select() (of
1097 course, programs shouldn't use select() anymore, and prefer
1098 poll()/epoll, but the call unfortunately remains undeservedly popular
1099 at this time). This change reflects the fact that file descriptor
1100 handling in the Linux kernel has been optimized in more recent
1101 kernels and allocating large numbers of them should be much cheaper
1102 both in memory and in performance than it used to be. Programs that
1103 want to take benefit of the increased limit have to "opt-in" into
1104 high file descriptors explicitly by raising their soft limit. Of
1105 course, when they do that they must acknowledge that they cannot use
1106 select() anymore (and neither can any shared library they use — or
1107 any shared library used by any shared library they use and so on).
1108 Which default hard limit is most appropriate is of course hard to
1109 decide. However, given reports that ~300K file descriptors are used
1110 in real-life applications we believe 512K is sufficiently high as new
1111 default for now. Note that there are also reports that using very
1112 high hard limits (e.g. 1G) is problematic: some software allocates
1113 large arrays with one element for each potential file descriptor
1114 (Java, …) — a high hard limit thus triggers excessively large memory
1115 allocations in these applications. Hopefully, the new default of 512K
1116 is a good middle ground: higher than what real-life applications
1117 currently need, and low enough for avoid triggering excessively large
1118 allocations in problematic software. (And yes, somebody should fix
1121 * The fs.nr_open and fs.file-max sysctls are now automatically bumped
1122 to the highest possible values, as separate accounting of file
1123 descriptors is no longer necessary, as memcg tracks them correctly as
1124 part of the memory accounting anyway. Thus, from the four limits on
1125 file descriptors currently enforced (fs.file-max, fs.nr_open,
1126 RLIMIT_NOFILE hard, RLIMIT_NOFILE soft) we turn off the first two,
1127 and keep only the latter two. A set of build-time options
1128 (-Dbump-proc-sys-fs-file-max=false and -Dbump-proc-sys-fs-nr-open=false)
1129 has been added to revert this change in behaviour, which might be
1130 an option for systems that turn off memcg in the kernel.
1132 * When no /etc/locale.conf file exists (and hence no locale settings
1133 are in place), systemd will now use the "C.UTF-8" locale by default,
1134 and set LANG= to it. This locale is supported by various
1135 distributions including Fedora, with clear indications that upstream
1136 glibc is going to make it available too. This locale enables UTF-8
1137 mode by default, which appears appropriate for 2018.
1139 * The "net.ipv4.conf.all.rp_filter" sysctl will now be set to 2 by
1140 default. This effectively switches the RFC3704 Reverse Path filtering
1141 from Strict mode to Loose mode. This is more appropriate for hosts
1142 that have multiple links with routes to the same networks (e.g.
1143 a client with a Wi-Fi and Ethernet both connected to the internet).
1145 Consult the kernel documentation for details on this sysctl:
1146 https://www.kernel.org/doc/Documentation/networking/ip-sysctl.txt
1148 * CPUAccounting=yes no longer enables the CPU controller when using
1149 kernel 4.15+ and the unified cgroup hierarchy, as required accounting
1150 statistics are now provided independently from the CPU controller.
1152 * Support for disabling a particular cgroup controller within a sub-tree
1153 has been added through the DisableControllers= directive.
1155 * cgroup_no_v1=all on the kernel command line now also implies
1156 using the unified cgroup hierarchy, unless one explicitly passes
1157 systemd.unified_cgroup_hierarchy=0 on the kernel command line.
1159 * The new "MemoryMin=" unit file property may now be used to set the
1160 memory usage protection limit of processes invoked by the unit. This
1161 controls the cgroup v2 memory.min attribute. Similarly, the new
1162 "IODeviceLatencyTargetSec=" property has been added, wrapping the new
1163 cgroup v2 io.latency cgroup property for configuring per-service I/O
1166 * systemd now supports the cgroup v2 devices BPF logic, as counterpart
1167 to the cgroup v1 "devices" cgroup controller.
1169 * systemd-escape now is able to combine --unescape with --template. It
1170 also learnt a new option --instance for extracting and unescaping the
1171 instance part of a unit name.
1173 * sd-bus now provides the sd_bus_message_readv() which is similar to
1174 sd_bus_message_read() but takes a va_list object. The pair
1175 sd_bus_set_method_call_timeout() and sd_bus_get_method_call_timeout()
1176 has been added for configuring the default method call timeout to
1177 use. sd_bus_error_move() may be used to efficiently move the contents
1178 from one sd_bus_error structure to another, invalidating the
1179 source. sd_bus_set_close_on_exit() and sd_bus_get_close_on_exit() may
1180 be used to control whether a bus connection object is automatically
1181 flushed when an sd-event loop is exited.
1183 * When processing classic BSD syslog log messages, journald will now
1184 save the original time-stamp string supplied in the new
1185 SYSLOG_TIMESTAMP= journal field. This permits consumers to
1186 reconstruct the original BSD syslog message more correctly.
1188 * StandardOutput=/StandardError= in service files gained support for
1189 new "append:…" parameters, for connecting STDOUT/STDERR of a service
1190 to a file, and appending to it.
1192 * The signal to use as last step of killing of unit processes is now
1193 configurable. Previously it was hard-coded to SIGKILL, which may now
1194 be overridden with the new KillSignal= setting. Note that this is the
1195 signal used when regular termination (i.e. SIGTERM) does not suffice.
1196 Similarly, the signal used when aborting a program in case of a
1197 watchdog timeout may now be configured too (WatchdogSignal=).
1199 * The XDG_SESSION_DESKTOP environment variable may now be configured in
1200 the pam_systemd argument line, using the new desktop= switch. This is
1201 useful to initialize it properly from a display manager without
1202 having to touch C code.
1204 * Most configuration options that previously accepted percentage values
1205 now also accept permille values with the '‰' suffix (instead of '%').
1207 * systemd-resolved may now optionally use OpenSSL instead of GnuTLS for
1210 * systemd-resolved's configuration file resolved.conf gained a new
1211 option ReadEtcHosts= which may be used to turn off processing and
1212 honoring /etc/hosts entries.
1214 * The "--wait" switch may now be passed to "systemctl
1215 is-system-running", in which case the tool will synchronously wait
1216 until the system finished start-up.
1218 * hostnamed gained a new bus call to determine the DMI product UUID.
1220 * On x86-64 systemd will now prefer using the RDRAND processor
1221 instruction over /dev/urandom whenever it requires randomness that
1222 neither has to be crypto-grade nor should be reproducible. This
1223 should substantially reduce the amount of entropy systemd requests
1224 from the kernel during initialization on such systems, though not
1225 reduce it to zero. (Why not zero? systemd still needs to allocate
1226 UUIDs and such uniquely, which require high-quality randomness.)
1228 * networkd gained support for Foo-Over-UDP, ERSPAN and ISATAP
1229 tunnels. It also gained a new option ForceDHCPv6PDOtherInformation=
1230 for forcing the "Other Information" bit in IPv6 RA messages. The
1231 bonding logic gained four new options AdActorSystemPriority=,
1232 AdUserPortKey=, AdActorSystem= for configuring various 802.3ad
1233 aspects, and DynamicTransmitLoadBalancing= for enabling dynamic
1234 shuffling of flows. The tunnel logic gained a new
1235 IPv6RapidDeploymentPrefix= option for configuring IPv6 Rapid
1236 Deployment. The policy rule logic gained four new options IPProtocol=,
1237 SourcePort= and DestinationPort=, InvertRule=. The bridge logic gained
1238 support for the MulticastToUnicast= option. networkd also gained
1239 support for configuring static IPv4 ARP or IPv6 neighbor entries.
1241 * .preset files (as read by 'systemctl preset') may now be used to
1242 instantiate services.
1244 * /etc/crypttab now understands the sector-size= option to configure
1245 the sector size for an encrypted partition.
1247 * Key material for encrypted disks may now be placed on a formatted
1248 medium, and referenced from /etc/crypttab by the UUID of the file
1249 system, followed by "=" suffixed by the path to the key file.
1251 * The "collect" udev component has been removed without replacement, as
1252 it is neither used nor maintained.
1254 * When the RuntimeDirectory=, StateDirectory=, CacheDirectory=,
1255 LogsDirectory=, ConfigurationDirectory= settings are used in a
1256 service the executed processes will now receive a set of environment
1257 variables containing the full paths of these directories.
1258 Specifically, RUNTIME_DIRECTORY=, STATE_DIRECTORY, CACHE_DIRECTORY,
1259 LOGS_DIRECTORY, CONFIGURATION_DIRECTORY are now set if these options
1260 are used. Note that these options may be used multiple times per
1261 service in which case the resulting paths will be concatenated and
1262 separated by colons.
1264 * Predictable interface naming has been extended to cover InfiniBand
1265 NICs. They will be exposed with an "ib" prefix.
1267 * tmpfiles.d/ line types may now be suffixed with a '-' character, in
1268 which case the respective line failing is ignored.
1270 * .link files may now be used to configure the equivalent to the
1271 "ethtool advertise" commands.
1273 * The sd-device.h and sd-hwdb.h APIs are now exported, as an
1274 alternative to libudev.h. Previously, the latter was just an internal
1275 wrapper around the former, but now these two APIs are exposed
1278 * sd-id128.h gained a new function sd_id128_get_boot_app_specific()
1279 which calculates an app-specific boot ID similar to how
1280 sd_id128_get_machine_app_specific() generates an app-specific machine
1283 * A new tool systemd-id128 has been added that can be used to determine
1284 and generate various 128bit IDs.
1286 * /etc/os-release gained two new standardized fields DOCUMENTATION_URL=
1289 * systemd-hibernate-resume-generator will now honor the "noresume"
1290 kernel command line option, in which case it will bypass resuming
1291 from any hibernated image.
1293 * The systemd-sleep.conf configuration file gained new options
1294 AllowSuspend=, AllowHibernation=, AllowSuspendThenHibernate=,
1295 AllowHybridSleep= for prohibiting specific sleep modes even if the
1296 kernel exports them.
1298 * portablectl is now officially supported and has thus moved to
1301 * bootctl learnt the two new commands "set-default" and "set-oneshot"
1302 for setting the default boot loader item to boot to (either
1303 persistently or only for the next boot). This is currently only
1304 compatible with sd-boot, but may be implemented on other boot loaders
1305 too, that follow the boot loader interface. The updated interface is
1306 now documented here:
1308 https://systemd.io/BOOT_LOADER_INTERFACE
1310 * A new kernel command line option systemd.early_core_pattern= is now
1311 understood which may be used to influence the core_pattern PID 1
1312 installs during early boot.
1314 * busctl learnt two new options -j and --json= for outputting method
1315 call replies, properties and monitoring output in JSON.
1317 * journalctl's JSON output now supports simple ANSI coloring as well as
1318 a new "json-seq" mode for generating RFC7464 output.
1320 * Unit files now support the %g/%G specifiers that resolve to the UNIX
1321 group/GID of the service manager runs as, similar to the existing
1322 %u/%U specifiers that resolve to the UNIX user/UID.
1324 * systemd-logind learnt a new global configuration option
1325 UserStopDelaySec= that may be set in logind.conf. It specifies how
1326 long the systemd --user instance shall remain started after a user
1327 logs out. This is useful to speed up repetitive re-connections of the
1328 same user, as it means the user's service manager doesn't have to be
1329 stopped/restarted on each iteration, but can be reused between
1330 subsequent options. This setting defaults to 10s. systemd-logind also
1331 exports two new properties on its Manager D-Bus objects indicating
1332 whether the system's lid is currently closed, and whether the system
1335 * systemd gained support for a generic boot counting logic, which
1336 generically permits automatic reverting to older boot loader entries
1337 if newer updated ones don't work. The boot loader side is implemented
1338 in sd-boot, but is kept open for other boot loaders too. For details
1341 https://systemd.io/AUTOMATIC_BOOT_ASSESSMENT
1343 * The SuccessAction=/FailureAction= unit file settings now learnt two
1344 new parameters: "exit" and "exit-force", which result in immediate
1345 exiting of the service manager, and are only useful in systemd --user
1346 and container environments.
1348 * Unit files gained support for a pair of options
1349 FailureActionExitStatus=/SuccessActionExitStatus= for configuring the
1350 exit status to use as service manager exit status when
1351 SuccessAction=/FailureAction= is set to exit or exit-force.
1353 * A pair of LogRateLimitIntervalSec=/LogRateLimitBurst= per-service
1354 options may now be used to configure the log rate limiting applied by
1355 journald per-service.
1357 * systemd-analyze gained a new verb "timespan" for parsing and
1358 normalizing time span values (i.e. strings like "5min 7s 8us").
1360 * systemd-analyze also gained a new verb "security" for analyzing the
1361 security and sand-boxing settings of services in order to determine an
1362 "exposure level" for them, indicating whether a service would benefit
1363 from more sand-boxing options turned on for them.
1365 * "systemd-analyze syscall-filter" will now also show system calls
1366 supported by the local kernel but not included in any of the defined
1369 * .nspawn files now understand the Ephemeral= setting, matching the
1370 --ephemeral command line switch.
1372 * sd-event gained the new APIs sd_event_source_get_floating() and
1373 sd_event_source_set_floating() for controlling whether a specific
1374 event source is "floating", i.e. destroyed along with the even loop
1377 * Unit objects on D-Bus gained a new "Refs" property that lists all
1378 clients that currently have a reference on the unit (to ensure it is
1381 * The JoinControllers= option in system.conf is no longer supported, as
1382 it didn't work correctly, is hard to support properly, is legacy (as
1383 the concept only exists on cgroup v1) and apparently wasn't used.
1385 * Journal messages that are generated whenever a unit enters the failed
1386 state are now tagged with a unique MESSAGE_ID. Similarly, messages
1387 generated whenever a service process exits are now made recognizable,
1388 too. A tagged message is also emitted whenever a unit enters the
1389 "dead" state on success.
1391 * systemd-run gained a new switch --working-directory= for configuring
1392 the working directory of the service to start. A shortcut -d is
1393 equivalent, setting the working directory of the service to the
1394 current working directory of the invoking program. The new --shell
1395 (or just -S) option has been added for invoking the $SHELL of the
1396 caller as a service, and implies --pty --same-dir --wait --collect
1397 --service-type=exec. Or in other words, "systemd-run -S" is now the
1398 quickest way to quickly get an interactive in a fully clean and
1399 well-defined system service context.
1401 * machinectl gained a new verb "import-fs" for importing an OS tree
1402 from a directory. Moreover, when a directory or tarball is imported
1403 and single top-level directory found with the OS itself below the OS
1404 tree is automatically mangled and moved one level up.
1406 * systemd-importd will no longer set up an implicit btrfs loop-back
1407 file system on /var/lib/machines. If one is already set up, it will
1408 continue to be used.
1410 * A new generator "systemd-run-generator" has been added. It will
1411 synthesize a unit from one or more program command lines included in
1412 the kernel command line. This is very useful in container managers
1415 # systemd-nspawn -i someimage.raw -b systemd.run='"some command line"'
1417 This will run "systemd-nspawn" on an image, invoke the specified
1418 command line and immediately shut down the container again, returning
1419 the command line's exit code.
1421 * The block device locking logic is now documented:
1423 https://systemd.io/BLOCK_DEVICE_LOCKING
1425 * loginctl and machinectl now optionally output the various tables in
1426 JSON using the --output= switch. It is our intention to add similar
1427 support to systemctl and all other commands.
1429 * udevadm's query and trigger verb now optionally take a .device unit
1432 * systemd-udevd's network naming logic now understands a new
1433 net.naming-scheme= kernel command line switch, which may be used to
1434 pick a specific version of the naming scheme. This helps stabilizing
1435 interface names even as systemd/udev are updated and the naming logic
1438 * sd-id128.h learnt two new auxiliary helpers: sd_id128_is_allf() and
1439 SD_ID128_ALLF to test if a 128bit ID is set to all 0xFF bytes, and to
1440 initialize one to all 0xFF.
1442 * After loading the SELinux policy systemd will now recursively relabel
1443 all files and directories listed in
1444 /run/systemd/relabel-extra.d/*.relabel (which should be simple
1445 newline separated lists of paths) in addition to the ones it already
1446 implicitly relabels in /run, /dev and /sys. After the relabelling is
1447 completed the *.relabel files (and /run/systemd/relabel-extra.d/) are
1448 removed. This is useful to permit initrds (i.e. code running before
1449 the SELinux policy is in effect) to generate files in the host
1450 filesystem safely and ensure that the correct label is applied during
1451 the transition to the host OS.
1453 * KERNEL API BREAKAGE: Linux kernel 4.18 changed behaviour regarding
1454 mknod() handling in user namespaces. Previously mknod() would always
1455 fail with EPERM in user namespaces. Since 4.18 mknod() will succeed
1456 but device nodes generated that way cannot be opened, and attempts to
1457 open them result in EPERM. This breaks the "graceful fallback" logic
1458 in systemd's PrivateDevices= sand-boxing option. This option is
1459 implemented defensively, so that when systemd detects it runs in a
1460 restricted environment (such as a user namespace, or an environment
1461 where mknod() is blocked through seccomp or absence of CAP_SYS_MKNOD)
1462 where device nodes cannot be created the effect of PrivateDevices= is
1463 bypassed (following the logic that 2nd-level sand-boxing is not
1464 essential if the system systemd runs in is itself already sand-boxed
1465 as a whole). This logic breaks with 4.18 in container managers where
1466 user namespacing is used: suddenly PrivateDevices= succeeds setting
1467 up a private /dev/ file system containing devices nodes — but when
1468 these are opened they don't work.
1470 At this point it is recommended that container managers utilizing
1471 user namespaces that intend to run systemd in the payload explicitly
1472 block mknod() with seccomp or similar, so that the graceful fallback
1475 We are very sorry for the breakage and the requirement to change
1476 container configurations for newer kernels. It's purely caused by an
1477 incompatible kernel change. The relevant kernel developers have been
1478 notified about this userspace breakage quickly, but they chose to
1481 * PermissionsStartOnly= setting is deprecated (but is still supported
1482 for backwards compatibility). The same functionality is provided by
1483 the more flexible "+", "!", and "!!" prefixes to ExecStart= and other
1486 * $DBUS_SESSION_BUS_ADDRESS environment variable is not set by
1487 pam_systemd anymore.
1489 * The naming scheme for network devices was changed to always rename
1490 devices, even if they were already renamed by userspace. The "kernel"
1491 policy was changed to only apply as a fallback, if no other naming
1494 * The requirements to build systemd is bumped to meson-0.46 and
1497 Contributions from: afg, Alan Jenkins, Aleksei Timofeyev, Alexander
1498 Filippov, Alexander Kurtz, Alexey Bogdanenko, Andreas Henriksson,
1499 Andrew Jorgensen, Anita Zhang, apnix-uk, Arkan49, Arseny Maslennikov,
1500 asavah, Asbjørn Apeland, aszlig, Bastien Nocera, Ben Boeckel, Benedikt
1501 Morbach, Benjamin Berg, Bruce Zhang, Carlo Caione, Cedric Viou, Chen
1502 Qi, Chris Chiu, Chris Down, Chris Morin, Christian Rebischke, Claudius
1503 Ellsel, Colin Guthrie, dana, Daniel, Daniele Medri, Daniel Kahn
1504 Gillmor, Daniel Rusek, Daniel van Vugt, Dariusz Gadomski, Dave Reisner,
1505 David Anderson, Davide Cavalca, David Leeds, David Malcolm, David
1506 Strauss, David Tardon, Dimitri John Ledkov, Dmitry Torokhov, dj-kaktus,
1507 Dongsu Park, Elias Probst, Emil Soleyman, Erik Kooistra, Ervin Peters,
1508 Evgeni Golov, Evgeny Vereshchagin, Fabrice Fontaine, Faheel Ahmad,
1509 Faizal Luthfi, Felix Yan, Filipe Brandenburger, Franck Bui, Frank
1510 Schaefer, Frantisek Sumsal, Gautier Husson, Gianluca Boiano, Giuseppe
1511 Scrivano, glitsj16, Hans de Goede, Harald Hoyer, Harry Mallon, Harshit
1512 Jain, Helmut Grohne, Henry Tung, Hui Yiqun, imayoda, Insun Pyo, Iwan
1513 Timmer, Jan Janssen, Jan Pokorný, Jan Synacek, Jason A. Donenfeld,
1514 javitoom, Jérémy Nouhaud, Jeremy Su, Jiuyang Liu, João Paulo Rechi
1515 Vita, Joe Hershberger, Joe Rayhawk, Joerg Behrmann, Joerg Steffens,
1516 Jonas Dorel, Jon Ringle, Josh Soref, Julian Andres Klode, Jun Bo Bi,
1517 Jürg Billeter, Keith Busch, Khem Raj, Kirill Marinushkin, Larry
1518 Bernstone, Lennart Poettering, Lion Yang, Li Song, Lorenz
1519 Hübschle-Schneider, Lubomir Rintel, Lucas Werkmeister, Ludwin Janvier,
1520 Lukáš Nykrýn, Luke Shumaker, mal, Marc-Antoine Perennou, Marcin
1521 Skarbek, Marco Trevisan (Treviño), Marian Cepok, Mario Hros, Marko
1522 Myllynen, Markus Grimm, Martin Pitt, Martin Sobotka, Martin Wilck,
1523 Mathieu Trudel-Lapierre, Matthew Leeds, Michael Biebl, Michael Olbrich,
1524 Michael 'pbone' Pobega, Michael Scherer, Michal Koutný, Michal
1525 Sekletar, Michal Soltys, Mike Gilbert, Mike Palmer, Muhammet Kara, Neal
1526 Gompa, Neil Brown, Network Silence, Niklas Tibbling, Nikolas Nyby,
1527 Nogisaka Sadata, Oliver Smith, Patrik Flykt, Pavel Hrdina, Paweł
1528 Szewczyk, Peter Hutterer, Piotr Drąg, Ray Strode, Reinhold Mueller,
1529 Renaud Métrich, Roman Gushchin, Ronny Chevalier, Rubén Suárez Alvarez,
1530 Ruixin Bao, RussianNeuroMancer, Ryutaroh Matsumoto, Saleem Rashid, Sam
1531 Morris, Samuel Morris, Sandy Carter, scootergrisen, Sébastien Bacher,
1532 Sergey Ptashnick, Shawn Landden, Shengyao Xue, Shih-Yuan Lee
1533 (FourDollars), Silvio Knizek, Sjoerd Simons, Stasiek Michalski, Stephen
1534 Gallagher, Steven Allen, Steve Ramage, Susant Sahani, Sven Joachim,
1535 Sylvain Plantefève, Tanu Kaskinen, Tejun Heo, Thiago Macieira, Thomas
1536 Blume, Thomas Haller, Thomas H. P. Andersen, Tim Ruffing, TJ, Tobias
1537 Jungel, Todd Walton, Tommi Rantala, Tomsod M, Tony Novak, Tore
1538 Anderson, Trevonn, Victor Laskurain, Victor Tapia, Violet Halo, Vojtech
1539 Trefny, welaq, William A. Kennington III, William Douglas, Wyatt Ward,
1540 Xiang Fan, Xi Ruoyao, Xuanwo, Yann E. Morin, YmrDtnJu, Yu Watanabe,
1541 Zbigniew Jędrzejewski-Szmek, Zhang Xianwei, Zsolt Dollenstein
1543 — Warsaw, 2018-12-21
1547 * NETWORK INTERFACE DEVICE NAMING CHANGES: systemd-udevd's "net_id"
1548 builtin will name network interfaces differently than in previous
1549 versions for virtual network interfaces created with SR-IOV and NPAR
1550 and for devices where the PCI network controller device does not have
1551 a slot number associated.
1553 SR-IOV virtual devices are now named based on the name of the parent
1554 interface, with a suffix of "v<N>", where <N> is the virtual device
1555 number. Previously those virtual devices were named as if completely
1558 The ninth and later NPAR virtual devices will be named following the
1559 scheme used for the first eight NPAR partitions. Previously those
1560 devices were not renamed and the kernel default (eth<n>) was used.
1562 "net_id" will also generate names for PCI devices where the PCI
1563 network controller device does not have an associated slot number
1564 itself, but one of its parents does. Previously those devices were
1565 not renamed and the kernel default (eth<n>) was used.
1567 * AF_INET and AF_INET6 are dropped from RestrictAddressFamilies= in
1568 systemd-logind.service. Since v235, IPAddressDeny=any has been set to
1569 the unit. So, it is expected that the default behavior of
1570 systemd-logind is not changed. However, if distribution packagers or
1571 administrators disabled or modified IPAddressDeny= setting by a
1572 drop-in config file, then it may be necessary to update the file to
1573 re-enable AF_INET and AF_INET6 to support network user name services,
1576 * When the RestrictNamespaces= unit property is specified multiple
1577 times, then the specified types are merged now. Previously, only the
1578 last assignment was used. So, if distribution packagers or
1579 administrators modified the setting by a drop-in config file, then it
1580 may be necessary to update the file.
1582 * When OnFailure= is used in combination with Restart= on a service
1583 unit, then the specified units will no longer be triggered on
1584 failures that result in restarting. Previously, the specified units
1585 would be activated each time the unit failed, even when the unit was
1586 going to be restarted automatically. This behaviour contradicted the
1587 documentation. With this release the code is adjusted to match the
1590 * systemd-tmpfiles will now print a notice whenever it encounters
1591 tmpfiles.d/ lines referencing the /var/run/ directory. It will
1592 recommend reworking them to use the /run/ directory instead (for
1593 which /var/run/ is simply a symlinked compatibility alias). This way
1594 systemd-tmpfiles can properly detect line conflicts and merge lines
1595 referencing the same file by two paths, without having to access
1598 * systemctl disable/unmask/preset/preset-all cannot be used with
1599 --runtime. Previously this was allowed, but resulted in unintuitive
1600 behaviour that wasn't useful. systemctl disable/unmask will now undo
1601 both runtime and persistent enablement/masking, i.e. it will remove
1602 any relevant symlinks both in /run and /etc.
1604 * Note that all long-running system services shipped with systemd will
1605 now default to a system call whitelist (rather than a blacklist, as
1606 before). In particular, systemd-udevd will now enforce one too. For
1607 most cases this should be safe, however downstream distributions
1608 which disabled sandboxing of systemd-udevd (specifically the
1609 MountFlags= setting), might want to disable this security feature
1610 too, as the default whitelisting will prohibit all mount, swap,
1611 reboot and clock changing operations from udev rules.
1613 * sd-boot acquired new loader configuration settings to optionally turn
1614 off Windows and MacOS boot partition discovery as well as
1615 reboot-into-firmware menu items. It is also able to pick a better
1616 screen resolution for HiDPI systems, and now provides loader
1617 configuration settings to change the resolution explicitly.
1619 * systemd-resolved now supports DNS-over-TLS. It's still
1620 turned off by default, use DNSOverTLS=opportunistic to turn it on in
1621 resolved.conf. We intend to make this the default as soon as couple
1622 of additional techniques for optimizing the initial latency caused by
1623 establishing a TLS/TCP connection are implemented.
1625 * systemd-resolved.service and systemd-networkd.service now set
1626 DynamicUser=yes. The users systemd-resolve and systemd-network are
1627 not created by systemd-sysusers anymore.
1629 NOTE: This has a chance of breaking nss-ldap and similar NSS modules
1630 that embed a network facing module into any process using getpwuid()
1631 or related call: the dynamic allocation of the user ID for
1632 systemd-resolved.service means the service manager has to check NSS
1633 if the user name is already taken when forking off the service. Since
1634 the user in the common case won't be defined in /etc/passwd the
1635 lookup is likely to trigger nss-ldap which in turn might use NSS to
1636 ask systemd-resolved for hostname lookups. This will hence result in
1637 a deadlock: a user name lookup in order to start
1638 systemd-resolved.service will result in a host name lookup for which
1639 systemd-resolved.service needs to be started already. There are
1640 multiple ways to work around this problem: pre-allocate the
1641 "systemd-resolve" user on such systems, so that nss-ldap won't be
1642 triggered; or use a different NSS package that doesn't do networking
1643 in-process but provides a local asynchronous name cache; or configure
1644 the NSS package to avoid lookups for UIDs in the range `pkg-config
1645 systemd --variable=dynamicuidmin` … `pkg-config systemd
1646 --variable=dynamicuidmax`, so that it does not consider itself
1647 authoritative for the same UID range systemd allocates dynamic users
1650 * The systemd-resolve tool has been renamed to resolvectl (it also
1651 remains available under the old name, for compatibility), and its
1652 interface is now verb-based, similar in style to the other <xyz>ctl
1653 tools, such as systemctl or loginctl.
1655 * The resolvectl/systemd-resolve tool also provides 'resolvconf'
1656 compatibility. It may be symlinked under the 'resolvconf' name, in
1657 which case it will take arguments and input compatible with the
1658 Debian and FreeBSD resolvconf tool.
1660 * Support for suspend-then-hibernate has been added, i.e. a sleep mode
1661 where the system initially suspends, and after a timeout resumes and
1664 * networkd's ClientIdentifier= now accepts a new option "duid-only". If
1665 set the client will only send a DUID as client identifier.
1667 * The nss-systemd glibc NSS module will now enumerate dynamic users and
1668 groups in effect. Previously, it could resolve UIDs/GIDs to user
1669 names/groups and vice versa, but did not support enumeration.
1671 * journald's Compress= configuration setting now optionally accepts a
1672 byte threshold value. All journal objects larger than this threshold
1673 will be compressed, smaller ones will not. Previously this threshold
1674 was not configurable and set to 512.
1676 * A new system.conf setting NoNewPrivileges= is now available which may
1677 be used to turn off acquisition of new privileges system-wide
1678 (i.e. set Linux' PR_SET_NO_NEW_PRIVS for PID 1 itself, and thus also
1679 for all its children). Note that turning this option on means setuid
1680 binaries and file system capabilities lose their special powers.
1681 While turning on this option is a big step towards a more secure
1682 system, doing so is likely to break numerous pre-existing UNIX tools,
1683 in particular su and sudo.
1685 * A new service systemd-time-sync-wait.service has been added. If
1686 enabled it will delay the time-sync.target unit at boot until time
1687 synchronization has been received from the network. This
1688 functionality is useful on systems lacking a local RTC or where it is
1689 acceptable that the boot process shall be delayed by external network
1692 * When hibernating, systemd will now inform the kernel of the image
1693 write offset, on kernels new enough to support this. This means swap
1694 files should work for hibernation now.
1696 * When loading unit files, systemd will now look for drop-in unit files
1697 extensions in additional places. Previously, for a unit file name
1698 "foo-bar-baz.service" it would look for dropin files in
1699 "foo-bar-baz.service.d/*.conf". Now, it will also look in
1700 "foo-bar-.service.d/*.conf" and "foo-.service.d/", i.e. at the
1701 service name truncated after all inner dashes. This scheme allows
1702 writing drop-ins easily that apply to a whole set of unit files at
1703 once. It's particularly useful for mount and slice units (as their
1704 naming is prefix based), but is also useful for service and other
1705 units, for packages that install multiple unit files at once,
1706 following a strict naming regime of beginning the unit file name with
1707 the package's name. Two new specifiers are now supported in unit
1708 files to match this: %j and %J are replaced by the part of the unit
1709 name following the last dash.
1711 * Unit files and other configuration files that support specifier
1712 expansion now understand another three new specifiers: %T and %V will
1713 resolve to /tmp and /var/tmp respectively, or whatever temporary
1714 directory has been set for the calling user. %E will expand to either
1715 /etc (for system units) or $XDG_CONFIG_HOME (for user units).
1717 * The ExecStart= lines of unit files are no longer required to
1718 reference absolute paths. If non-absolute paths are specified the
1719 specified binary name is searched within the service manager's
1720 built-in $PATH, which may be queried with 'systemd-path
1721 search-binaries-default'. It's generally recommended to continue to
1722 use absolute paths for all binaries specified in unit files.
1724 * Units gained a new load state "bad-setting", which is used when a
1725 unit file was loaded, but contained fatal errors which prevent it
1726 from being started (for example, a service unit has been defined
1727 lacking both ExecStart= and ExecStop= lines).
1729 * coredumpctl's "gdb" verb has been renamed to "debug", in order to
1730 support alternative debuggers, for example lldb. The old name
1731 continues to be available however, for compatibility reasons. Use the
1732 new --debugger= switch or the $SYSTEMD_DEBUGGER environment variable
1733 to pick an alternative debugger instead of the default gdb.
1735 * systemctl and the other tools will now output escape sequences that
1736 generate proper clickable hyperlinks in various terminal emulators
1737 where useful (for example, in the "systemctl status" output you can
1738 now click on the unit file name to quickly open it in the
1739 editor/viewer of your choice). Note that not all terminal emulators
1740 support this functionality yet, but many do. Unfortunately, the
1741 "less" pager doesn't support this yet, hence this functionality is
1742 currently automatically turned off when a pager is started (which
1743 happens quite often due to auto-paging). We hope to remove this
1744 limitation as soon as "less" learns these escape sequences. This new
1745 behaviour may also be turned off explicitly with the $SYSTEMD_URLIFY
1746 environment variable. For details on these escape sequences see:
1747 https://gist.github.com/egmontkob/eb114294efbcd5adb1944c9f3cb5feda
1749 * networkd's .network files now support a new IPv6MTUBytes= option for
1750 setting the MTU used by IPv6 explicitly as well as a new MTUBytes=
1751 option in the [Route] section to configure the MTU to use for
1752 specific routes. It also gained support for configuration of the DHCP
1753 "UserClass" option through the new UserClass= setting. It gained
1754 three new options in the new [CAN] section for configuring CAN
1755 networks. The MULTICAST and ALLMULTI interface flags may now be
1756 controlled explicitly with the new Multicast= and AllMulticast=
1759 * networkd will now automatically make use of the kernel's route
1760 expiration feature, if it is available.
1762 * udevd's .link files now support setting the number of receive and
1763 transmit channels, using the RxChannels=, TxChannels=,
1764 OtherChannels=, CombinedChannels= settings.
1766 * Support for UDPSegmentationOffload= has been removed, given its
1767 limited support in hardware, and waning software support.
1769 * networkd's .netdev files now support creating "netdevsim" interfaces.
1771 * PID 1 learnt a new bus call GetUnitByControlGroup() which may be used
1772 to query the unit belonging to a specific kernel control group.
1774 * systemd-analyze gained a new verb "cat-config", which may be used to
1775 dump the contents of any configuration file, with all its matching
1776 drop-in files added in, and honouring the usual search and masking
1777 logic applied to systemd configuration files. For example use
1778 "systemd-analyze cat-config systemd/system.conf" to get the complete
1779 system configuration file of systemd how it would be loaded by PID 1
1780 itself. Similar to this, various tools such as systemd-tmpfiles or
1781 systemd-sysusers, gained a new option "--cat-config", which does the
1782 corresponding operation for their own configuration settings. For
1783 example, "systemd-tmpfiles --cat-config" will now output the full
1784 list of tmpfiles.d/ lines in place.
1786 * timedatectl gained three new verbs: "show" shows bus properties of
1787 systemd-timedated, "timesync-status" shows the current NTP
1788 synchronization state of systemd-timesyncd, and "show-timesync"
1789 shows bus properties of systemd-timesyncd.
1791 * systemd-timesyncd gained a bus interface on which it exposes details
1794 * A new environment variable $SYSTEMD_TIMEDATED_NTP_SERVICES is now
1795 understood by systemd-timedated. It takes a colon-separated list of
1796 unit names of NTP client services. The list is used by
1797 "timedatectl set-ntp".
1799 * systemd-nspawn gained a new --rlimit= switch for setting initial
1800 resource limits for the container payload. There's a new switch
1801 --hostname= to explicitly override the container's hostname. A new
1802 --no-new-privileges= switch may be used to control the
1803 PR_SET_NO_NEW_PRIVS flag for the container payload. A new
1804 --oom-score-adjust= switch controls the OOM scoring adjustment value
1805 for the payload. The new --cpu-affinity= switch controls the CPU
1806 affinity of the container payload. The new --resolv-conf= switch
1807 allows more detailed control of /etc/resolv.conf handling of the
1808 container. Similarly, the new --timezone= switch allows more detailed
1809 control of /etc/localtime handling of the container.
1811 * systemd-detect-virt gained a new --list switch, which will print a
1812 list of all currently known VM and container environments.
1814 * Support for "Portable Services" has been added, see
1815 doc/PORTABLE_SERVICES.md for details. Currently, the support is still
1816 experimental, but this is expected to change soon. Reflecting this
1817 experimental state, the "portablectl" binary is not installed into
1818 /usr/bin yet. The binary has to be called with the full path
1819 /usr/lib/systemd/portablectl instead.
1821 * journalctl's and systemctl's -o switch now knows a new log output
1822 mode "with-unit". The output it generates is very similar to the
1823 regular "short" mode, but displays the unit name instead of the
1824 syslog tag for each log line. Also, the date is shown with timezone
1825 information. This mode is probably more useful than the classic
1826 "short" output mode for most purposes, except where pixel-perfect
1827 compatibility with classic /var/log/messages formatting is required.
1829 * A new --dump-bus-properties switch has been added to the systemd
1830 binary, which may be used to dump all supported D-Bus properties.
1831 (Options which are still supported, but are deprecated, are *not*
1834 * sd-bus gained a set of new calls:
1835 sd_bus_slot_set_floating()/sd_bus_slot_get_floating() may be used to
1836 enable/disable the "floating" state of a bus slot object,
1837 i.e. whether the slot object pins the bus it is allocated for into
1838 memory or if the bus slot object gets disconnected when the bus goes
1839 away. sd_bus_open_with_description(),
1840 sd_bus_open_user_with_description(),
1841 sd_bus_open_system_with_description() may be used to allocate bus
1842 objects and set their description string already during allocation.
1844 * sd-event gained support for watching inotify events from the event
1845 loop, in an efficient way, sharing inotify handles between multiple
1846 users. For this a new function sd_event_add_inotify() has been added.
1848 * sd-event and sd-bus gained support for calling special user-supplied
1849 destructor functions for userdata pointers associated with
1850 sd_event_source, sd_bus_slot, and sd_bus_track objects. For this new
1851 functions sd_bus_slot_set_destroy_callback,
1852 sd_bus_slot_get_destroy_callback, sd_bus_track_set_destroy_callback,
1853 sd_bus_track_get_destroy_callback,
1854 sd_event_source_set_destroy_callback,
1855 sd_event_source_get_destroy_callback have been added.
1857 * The "net.ipv4.tcp_ecn" sysctl will now be turned on by default.
1859 * PID 1 will now automatically reschedule .timer units whenever the
1860 local timezone changes. (They previously got rescheduled
1861 automatically when the system clock changed.)
1863 * New documentation has been added to document cgroups delegation,
1864 portable services and the various code quality tools we have set up:
1866 https://github.com/systemd/systemd/blob/master/docs/CGROUP_DELEGATION.md
1867 https://github.com/systemd/systemd/blob/master/docs/PORTABLE_SERVICES.md
1868 https://github.com/systemd/systemd/blob/master/docs/CODE_QUALITY.md
1870 * The Boot Loader Specification has been added to the source tree.
1872 https://github.com/systemd/systemd/blob/master/docs/BOOT_LOADER_SPECIFICATION.md
1874 While moving it into our source tree we have updated it and further
1875 changes are now accepted through the usual github PR workflow.
1877 * pam_systemd will now look for PAM userdata fields systemd.memory_max,
1878 systemd.tasks_max, systemd.cpu_weight, systemd.io_weight set by
1879 earlier PAM modules. The data in these fields is used to initialize
1880 the session scope's resource properties. Thus external PAM modules
1881 may now configure per-session limits, for example sourced from
1882 external user databases.
1884 * socket units with Accept=yes will now maintain a "refused" counter in
1885 addition to the existing "accepted" counter, counting connections
1886 refused due to the enforced limits.
1888 * The "systemd-path search-binaries-default" command may now be use to
1889 query the default, built-in $PATH PID 1 will pass to the services it
1892 * A new unit file setting PrivateMounts= has been added. It's a boolean
1893 option. If enabled the unit's processes are invoked in their own file
1894 system namespace. Note that this behaviour is also implied if any
1895 other file system namespacing options (such as PrivateTmp=,
1896 PrivateDevices=, ProtectSystem=, …) are used. This option is hence
1897 primarily useful for services that do not use any of the other file
1898 system namespacing options. One such service is systemd-udevd.service
1899 where this is now used by default.
1901 * ConditionSecurity= gained a new value "uefi-secureboot" that is true
1902 when the system is booted in UEFI "secure mode".
1904 * A new unit "system-update-pre.target" is added, which defines an
1905 optional synchronization point for offline system updates, as
1906 implemented by the pre-existing "system-update.target" unit. It
1907 allows ordering services before the service that executes the actual
1908 update process in a generic way.
1910 * Systemd now emits warnings whenever .include syntax is used.
1912 Contributions from: Adam Duskett, Alan Jenkins, Alessandro Casale,
1913 Alexander Kurtz, Alex Gartrell, Anssi Hannula, Arnaud Rebillout, Brian
1914 J. Murrell, Bruno Vernay, Chris Lamb, Chris Lesiak, Christian Brauner,
1915 Christian Hesse, Christian Rebischke, Colin Guthrie, Daniel Dao, Daniel
1916 Lin, Danylo Korostil, Davide Cavalca, David Tardon, Dimitri John
1917 Ledkov, Dmitriy Geels, Douglas Christman, Elia Geretto, emelenas, Emil
1918 Velikov, Evgeny Vereshchagin, Felipe Sateler, Feng Sun, Filipe
1919 Brandenburger, Franck Bui, futpib, Giuseppe Scrivano, Guillem Jover,
1920 guixxx, Hannes Reinecke, Hans de Goede, Harald Hoyer, Henrique Dante de
1921 Almeida, Hiram van Paassen, Ian Miell, Igor Gnatenko, Ivan Shapovalov,
1922 Iwan Timmer, James Cowgill, Jan Janssen, Jan Synacek, Jared Kazimir,
1923 Jérémy Rosen, João Paulo Rechi Vita, Joost Heitbrink, Jui-Chi Ricky
1924 Liang, Jürg Billeter, Kai-Heng Feng, Karol Augustin, Kay Sievers,
1925 Krzysztof Nowicki, Lauri Tirkkonen, Lennart Poettering, Leonard König,
1926 Long Li, Luca Boccassi, Lucas Werkmeister, Marcel Hoppe, Marc
1927 Kleine-Budde, Mario Limonciello, Martin Jansa, Martin Wilck, Mathieu
1928 Malaterre, Matteo F. Vescovi, Matthew McGinn, Matthias-Christian Ott,
1929 Michael Biebl, Michael Olbrich, Michael Prokop, Michal Koutný, Michal
1930 Sekletar, Mike Gilbert, Mikhail Kasimov, Milan Broz, Milan Pässler,
1931 Mladen Pejaković, Muhammet Kara, Nicolas Boichat, Omer Katz, Paride
1932 Legovini, Paul Menzel, Paul Milliken, Pavel Hrdina, Peter A. Bigot,
1933 Peter D'Hoye, Peter Hutterer, Peter Jones, Philip Sequeira, Philip
1934 Withnall, Piotr Drąg, Radostin Stoyanov, Ricardo Salveti de Araujo,
1935 Ronny Chevalier, Rosen Penev, Rubén Suárez Alvarez, Ryan Gonzalez,
1936 Salvo Tomaselli, Sebastian Reichel, Sergey Ptashnick, Sergio Lindo
1937 Mansilla, Stefan Schweter, Stephen Hemminger, Stuart Hayes, Susant
1938 Sahani, Sylvain Plantefève, Thomas H. P. Andersen, Tobias Jungel,
1939 Tomasz Torcz, Vito Caputo, Will Dietz, Will Thompson, Wim van Mourik,
1940 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
1942 — Berlin, 2018-06-22
1946 * The MemoryAccounting= unit property now defaults to on. After
1947 discussions with the upstream control group maintainers we learnt
1948 that the negative impact of cgroup memory accounting on current
1949 kernels is finally relatively minimal, so that it should be safe to
1950 enable this by default without affecting system performance. Besides
1951 memory accounting only task accounting is turned on by default, all
1952 other forms of resource accounting (CPU, IO, IP) remain off for now,
1953 because it's not clear yet that their impact is small enough to move
1954 from opt-in to opt-out. We recommend downstreams to leave memory
1955 accounting on by default if kernel 4.14 or higher is primarily
1956 used. On very resource constrained systems or when support for old
1957 kernels is a necessity, -Dmemory-accounting-default=false can be used
1958 to revert this change.
1960 * rpm scriptlets to update the udev hwdb and rules (%udev_hwdb_update,
1961 %udev_rules_update) and the journal catalog (%journal_catalog_update)
1962 from the upgrade scriptlets of individual packages now do nothing.
1963 Transfiletriggers have been added which will perform those updates
1964 once at the end of the transaction.
1966 Similar transfiletriggers have been added to execute any sysctl.d
1967 and binfmt.d rules. Thus, it should be unnecessary to provide any
1968 scriptlets to execute this configuration from package installation
1971 * systemd-sysusers gained a mode where the configuration to execute is
1972 specified on the command line, but this configuration is not executed
1973 directly, but instead it is merged with the configuration on disk,
1974 and the result is executed. This is useful for package installation
1975 scripts which want to create the user before installing any files on
1976 disk (in case some of those files are owned by that user), while
1977 still allowing local admin overrides.
1979 This functionality is exposed to rpm scriptlets through a new
1980 %sysusers_create_package macro. Old %sysusers_create and
1981 %sysusers_create_inline macros are deprecated.
1983 A transfiletrigger for sysusers.d configuration is now installed,
1984 which means that it should be unnecessary to call systemd-sysusers from
1985 package installation scripts, unless the package installs any files
1986 owned by those newly-created users, in which case
1987 %sysusers_create_package should be used.
1989 * Analogous change has been done for systemd-tmpfiles: it gained a mode
1990 where the command-line configuration is merged with the configuration
1991 on disk. This is exposed as the new %tmpfiles_create_package macro,
1992 and %tmpfiles_create is deprecated. A transfiletrigger is installed
1993 for tmpfiles.d, hence it should be unnecessary to call systemd-tmpfiles
1994 from package installation scripts.
1996 * sysusers.d configuration for a user may now also specify the group
1997 number, in addition to the user number ("u username 123:456"), or
1998 without the user number ("u username -:456").
2000 * Configution items for systemd-sysusers can now be specified as
2001 positional arguments when the new --inline switch is used.
2003 * The login shell of users created through sysusers.d may now be
2004 specified (previously, it was always /bin/sh for root and
2005 /sbin/nologin for other users).
2007 * systemd-analyze gained a new --global switch to look at global user
2008 configuration. It also gained a unit-paths verb to list the unit load
2009 paths that are compiled into systemd (which can be used with
2010 --systemd, --user, or --global).
2012 * udevadm trigger gained a new --settle/-w option to wait for any
2013 triggered events to finish (but just those, and not any other events
2014 which are triggered meanwhile).
2016 * The action that systemd-logind takes when the lid is closed and the
2017 machine is connected to external power can now be configured using
2018 HandleLidSwitchExternalPower= in logind.conf. Previously, this action
2019 was determined by HandleLidSwitch=, and, for backwards compatibility,
2020 is still is, if HandleLidSwitchExternalPower= is not explicitly set.
2022 * journalctl will periodically call sd_journal_process() to make it
2023 resilient against inotify queue overruns when journal files are
2024 rotated very quickly.
2026 * Two new functions in libsystemd — sd_bus_get_n_queued_read and
2027 sd_bus_get_n_queued_write — may be used to check the number of
2028 pending bus messages.
2030 * systemd gained a new
2031 org.freedesktop.systemd1.Manager.AttachProcessesToUnit dbus call
2032 which can be used to migrate foreign processes to scope and service
2033 units. The primary user for this new API is systemd itself: the
2034 systemd --user instance uses this call of the systemd --system
2035 instance to migrate processes if it itself gets the request to
2036 migrate processes and the kernel refuses this due to access
2037 restrictions. Thanks to this "systemd-run --scope --user …" works
2038 again in pure cgroup v2 environments when invoked from the user
2041 * A new TemporaryFileSystem= setting can be used to mask out part of
2042 the real file system tree with tmpfs mounts. This may be combined
2043 with BindPaths= and BindReadOnlyPaths= to hide files or directories
2044 not relevant to the unit, while still allowing some paths lower in
2045 the tree to be accessed.
2047 ProtectHome=tmpfs may now be used to hide user home and runtime
2048 directories from units, in a way that is mostly equivalent to
2049 "TemporaryFileSystem=/home /run/user /root".
2051 * Non-service units are now started with KeyringMode=shared by default.
2052 This means that mount and swapon and other mount tools have access
2053 to keys in the main keyring.
2055 * /sys/fs/bpf is now mounted automatically.
2057 * QNX virtualization is now detected by systemd-detect-virt and may
2058 be used in ConditionVirtualization=.
2060 * IPAccounting= may now be enabled also for slice units.
2062 * A new -Dsplit-bin= build configuration switch may be used to specify
2063 whether bin and sbin directories are merged, or if they should be
2064 included separately in $PATH and various listings of executable
2065 directories. The build configuration scripts will try to autodetect
2066 the proper values of -Dsplit-usr= and -Dsplit-bin= based on build
2067 system, but distributions are encouraged to configure this
2070 * A new -Dok-color= build configuration switch may be used to change
2071 the colour of "OK" status messages.
2073 * UPGRADE ISSUE: serialization of units using JoinsNamespaceOf= with
2074 PrivateNetwork=yes was buggy in previous versions of systemd. This
2075 means that after the upgrade and daemon-reexec, any such units must
2078 * INCOMPATIBILITY: as announced in the NEWS for 237, systemd-tmpfiles
2079 will not exclude read-only files owned by root from cleanup.
2081 Contributions from: Alan Jenkins, Alexander F Rødseth, Alexis Jeandet,
2082 Andika Triwidada, Andrei Gherzan, Ansgar Burchardt, antizealot1337,
2083 Batuhan Osman Taşkaya, Beniamino Galvani, Bill Yodlowsky, Caio Marcelo
2084 de Oliveira Filho, CuBiC, Daniele Medri, Daniel Mouritzen, Daniel
2085 Rusek, Davide Cavalca, Dimitri John Ledkov, Douglas Christman, Evgeny
2086 Vereshchagin, Faalagorn, Filipe Brandenburger, Franck Bui, futpib,
2087 Giacomo Longo, Gunnar Hjalmarsson, Hans de Goede, Hermann Gausterer,
2088 Iago López Galeiras, Jakub Filak, Jan Synacek, Jason A. Donenfeld,
2089 Javier Martinez Canillas, Jérémy Rosen, Lennart Poettering, Lucas
2090 Werkmeister, Mao Huang, Marco Gulino, Michael Biebl, Michael Vogt,
2091 MilhouseVH, Neal Gompa (ニール・ゴンパ), Oleander Reis, Olof Mogren,
2092 Patrick Uiterwijk, Peter Hutterer, Peter Portante, Piotr Drąg, Robert
2093 Antoni Buj Gelonch, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
2094 Fowler, SjonHortensius, snorreflorre, Susant Sahani, Sylvain
2095 Plantefève, Thomas Blume, Thomas Haller, Vito Caputo, Yu Watanabe,
2096 Zbigniew Jędrzejewski-Szmek, Марко М. Костић (Marko M. Kostić)
2098 — Warsaw, 2018-03-05
2102 * Some keyboards come with a zoom see-saw or rocker which until now got
2103 mapped to the Linux "zoomin/out" keys in hwdb. However, these
2104 keycodes are not recognized by any major desktop. They now produce
2105 Up/Down key events so that they can be used for scrolling.
2107 * INCOMPATIBILITY: systemd-tmpfiles' "f" lines changed behaviour
2108 slightly: previously, if an argument was specified for lines of this
2109 type (i.e. the right-most column was set) this string was appended to
2110 existing files each time systemd-tmpfiles was run. This behaviour was
2111 different from what the documentation said, and not particularly
2112 useful, as repeated systemd-tmpfiles invocations would not be
2113 idempotent and grow such files without bounds. With this release
2114 behaviour has been altered to match what the documentation says:
2115 lines of this type only have an effect if the indicated files don't
2116 exist yet, and only then the argument string is written to the file.
2118 * FUTURE INCOMPATIBILITY: In systemd v238 we intend to slightly change
2119 systemd-tmpfiles behaviour: previously, read-only files owned by root
2120 were always excluded from the file "aging" algorithm (i.e. the
2121 automatic clean-up of directories like /tmp based on
2122 atime/mtime/ctime). We intend to drop this restriction, and age files
2123 by default even when owned by root and read-only. This behaviour was
2124 inherited from older tools, but there have been requests to remove
2125 it, and it's not obvious why this restriction was made in the first
2126 place. Please speak up now, if you are aware of software that reqires
2127 this behaviour, otherwise we'll remove the restriction in v238.
2129 * A new environment variable $SYSTEMD_OFFLINE is now understood by
2130 systemctl. It takes a boolean argument. If on, systemctl assumes it
2131 operates on an "offline" OS tree, and will not attempt to talk to the
2132 service manager. Previously, this mode was implicitly enabled if a
2133 chroot() environment was detected, and this new environment variable
2134 now provides explicit control.
2136 * .path and .socket units may now be created transiently, too.
2137 Previously only service, mount, automount and timer units were
2138 supported as transient units. The systemd-run tool has been updated
2139 to expose this new functionality, you may hence use it now to bind
2140 arbitrary commands to path or socket activation on-the-fly from the
2141 command line. Moreover, almost all properties are now exposed for the
2142 unit types that already supported transient operation.
2144 * The systemd-mount command gained support for a new --owner= parameter
2145 which takes a user name, which is then resolved and included in uid=
2146 and gid= mount options string of the file system to mount.
2148 * A new unit condition ConditionControlGroupController= has been added
2149 that checks whether a specific cgroup controller is available.
2151 * Unit files, udev's .link files, and systemd-networkd's .netdev and
2152 .network files all gained support for a new condition
2153 ConditionKernelVersion= for checking against specific kernel
2156 * In systemd-networkd, the [IPVLAN] section in .netdev files gained
2157 support for configuring device flags in the Flags= setting. In the
2158 same files, the [Tunnel] section gained support for configuring
2159 AllowLocalRemote=. The [Route] section in .network files gained
2160 support for configuring InitialCongestionWindow=,
2161 InitialAdvertisedReceiveWindow= and QuickAck=. The [DHCP] section now
2162 understands RapidCommit=.
2164 * systemd-networkd's DHCPv6 support gained support for Prefix
2167 * sd-bus gained support for a new "watch-bind" feature. When this
2168 feature is enabled, an sd_bus connection may be set up to connect to
2169 an AF_UNIX socket in the file system as soon as it is created. This
2170 functionality is useful for writing early-boot services that
2171 automatically connect to the system bus as soon as it is started,
2172 without ugly time-based polling. systemd-networkd and
2173 systemd-resolved have been updated to make use of this
2174 functionality. busctl exposes this functionality in a new
2175 --watch-bind= command line switch.
2177 * sd-bus will now optionally synthesize a local "Connected" signal as
2178 soon as a D-Bus connection is set up fully. This message mirrors the
2179 already existing "Disconnected" signal which is synthesized when the
2180 connection is terminated. This signal is generally useful but
2181 particularly handy in combination with the "watch-bind" feature
2182 described above. Synthesizing of this message has to be requested
2183 explicitly through the new API call sd_bus_set_connected_signal(). In
2184 addition a new call sd_bus_is_ready() has been added that checks
2185 whether a connection is fully set up (i.e. between the "Connected" and
2186 "Disconnected" signals).
2188 * sd-bus gained two new calls sd_bus_request_name_async() and
2189 sd_bus_release_name_async() for asynchronously registering bus
2190 names. Similar, there is now sd_bus_add_match_async() for installing
2191 a signal match asynchronously. All of systemd's own services have
2192 been updated to make use of these calls. Doing these operations
2193 asynchronously has two benefits: it reduces the risk of deadlocks in
2194 case of cyclic dependencies between bus services, and it speeds up
2195 service initialization since synchronization points for bus
2196 round-trips are removed.
2198 * sd-bus gained two new calls sd_bus_match_signal() and
2199 sd_bus_match_signal_async(), which are similar to sd_bus_add_match()
2200 and sd_bus_add_match_async() but instead of taking a D-Bus match
2201 string take match fields as normal function parameters.
2203 * sd-bus gained two new calls sd_bus_set_sender() and
2204 sd_bus_message_set_sender() for setting the sender name of outgoing
2205 messages (either for all outgoing messages or for just one specific
2206 one). These calls are only useful in direct connections as on
2207 brokered connections the broker fills in the sender anyway,
2208 overwriting whatever the client filled in.
2210 * sd-event gained a new pseudo-handle that may be specified on all API
2211 calls where an "sd_event*" object is expected: SD_EVENT_DEFAULT. When
2212 used this refers to the default event loop object of the calling
2213 thread. Note however that this does not implicitly allocate one —
2214 which has to be done prior by using sd_event_default(). Similarly
2215 sd-bus gained three new pseudo-handles SD_BUS_DEFAULT,
2216 SD_BUS_DEFAULT_USER, SD_BUS_DEFAULT_SYSTEM that may be used to refer
2217 to the default bus of the specified type of the calling thread. Here
2218 too this does not implicitly allocate bus connection objects, this
2219 has to be done prior with sd_bus_default() and friends.
2221 * sd-event gained a new call pair
2222 sd_event_source_{get|set}_io_fd_own(). This may be used to request
2223 automatic closure of the file descriptor an IO event source watches
2224 when the event source is destroyed.
2226 * systemd-networkd gained support for natively configuring WireGuard
2229 * In previous versions systemd synthesized user records both for the
2230 "nobody" (UID 65534) and "root" (UID 0) users in nss-systemd and
2231 internally. In order to simplify distribution-wide renames of the
2232 "nobody" user (like it is planned in Fedora: nfsnobody → nobody), a
2233 new transitional flag file has been added: if
2234 /etc/systemd/dont-synthesize-nobody exists synthesizing of the 65534
2235 user and group record within the systemd codebase is disabled.
2237 * systemd-notify gained a new --uid= option for selecting the source
2238 user/UID to use for notification messages sent to the service
2241 * journalctl gained a new --grep= option to list only entries in which
2242 the message matches a certain pattern. By default matching is case
2243 insensitive if the pattern is lowercase, and case sensitive
2244 otherwise. Option --case-sensitive=yes|no can be used to override
2245 this an specify case sensitivity or case insensitivity.
2247 * There's now a "systemd-analyze service-watchdogs" command for printing
2248 the current state of the service runtime watchdog, and optionally
2249 enabling or disabling the per-service watchdogs system-wide if given a
2250 boolean argument (i.e. the concept you configure in WatchdogSec=), for
2251 debugging purposes. There's also a kernel command line option
2252 systemd.service_watchdogs= for controlling the same.
2254 * Two new "log-level" and "log-target" options for systemd-analyze were
2255 added that merge the now deprecated get-log-level, set-log-level and
2256 get-log-target, set-log-target pairs. The deprecated options are still
2257 understood for backwards compatibility. The two new options print the
2258 current value when no arguments are given, and set them when a
2259 level/target is given as an argument.
2261 * sysusers.d's "u" lines now optionally accept both a UID and a GID
2262 specification, separated by a ":" character, in order to create users
2263 where UID and GID do not match.
2265 Contributions from: Adam Duskett, Alan Jenkins, Alexander Kuleshov,
2266 Alexis Deruelle, Andrew Jeddeloh, Armin Widegreen, Batuhan Osman
2267 Taşkaya, Björn Esser, bleep_blop, Bruce A. Johnson, Chris Down, Clinton
2268 Roy, Colin Walters, Daniel Rusek, Dimitri John Ledkov, Dmitry Rozhkov,
2269 Evgeny Vereshchagin, Ewout van Mansom, Felipe Sateler, Franck Bui,
2270 Frantisek Sumsal, George Gaydarov, Gianluca Boiano, Hans-Christian
2271 Noren Egtvedt, Hans de Goede, Henrik Grindal Bakken, Jan Alexander
2272 Steffens, Jan Klötzke, Jason A. Donenfeld, jdkbx, Jérémy Rosen,
2273 Jerónimo Borque, John Lin, John Paul Herold, Jonathan Rudenberg, Jörg
2274 Thalheim, Ken (Bitsko) MacLeod, Larry Bernstone, Lennart Poettering,
2275 Lucas Werkmeister, Maciej S. Szmigiero, Marek Čermák, Martin Pitt,
2276 Mathieu Malaterre, Matthew Thode, Matthias-Christian Ott, Max Harmathy,
2277 Michael Biebl, Michael Vogt, Michal Koutný, Michal Sekletar, Michał
2278 Szczepański, Mike Gilbert, Nathaniel McCallum, Nicolas Chauvet, Olaf
2279 Hering, Olivier Schwander, Patrik Flykt, Paul Cercueil, Peter Hutterer,
2280 Piotr Drąg, Raphael Vogelgsang, Reverend Homer, Robert Kolchmeyer,
2281 Samuel Dionne-Riel, Sergey Ptashnick, Shawn Landden, Susant Sahani,
2282 Sylvain Plantefève, Thomas H. P. Andersen, Thomas Huth, Tomasz
2283 Bachorski, Vladislav Vishnyakov, Wieland Hoffmann, Yu Watanabe, Zachary
2284 Winnerman, Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски, Дилян
2291 * The modprobe.d/ drop-in for the bonding.ko kernel module introduced
2292 in v235 has been extended to also set the dummy.ko module option
2293 numdummies=0, preventing the kernel from automatically creating
2294 dummy0. All dummy interfaces must now be explicitly created.
2296 * Unknown '%' specifiers in configuration files are now rejected. This
2297 applies to units and tmpfiles.d configuration. Any percent characters
2298 that are followed by a letter or digit that are not supposed to be
2299 interpreted as the beginning of a specifier should be escaped by
2300 doubling ("%%"). (So "size=5%" is still accepted, as well as
2301 "size=5%,foo=bar", but not "LABEL=x%y%z" since %y and %z are not
2302 valid specifiers today.)
2304 * systemd-resolved now maintains a new dynamic
2305 /run/systemd/resolve/stub-resolv.conf compatibility file. It is
2306 recommended to make /etc/resolv.conf a symlink to it. This file
2307 points at the systemd-resolved stub DNS 127.0.0.53 resolver and
2308 includes dynamically acquired search domains, achieving more correct
2309 DNS resolution by software that bypasses local DNS APIs such as NSS.
2311 * The "uaccess" udev tag has been dropped from /dev/kvm and
2312 /dev/dri/renderD*. These devices now have the 0666 permissions by
2313 default (but this may be changed at build-time). /dev/dri/renderD*
2314 will now be owned by the "render" group along with /dev/kfd.
2316 * "DynamicUser=yes" has been enabled for systemd-timesyncd.service,
2317 systemd-journal-gatewayd.service and
2318 systemd-journal-upload.service. This means "nss-systemd" must be
2319 enabled in /etc/nsswitch.conf to ensure the UIDs assigned to these
2320 services are resolved properly.
2322 * In /etc/fstab two new mount options are now understood:
2323 x-systemd.makefs and x-systemd.growfs. The former has the effect that
2324 the configured file system is formatted before it is mounted, the
2325 latter that the file system is resized to the full block device size
2326 after it is mounted (i.e. if the file system is smaller than the
2327 partition it resides on, it's grown). This is similar to the fsck
2328 logic in /etc/fstab, and pulls in systemd-makefs@.service and
2329 systemd-growfs@.service as necessary, similar to
2330 systemd-fsck@.service. Resizing is currently only supported on ext4
2333 * In systemd-networkd, the IPv6 RA logic now optionally may announce
2334 DNS server and domain information.
2336 * Support for the LUKS2 on-disk format for encrypted partitions has
2337 been added. This requires libcryptsetup2 during compilation and
2340 * The systemd --user instance will now signal "readiness" when its
2341 basic.target unit has been reached, instead of when the run queue ran
2342 empty for the first time.
2344 * Tmpfiles.d with user configuration are now also supported.
2345 systemd-tmpfiles gained a new --user switch, and snippets placed in
2346 ~/.config/user-tmpfiles.d/ and corresponding directories will be
2347 executed by systemd-tmpfiles --user running in the new
2348 systemd-tmpfiles-setup.service and systemd-tmpfiles-clean.service
2349 running in the user session.
2351 * Unit files and tmpfiles.d snippets learnt three new % specifiers:
2352 %S resolves to the top-level state directory (/var/lib for the system
2353 instance, $XDG_CONFIG_HOME for the user instance), %C resolves to the
2354 top-level cache directory (/var/cache for the system instance,
2355 $XDG_CACHE_HOME for the user instance), %L resolves to the top-level
2356 logs directory (/var/log for the system instance,
2357 $XDG_CONFIG_HOME/log/ for the user instance). This matches the
2358 existing %t specifier, that resolves to the top-level runtime
2359 directory (/run for the system instance, and $XDG_RUNTIME_DIR for the
2362 * journalctl learnt a new parameter --output-fields= for limiting the
2363 set of journal fields to output in verbose and JSON output modes.
2365 * systemd-timesyncd's configuration file gained a new option
2366 RootDistanceMaxSec= for setting the maximum root distance of servers
2367 it'll use, as well as the new options PollIntervalMinSec= and
2368 PollIntervalMaxSec= to tweak the minimum and maximum poll interval.
2370 * bootctl gained a new command "list" for listing all available boot
2371 menu items on systems that follow the boot loader specification.
2373 * systemctl gained a new --dry-run switch that shows what would be done
2374 instead of doing it, and is currently supported by the shutdown and
2377 * ConditionSecurity= can now detect the TOMOYO security module.
2379 * Unit file [Install] sections are now also respected in unit drop-in
2380 files. This is intended to be used by drop-ins under /usr/lib/.
2382 * systemd-firstboot may now also set the initial keyboard mapping.
2384 * Udev "changed" events for devices which are exposed as systemd
2385 .device units are now propagated to units specified in
2386 ReloadPropagatedFrom= as reload requests.
2388 * If a udev device has a SYSTEMD_WANTS= property containing a systemd
2389 unit template name (i.e. a name in the form of 'foobar@.service',
2390 without the instance component between the '@' and - the '.'), then
2391 the escaped sysfs path of the device is automatically used as the
2394 * SystemCallFilter= in unit files has been extended so that an "errno"
2395 can be specified individually for each system call. Example:
2396 SystemCallFilter=~uname:EILSEQ.
2398 * The cgroup delegation logic has been substantially updated. Delegate=
2399 now optionally takes a list of controllers (instead of a boolean, as
2400 before), which lists the controllers to delegate at least.
2402 * The networkd DHCPv6 client now implements the FQDN option (RFC 4704).
2404 * A new LogLevelMax= setting configures the maximum log level any
2405 process of the service may log at (i.e. anything with a lesser
2406 priority than what is specified is automatically dropped). A new
2407 LogExtraFields= setting allows configuration of additional journal
2408 fields to attach to all log records generated by any of the unit's
2411 * New StandardInputData= and StandardInputText= settings along with the
2412 new option StandardInput=data may be used to configure textual or
2413 binary data that shall be passed to the executed service process via
2414 standard input, encoded in-line in the unit file.
2416 * StandardInput=, StandardOutput= and StandardError= may now be used to
2417 connect stdin/stdout/stderr of executed processes directly with a
2418 file or AF_UNIX socket in the file system, using the new "file:" option.
2420 * A new unit file option CollectMode= has been added, that allows
2421 tweaking the garbage collection logic for units. It may be used to
2422 tell systemd to garbage collect units that have failed automatically
2423 (normally it only GCs units that exited successfully). systemd-run
2424 and systemd-mount expose this new functionality with a new -G option.
2426 * "machinectl bind" may now be used to bind mount non-directories
2427 (i.e. regularfiles, devices, fifos, sockets).
2429 * systemd-analyze gained a new verb "calendar" for validating and
2430 testing calendar time specifications to use for OnCalendar= in timer
2431 units. Besides validating the expression it will calculate the next
2432 time the specified expression would elapse.
2434 * In addition to the pre-existing FailureAction= unit file setting
2435 there's now SuccessAction=, for configuring a shutdown action to
2436 execute when a unit completes successfully. This is useful in
2437 particular inside containers that shall terminate after some workload
2438 has been completed. Also, both options are now supported for all unit
2439 types, not just services.
2441 * networkds's IP rule support gained two new options
2442 IncomingInterface= and OutgoingInterface= for configuring the incoming
2443 and outgoing interfaces of configured rules. systemd-networkd also
2444 gained support for "vxcan" network devices.
2446 * networkd gained a new setting RequiredForOnline=, taking a
2447 boolean. If set, systemd-wait-online will take it into consideration
2448 when determining that the system is up, otherwise it will ignore the
2449 interface for this purpose.
2451 * The sd_notify() protocol gained support for a new operation: with
2452 FDSTOREREMOVE=1 file descriptors may be removed from the per-service
2453 store again, ahead of POLLHUP or POLLERR when they are removed
2456 * A new document doc/UIDS-GIDS.md has been added to the source tree,
2457 that documents the UID/GID range and assignment assumptions and
2458 requirements of systemd.
2460 * The watchdog device PID 1 will ping may now be configured through the
2461 WatchdogDevice= configuration file setting, or by setting the
2462 systemd.watchdog_service= kernel commandline option.
2464 * systemd-resolved's gained support for registering DNS-SD services on
2465 the local network using MulticastDNS. Services may either be
2466 registered by dropping in a .dnssd file in /etc/systemd/dnssd/ (or
2467 the same dir below /run, /usr/lib), or through its D-Bus API.
2469 * The sd_notify() protocol can now with EXTEND_TIMEOUT_USEC=microsecond
2470 extend the effective start, runtime, and stop time. The service must
2471 continue to send EXTEND_TIMEOUT_USEC within the period specified to
2472 prevent the service manager from making the service as timedout.
2474 * systemd-resolved's DNSSEC support gained support for RFC 8080
2475 (Ed25519 keys and signatures).
2477 * The systemd-resolve command line tool gained a new set of options
2478 --set-dns=, --set-domain=, --set-llmnr=, --set-mdns=, --set-dnssec=,
2479 --set-nta= and --revert to configure per-interface DNS configuration
2480 dynamically during runtime. It's useful for pushing DNS information
2481 into systemd-resolved from DNS hook scripts that various interface
2482 managing software supports (such as pppd).
2484 * systemd-nspawn gained a new --network-namespace-path= command line
2485 option, which may be used to make a container join an existing
2486 network namespace, by specifying a path to a "netns" file.
2488 Contributions from: Alan Jenkins, Alan Robertson, Alessandro Ghedini,
2489 Andrew Jeddeloh, Antonio Rojas, Ari, asavah, bleep_blop, Carsten
2490 Strotmann, Christian Brauner, Christian Hesse, Clinton Roy, Collin
2491 Eggert, Cong Wang, Daniel Black, Daniel Lockyer, Daniel Rusek, Dimitri
2492 John Ledkov, Dmitry Rozhkov, Dongsu Park, Edward A. James, Evgeny
2493 Vereshchagin, Florian Klink, Franck Bui, Gwendal Grignou, Hans de
2494 Goede, Harald Hoyer, Hristo Venev, Iago López Galeiras, Ikey Doherty,
2495 Jakub Wilk, Jérémy Rosen, Jiahui Xie, John Lin, José Bollo, Josef
2496 Andersson, juga0, Krzysztof Nowicki, Kyle Walker, Lars Karlitski, Lars
2497 Kellogg-Stedman, Lauri Tirkkonen, Lennart Poettering, Lubomir Rintel,
2498 Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn, Lukáš Říha, Lukasz
2499 Rubaszewski, Maciej S. Szmigiero, Mantas Mikulėnas, Marcus Folkesson,
2500 Martin Steuer, Mathieu Trudel-Lapierre, Matija Skala,
2501 Matthias-Christian Ott, Max Resch, Michael Biebl, Michael Vogt, Michal
2502 Koutný, Michal Sekletar, Mike Gilbert, Muhammet Kara, Neil Brown, Olaf
2503 Hering, Ondrej Kozina, Patrik Flykt, Patryk Kocielnik, Peter Hutterer,
2504 Piotr Drąg, Razvan Cojocaru, Robin McCorkell, Roland Hieber, Saran
2505 Tunyasuvunakool, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
2506 Arlott, Simon Peeters, Stanislav Angelovič, Stefan Agner, Susant
2507 Sahani, Sylvain Plantefève, Thomas Blume, Thomas Haller, Tiago Salem
2508 Herrmann, Tinu Weber, Tom Stellard, Topi Miettinen, Torsten Hilbrich,
2509 Vito Caputo, Vladislav Vishnyakov, WaLyong Cho, Yu Watanabe, Zbigniew
2510 Jędrzejewski-Szmek, Zeal Jagannatha
2512 — Berlin, 2017-12-14
2516 * INCOMPATIBILITY: systemd-logind.service and other long-running
2517 services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP
2518 communication with the outside. This generally improves security of
2519 the system, and is in almost all cases a safe and good choice, as
2520 these services do not and should not provide any network-facing
2521 functionality. However, systemd-logind uses the glibc NSS API to
2522 query the user database. This creates problems on systems where NSS
2523 is set up to directly consult network services for user database
2524 lookups. In particular, this creates incompatibilities with the
2525 "nss-nis" module, which attempts to directly contact the NIS/YP
2526 network servers it is configured for, and will now consistently
2527 fail. In such cases, it is possible to turn off IP sandboxing for
2528 systemd-logind.service (set IPAddressDeny= in its [Service] section
2529 to the empty string, via a .d/ unit file drop-in). Downstream
2530 distributions might want to update their nss-nis packaging to include
2531 such a drop-in snippet, accordingly, to hide this incompatibility
2532 from the user. Another option is to make use of glibc's nscd service
2533 to proxy such network requests through a privilege-separated, minimal
2534 local caching daemon, or to switch to more modern technologies such
2535 sssd, whose NSS hook-ups generally do not involve direct network
2536 access. In general, we think it's definitely time to question the
2537 implementation choices of nss-nis, i.e. whether it's a good idea
2538 today to embed a network-facing loadable module into all local
2539 processes that need to query the user database, including the most
2540 trivial and benign ones, such as "ls". For more details about
2541 IPAddressDeny= see below.
2543 * A new modprobe.d drop-in is now shipped by default that sets the
2544 bonding module option max_bonds=0. This overrides the kernel default,
2545 to avoid conflicts and ambiguity as to whether or not bond0 should be
2546 managed by systemd-networkd or not. This resolves multiple issues
2547 with bond0 properties not being applied, when bond0 is configured
2548 with systemd-networkd. Distributors may choose to not package this,
2549 however in that case users will be prevented from correctly managing
2550 bond0 interface using systemd-networkd.
2552 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
2553 which print the logging level and target of the system manager. They
2554 complement the existing "set-log-level" and "set-log-target" verbs
2555 used to change those values.
2557 * journald.conf gained a new boolean setting ReadKMsg= which defaults
2558 to on. If turned off kernel log messages will not be read by
2559 systemd-journald or included in the logs. It also gained a new
2560 setting LineMax= for configuring the maximum line length in
2561 STDOUT/STDERR log streams. The new default for this value is 48K, up
2562 from the previous hardcoded 2048.
2564 * A new unit setting RuntimeDirectoryPreserve= has been added, which
2565 allows more detailed control of what to do with a runtime directory
2566 configured with RuntimeDirectory= (i.e. a directory below /run or
2567 $XDG_RUNTIME_DIR) after a unit is stopped.
2569 * The RuntimeDirectory= setting for units gained support for creating
2570 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
2571 one top-level directory.
2573 * Units gained new options StateDirectory=, CacheDirectory=,
2574 LogsDirectory= and ConfigurationDirectory= which are closely related
2575 to RuntimeDirectory= but manage per-service directories below
2576 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
2577 possible to write unit files which when activated automatically gain
2578 properly owned service specific directories in these locations, thus
2579 making unit files self-contained and increasing compatibility with
2580 stateless systems and factory reset where /etc or /var are
2581 unpopulated at boot. Matching these new settings there's also
2582 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
2583 ConfigurationDirectoryMode= for configuring the access mode of these
2584 directories. These settings are particularly useful in combination
2585 with DynamicUser=yes as they provide secure, properly-owned,
2586 writable, and stateful locations for storage, excluded from the
2587 sandbox that such services live in otherwise.
2589 * Automake support has been removed from this release. systemd is now
2592 * systemd-journald will now aggressively cache client metadata during
2593 runtime, speeding up log write performance under pressure. This comes
2594 at a small price though: as much of the metadata is read
2595 asynchronously from /proc/ (and isn't implicitly attached to log
2596 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
2597 metadata stored alongside a log entry might be slightly
2598 out-of-date. Previously it could only be slightly newer than the log
2599 message. The time window is small however, and given that the kernel
2600 is unlikely to be improved anytime soon in this regard, this appears
2603 * nss-myhostname/systemd-resolved will now by default synthesize an
2604 A/AAAA resource record for the "_gateway" hostname, pointing to the
2605 current default IP gateway. Previously it did that for the "gateway"
2606 name, hampering adoption, as some distributions wanted to leave that
2607 host name open for local use. The old behaviour may still be
2608 requested at build time.
2610 * systemd-networkd's [Address] section in .network files gained a new
2611 Scope= setting for configuring the IP address scope. The [Network]
2612 section gained a new boolean setting ConfigureWithoutCarrier= that
2613 tells systemd-networkd to ignore link sensing when configuring the
2614 device. The [DHCP] section gained a new Anonymize= boolean option for
2615 turning on a number of options suggested in RFC 7844. A new
2616 [RoutingPolicyRule] section has been added for configuring the IP
2617 routing policy. The [Route] section has gained support for a new
2618 Type= setting which permits configuring
2619 blackhole/unreachable/prohibit routes.
2621 * The [VRF] section in .netdev files gained a new Table= setting for
2622 configuring the routing table to use. The [Tunnel] section gained a
2623 new Independent= boolean field for configuring tunnels independent of
2624 an underlying network interface. The [Bridge] section gained a new
2625 GroupForwardMask= option for configuration of propagation of link
2626 local frames between bridge ports.
2628 * The WakeOnLan= setting in .link files gained support for a number of
2629 new modes. A new TCP6SegmentationOffload= setting has been added for
2630 configuring TCP/IPv6 hardware segmentation offload.
2632 * The IPv6 RA sender implementation may now optionally send out RDNSS
2633 and RDNSSL records to supply DNS configuration to peers.
2635 * systemd-nspawn gained support for a new --system-call-filter= command
2636 line option for adding and removing entries in the default system
2637 call filter it applies. Moreover systemd-nspawn has been changed to
2638 implement a system call whitelist instead of a blacklist.
2640 * systemd-run gained support for a new --pipe command line option. If
2641 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
2642 are directly passed on to the activated transient service
2643 executable. This allows invoking arbitrary processes as systemd
2644 services (for example to take benefit of dependency management,
2645 accounting management, resource management or log management that is
2646 done automatically for services) — while still allowing them to be
2647 integrated in a classic UNIX shell pipeline.
2649 * When a service sends RELOAD=1 via sd_notify() and reload propagation
2650 using ReloadPropagationTo= is configured, a reload is now propagated
2651 to configured units. (Previously this was only done on explicitly
2652 requested reloads, using "systemctl reload" or an equivalent
2655 * For each service unit a restart counter is now kept: it is increased
2656 each time the service is restarted due to Restart=, and may be
2657 queried using "systemctl show -p NRestarts …".
2659 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
2660 @signal and @timer have been added, for usage with SystemCallFilter=
2661 in unit files and the new --system-call-filter= command line option
2662 of systemd-nspawn (see above).
2664 * ExecStart= lines in unit files gained two new modifiers: when a
2665 command line is prefixed with "!" the command will be executed as
2666 configured, except for the credentials applied by
2667 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
2668 "+", but does still apply namespacing options unlike "+". There's
2669 also "!!" now, which is mostly identical, but becomes a NOP on
2670 systems that support ambient capabilities. This is useful to write
2671 unit files that work with ambient capabilities where possible but
2672 automatically fall back to traditional privilege dropping mechanisms
2673 on systems where this is not supported.
2675 * ListenNetlink= settings in socket units now support RDMA netlink
2678 * A new unit file setting LockPersonality= has been added which permits
2679 locking down the chosen execution domain ("personality") of a service
2682 * A new special target "getty-pre.target" has been added, which is
2683 ordered before all text logins, and may be used to order services
2684 before textual logins acquire access to the console.
2686 * systemd will now attempt to load the virtio-rng.ko kernel module very
2687 early on if a VM environment supporting this is detected. This should
2688 improve entropy during early boot in virtualized environments.
2690 * A _netdev option is now supported in /etc/crypttab that operates in a
2691 similar way as the same option in /etc/fstab: it permits configuring
2692 encrypted devices that need to be ordered after the network is up.
2693 Following this logic, two new special targets
2694 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
2695 added that are to cryptsetup.target what remote-fs.target and
2696 remote-fs-pre.target are to local-fs.target.
2698 * Service units gained a new UnsetEnvironment= setting which permits
2699 unsetting specific environment variables for services that are
2700 normally passed to it (for example in order to mask out locale
2701 settings for specific services that can't deal with it).
2703 * Units acquired a new boolean option IPAccounting=. When turned on, IP
2704 traffic accounting (packet count as well as byte count) is done for
2705 the service, and shown as part of "systemctl status" or "systemd-run
2708 * Service units acquired two new options IPAddressAllow= and
2709 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
2710 for configuring a simple IP access control list for all sockets of
2711 the unit. These options are available also on .slice and .socket
2712 units, permitting flexible access list configuration for individual
2713 services as well as groups of services (as defined by a slice unit),
2714 including system-wide. Note that IP ACLs configured this way are
2715 enforced on every single IPv4 and IPv6 socket created by any process
2716 of the service unit, and apply to ingress as well as egress traffic.
2718 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
2719 structured log message is generated each time the unit is stopped,
2720 containing information about the consumed resources of this
2723 * A new setting KeyringMode= has been added to unit files, which may be
2724 used to control how the kernel keyring is set up for executed
2727 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
2728 "systemctl kexec" and "systemctl exit" are now always asynchronous in
2729 behaviour (that is: these commands return immediately after the
2730 operation was enqueued instead of waiting for the operation to
2731 complete). Previously, "systemctl poweroff" and "systemctl reboot"
2732 were asynchronous on systems using systemd-logind (i.e. almost
2733 always, and like they were on sysvinit), and the other three commands
2734 were unconditionally synchronous. With this release this is cleaned
2735 up, and callers will see the same asynchronous behaviour on all
2736 systems for all five operations.
2738 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
2741 * .timer units now accept calendar specifications in other timezones
2742 than UTC or the local timezone.
2744 * The tmpfiles snippet var.conf has been changed to create
2745 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
2746 the "utmp" group already, and it appears to be generally understood
2747 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
2748 databases. Previously this was implemented correctly for all these
2749 databases excepts btmp, which has been opened up like this now
2750 too. Note that while the other databases are world-readable
2751 (i.e. 0644), btmp is not and remains more restrictive.
2753 * The systemd-resolve tool gained a new --reset-server-features
2754 switch. When invoked like this systemd-resolved will forget
2755 everything it learnt about the features supported by the configured
2756 upstream DNS servers, and restarts the feature probing logic on the
2757 next resolver look-up for them at the highest feature level
2760 * The status dump systemd-resolved sends to the logs upon receiving
2761 SIGUSR1 now also includes information about all DNS servers it is
2762 configured to use, and the features levels it probed for them.
2764 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
2765 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
2766 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
2767 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
2768 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
2769 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
2770 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
2771 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
2772 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
2773 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
2774 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
2775 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
2776 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
2777 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
2778 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
2779 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
2780 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
2781 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
2782 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
2783 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
2785 — Berlin, 2017-10-06
2789 * Meson is now supported as build system in addition to Automake. It is
2790 our plan to remove Automake in one of our next releases, so that
2791 Meson becomes our exclusive build system. Hence, please start using
2792 the Meson build system in your downstream packaging. There's plenty
2793 of documentation around how to use Meson, the extremely brief
2796 ./autogen.sh && ./configure && make && sudo make install
2800 meson build && ninja -C build && sudo ninja -C build install
2802 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
2803 which permits configuring a timeout on the time a job is
2804 running. This is particularly useful for setting timeouts on jobs for
2807 * Unit files gained two new options ConditionUser= and ConditionGroup=
2808 for conditionalizing units based on the identity of the user/group
2809 running a systemd user instance.
2811 * systemd-networkd now understands a new FlowLabel= setting in the
2812 [VXLAN] section of .network files, as well as a Priority= in
2813 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
2814 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
2815 gained support for configuration of GENEVE links, and IPv6 address
2816 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
2818 * .link files now understand a new Port= setting.
2820 * systemd-networkd's DHCP support gained support for DHCP option 119
2821 (domain search list).
2823 * systemd-networkd gained support for serving IPv6 address ranges using
2824 the Router Advertisement protocol. The new .network configuration
2825 section [IPv6Prefix] may be used to configure the ranges to
2826 serve. This is implemented based on a new, minimal, native server
2827 implementation of RA.
2829 * journalctl's --output= switch gained support for a new parameter
2830 "short-iso-precise" for a mode where timestamps are shown as precise
2833 * systemd-udevd's "net_id" builtin may now generate stable network
2834 interface names from IBM PowerVM VIO devices as well as ACPI platform
2837 * MulticastDNS support in systemd-resolved may now be explicitly
2838 enabled/disabled using the new MulticastDNS= configuration file
2841 * systemd-resolved may now optionally use libidn2 instead of the libidn
2842 for processing internationalized domain names. Support for libidn2
2843 should be considered experimental and should not be enabled by
2846 * "machinectl pull-tar" and related call may now do verification of
2847 downloaded images using SUSE-style .sha256 checksum files in addition
2848 to the already existing support for validating using Ubuntu-style
2851 * sd-bus gained support for a new sd_bus_message_appendv() call which
2852 is va_list equivalent of sd_bus_message_append().
2854 * sd-boot gained support for validating images using SHIM/MOK.
2856 * The SMACK code learnt support for "onlycap".
2858 * systemd-mount --umount is now much smarter in figuring out how to
2859 properly unmount a device given its mount or device path.
2861 * The code to call libnss_dns as a fallback from libnss_resolve when
2862 the communication with systemd-resolved fails was removed. This
2863 fallback was redundant and interfered with the [!UNAVAIL=return]
2864 suffix. See nss-resolve(8) for the recommended configuration.
2866 * systemd-logind may now be restarted without losing state. It stores
2867 the file descriptors for devices it manages in the system manager
2868 using the FDSTORE= mechanism. Please note that further changes in
2869 other components may be required to make use of this (for example
2870 Xorg has code to listen for stops of systemd-logind and terminate
2871 itself when logind is stopped or restarted, in order to avoid using
2872 stale file descriptors for graphical devices, which is now
2873 counterproductive and must be reverted in order for restarts of
2874 systemd-logind to be safe. See
2875 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
2877 * All kernel-install plugins are called with the environment variable
2878 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
2879 /etc/machine-id. If the machine ID could not be determined,
2880 $KERNEL_INSTALL_MACHINE_ID will be empty. Plugins should not put
2881 anything in the entry directory (passed as the second argument) if
2882 $KERNEL_INSTALL_MACHINE_ID is empty. For backwards compatibility, a
2883 temporary directory is passed as the entry directory and removed
2884 after all the plugins exit.
2886 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
2887 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
2888 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
2889 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
2890 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
2891 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
2892 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
2893 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
2894 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
2895 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
2896 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
2897 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
2898 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
2899 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
2900 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
2901 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
2902 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
2903 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
2904 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
2905 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
2906 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
2907 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
2908 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
2909 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
2910 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
2911 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
2912 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
2913 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
2914 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
2917 — Berlin, 2017-07-12
2921 * The "hybrid" control group mode has been modified to improve
2922 compatibility with "legacy" cgroups-v1 setups. Specifically, the
2923 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
2924 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
2925 cgroups-v1 hierarchy), the only externally visible change being that
2926 the cgroups-v2 hierarchy is also mounted, to
2927 /sys/fs/cgroup/unified. This should provide a large degree of
2928 compatibility with "legacy" cgroups-v1, while taking benefit of the
2929 better management capabilities of cgroups-v2.
2931 * The default control group setup mode may be selected both a boot-time
2932 via a set of kernel command line parameters (specifically:
2933 systemd.unified_cgroup_hierarchy= and
2934 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
2935 default selected on the configure command line
2936 (--with-default-hierarchy=). The upstream default is "hybrid"
2937 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
2938 this will change in a future systemd version to be "unified" (pure
2939 cgroups-v2 mode). The third option for the compile time option is
2940 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
2941 distributions to default to "hybrid" mode for release distributions,
2942 starting with v233. We recommend "unified" for development
2943 distributions (specifically: distributions such as Fedora's rawhide)
2944 as that's where things are headed in the long run. Use "legacy" for
2945 greatest stability and compatibility only.
2947 * Note one current limitation of "unified" and "hybrid" control group
2948 setup modes: the kernel currently does not permit the systemd --user
2949 instance (i.e. unprivileged code) to migrate processes between two
2950 disconnected cgroup subtrees, even if both are managed and owned by
2951 the user. This effectively means "systemd-run --user --scope" doesn't
2952 work when invoked from outside of any "systemd --user" service or
2953 scope. Specifically, it is not supported from session scopes. We are
2954 working on fixing this in a future systemd version. (See #3388 for
2955 further details about this.)
2957 * DBus policy files are now installed into /usr rather than /etc. Make
2958 sure your system has dbus >= 1.9.18 running before upgrading to this
2959 version, or override the install path with --with-dbuspolicydir= .
2961 * All python scripts shipped with systemd (specifically: the various
2962 tests written in Python) now require Python 3.
2964 * systemd unit tests can now run standalone (without the source or
2965 build directories), and can be installed into /usr/lib/systemd/tests/
2966 with 'make install-tests'.
2968 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
2969 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
2972 * Support for the %c, %r, %R specifiers in unit files has been
2973 removed. Specifiers are not supposed to be dependent on configuration
2974 in the unit file itself (so that they resolve the same regardless
2975 where used in the unit files), but these specifiers were influenced
2976 by the Slice= option.
2978 * The shell invoked by debug-shell.service now defaults to /bin/sh in
2979 all cases. If distributions want to use a different shell for this
2980 purpose (for example Fedora's /sbin/sushell) they need to specify
2981 this explicitly at configure time using --with-debug-shell=.
2983 * The confirmation spawn prompt has been reworked to offer the
2986 (c)ontinue, proceed without asking anymore
2987 (D)ump, show the state of the unit
2988 (f)ail, don't execute the command and pretend it failed
2990 (i)nfo, show a short summary of the unit
2991 (j)obs, show jobs that are in progress
2992 (s)kip, don't execute the command and pretend it succeeded
2993 (y)es, execute the command
2995 The 'n' choice for the confirmation spawn prompt has been removed,
2996 because its meaning was confusing.
2998 The prompt may now also be redirected to an alternative console by
2999 specifying the console as parameter to systemd.confirm_spawn=.
3001 * Services of Type=notify require a READY=1 notification to be sent
3002 during startup. If no such message is sent, the service now fails,
3003 even if the main process exited with a successful exit code.
3005 * Services that fail to start up correctly now always have their
3006 ExecStopPost= commands executed. Previously, they'd enter "failed"
3007 state directly, without executing these commands.
3009 * The option MulticastDNS= of network configuration files has acquired
3010 an actual implementation. With MulticastDNS=yes a host can resolve
3011 names of remote hosts and reply to mDNS A and AAAA requests.
3013 * When units are about to be started an additional check is now done to
3014 ensure that all dependencies of type BindsTo= (when used in
3015 combination with After=) have been started.
3017 * systemd-analyze gained a new verb "syscall-filter" which shows which
3018 system call groups are defined for the SystemCallFilter= unit file
3019 setting, and which system calls they contain.
3021 * A new system call filter group "@filesystem" has been added,
3022 consisting of various file system related system calls. Group
3023 "@reboot" has been added, covering reboot, kexec and shutdown related
3024 calls. Finally, group "@swap" has been added covering swap
3025 configuration related calls.
3027 * A new unit file option RestrictNamespaces= has been added that may be
3028 used to restrict access to the various process namespace types the
3029 Linux kernel provides. Specifically, it may be used to take away the
3030 right for a service unit to create additional file system, network,
3031 user, and other namespaces. This sandboxing option is particularly
3032 relevant due to the high amount of recently discovered namespacing
3033 related vulnerabilities in the kernel.
3035 * systemd-udev's .link files gained support for a new AutoNegotiation=
3036 setting for configuring Ethernet auto-negotiation.
3038 * systemd-networkd's .network files gained support for a new
3039 ListenPort= setting in the [DHCP] section to explicitly configure the
3040 UDP client port the DHCP client shall listen on.
3042 * .network files gained a new Unmanaged= boolean setting for explicitly
3043 excluding one or more interfaces from management by systemd-networkd.
3045 * The systemd-networkd ProxyARP= option has been renamed to
3046 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
3047 renamed to ReduceARPProxy=. The old names continue to be available
3050 * systemd-networkd gained support for configuring IPv6 Proxy NDP
3051 addresses via the new IPv6ProxyNDPAddress= .network file setting.
3053 * systemd-networkd's bonding device support gained support for two new
3054 configuration options ActiveSlave= and PrimarySlave=.
3056 * The various options in the [Match] section of .network files gained
3057 support for negative matching.
3059 * New systemd-specific mount options are now understood in /etc/fstab:
3061 x-systemd.mount-timeout= may be used to configure the maximum
3062 permitted runtime of the mount command.
3064 x-systemd.device-bound may be set to bind a mount point to its
3065 backing device unit, in order to automatically remove a mount point
3066 if its backing device is unplugged. This option may also be
3067 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
3068 on the block device, which is now automatically set for all CDROM
3069 drives, so that mounted CDs are automatically unmounted when they are
3070 removed from the drive.
3072 x-systemd.after= and x-systemd.before= may be used to explicitly
3073 order a mount after or before another unit or mount point.
3075 * Enqueued start jobs for device units are now automatically garbage
3076 collected if there are no jobs waiting for them anymore.
3078 * systemctl list-jobs gained two new switches: with --after, for every
3079 queued job the jobs it's waiting for are shown; with --before the
3080 jobs which it's blocking are shown.
3082 * systemd-nspawn gained support for ephemeral boots from disk images
3083 (or in other words: --ephemeral and --image= may now be
3084 combined). Moreover, ephemeral boots are now supported for normal
3085 directories, even if the backing file system is not btrfs. Of course,
3086 if the file system does not support file system snapshots or
3087 reflinks, the initial copy operation will be relatively expensive, but
3088 this should still be suitable for many use cases.
3090 * Calendar time specifications in .timer units now support
3091 specifications relative to the end of a month by using "~" instead of
3092 "-" as separator between month and day. For example, "*-02~03" means
3093 "the third last day in February". In addition a new syntax for
3094 repeated events has been added using the "/" character. For example,
3095 "9..17/2:00" means "every two hours from 9am to 5pm".
3097 * systemd-socket-proxyd gained a new parameter --connections-max= for
3098 configuring the maximum number of concurrent connections.
3100 * sd-id128 gained a new API for generating unique IDs for the host in a
3101 way that does not leak the machine ID. Specifically,
3102 sd_id128_get_machine_app_specific() derives an ID based on the
3103 machine ID a in well-defined, non-reversible, stable way. This is
3104 useful whenever an identifier for the host is needed but where the
3105 identifier shall not be useful to identify the system beyond the
3106 scope of the application itself. (Internally this uses HMAC-SHA256 as
3107 keyed hash function using the machine ID as input.)
3109 * NotifyAccess= gained a new supported value "exec". When set
3110 notifications are accepted from all processes systemd itself invoked,
3111 including all control processes.
3113 * .nspawn files gained support for defining overlay mounts using the
3114 Overlay= and OverlayReadOnly= options. Previously this functionality
3115 was only available on the systemd-nspawn command line.
3117 * systemd-nspawn's --bind= and --overlay= options gained support for
3118 bind/overlay mounts whose source lies within the container tree by
3119 prefixing the source path with "+".
3121 * systemd-nspawn's --bind= and --overlay= options gained support for
3122 automatically allocating a temporary source directory in /var/tmp
3123 that is removed when the container dies. Specifically, if the source
3124 directory is specified as empty string this mechanism is selected. An
3125 example usage is --overlay=+/var::/var, which creates an overlay
3126 mount based on the original /var contained in the image, overlayed
3127 with a temporary directory in the host's /var/tmp. This way changes
3128 to /var are automatically flushed when the container shuts down.
3130 * systemd-nspawn --image= option does now permit raw file system block
3131 devices (in addition to images containing partition tables, as
3134 * The disk image dissection logic in systemd-nspawn gained support for
3135 automatically setting up LUKS encrypted as well as Verity protected
3136 partitions. When a container is booted from an encrypted image the
3137 passphrase is queried at start-up time. When a container with Verity
3138 data is started, the root hash is search in a ".roothash" file
3139 accompanying the disk image (alternatively, pass the root hash via
3140 the new --root-hash= command line option).
3142 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
3143 be used to dissect disk images the same way as systemd-nspawn does
3144 it, following the Bootable Partition Specification. It may even be
3145 used to mount disk images with complex partition setups (including
3146 LUKS and Verity partitions) to a local host directory, in order to
3147 inspect them. This tool is not considered public API (yet), and is
3148 thus not installed into /usr/bin. Please do not rely on its
3149 existence, since it might go away or be changed in later systemd
3152 * A new generator "systemd-verity-generator" has been added, similar in
3153 style to "systemd-cryptsetup-generator", permitting automatic setup of
3154 Verity root partitions when systemd boots up. In order to make use of
3155 this your partition setup should follow the Discoverable Partitions
3156 Specification, and the GPT partition ID of the root file system
3157 partition should be identical to the upper 128bit of the Verity root
3158 hash. The GPT partition ID of the Verity partition protecting it
3159 should be the lower 128bit of the Verity root hash. If the partition
3160 image follows this model it is sufficient to specify a single
3161 "roothash=" kernel command line argument to both configure which root
3162 image and verity partition to use as well as the root hash for
3163 it. Note that systemd-nspawn's Verity support follows the same
3164 semantics, meaning that disk images with proper Verity data in place
3165 may be booted in containers with systemd-nspawn as well as on
3166 physical systems via the verity generator. Also note that the "mkosi"
3167 tool available at https://github.com/systemd/mkosi has been updated
3168 to generate Verity protected disk images following this scheme. In
3169 fact, it has been updated to generate disk images that optionally
3170 implement a complete UEFI SecureBoot trust chain, involving a signed
3171 kernel and initrd image that incorporates such a root hash as well as
3172 a Verity-enabled root partition.
3174 * The hardware database (hwdb) udev supports has been updated to carry
3175 accelerometer quirks.
3177 * All system services are now run with a fresh kernel keyring set up
3178 for them. The invocation ID is stored by default in it, thus
3179 providing a safe, non-overridable way to determine the invocation
3182 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
3183 options for bind mounting arbitrary paths in a service-specific
3184 way. When these options are used, arbitrary host or service files and
3185 directories may be mounted to arbitrary locations in the service's
3188 * Documentation has been added that lists all of systemd's low-level
3189 environment variables:
3191 https://github.com/systemd/systemd/blob/master/docs/ENVIRONMENT.md
3193 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
3194 whether a specific socket file descriptor matches a specified socket
3197 * systemd-firstboot has been updated to check for the
3198 systemd.firstboot= kernel command line option. It accepts a boolean
3199 and when set to false the first boot questions are skipped.
3201 * systemd-fstab-generator has been updated to check for the
3202 systemd.volatile= kernel command line option, which either takes an
3203 optional boolean parameter or the special value "state". If used the
3204 system may be booted in a "volatile" boot mode. Specifically,
3205 "systemd.volatile" is used, the root directory will be mounted as
3206 tmpfs, and only /usr is mounted from the actual root file system. If
3207 "systemd.volatile=state" is used, the root directory will be mounted
3208 as usual, but /var is mounted as tmpfs. This concept provides similar
3209 functionality as systemd-nspawn's --volatile= option, but provides it
3210 on physical boots. Use this option for implementing stateless
3211 systems, or testing systems with all state and/or configuration reset
3212 to the defaults. (Note though that many distributions are not
3213 prepared to boot up without a populated /etc or /var, though.)
3215 * systemd-gpt-auto-generator gained support for LUKS encrypted root
3216 partitions. Previously it only supported LUKS encrypted partitions
3217 for all other uses, except for the root partition itself.
3219 * Socket units gained support for listening on AF_VSOCK sockets for
3220 communication in virtualized QEMU environments.
3222 * The "configure" script gained a new option --with-fallback-hostname=
3223 for specifying the fallback hostname to use if none is configured in
3224 /etc/hostname. For example, by specifying
3225 --with-fallback-hostname=fedora it is possible to default to a
3226 hostname of "fedora" on pristine installations.
3228 * systemd-cgls gained support for a new --unit= switch for listing only
3229 the control groups of a specific unit. Similar --user-unit= has been
3230 added for listing only the control groups of a specific user unit.
3232 * systemd-mount gained a new --umount switch for unmounting a mount or
3233 automount point (and all mount/automount points below it).
3235 * systemd will now refuse full configuration reloads (via systemctl
3236 daemon-reload and related calls) unless at least 16MiB of free space
3237 are available in /run. This is a safety precaution in order to ensure
3238 that generators can safely operate after the reload completed.
3240 * A new unit file option RootImage= has been added, which has a similar
3241 effect as RootDirectory= but mounts the service's root directory from
3242 a disk image instead of plain directory. This logic reuses the same
3243 image dissection and mount logic that systemd-nspawn already uses,
3244 and hence supports any disk images systemd-nspawn supports, including
3245 those following the Discoverable Partition Specification, as well as
3246 Verity enabled images. This option enables systemd to run system
3247 services directly off disk images acting as resource bundles,
3248 possibly even including full integrity data.
3250 * A new MountAPIVFS= unit file option has been added, taking a boolean
3251 argument. If enabled /proc, /sys and /dev (collectively called the
3252 "API VFS") will be mounted for the service. This is only relevant if
3253 RootDirectory= or RootImage= is used for the service, as these mounts
3254 are of course in place in the host mount namespace anyway.
3256 * systemd-nspawn gained support for a new --pivot-root= switch. If
3257 specified the root directory within the container image is pivoted to
3258 the specified mount point, while the original root disk is moved to a
3259 different place. This option enables booting of ostree images
3260 directly with systemd-nspawn.
3262 * The systemd build scripts will no longer complain if the NTP server
3263 addresses are not changed from the defaults. Google now supports
3264 these NTP servers officially. We still recommend downstreams to
3265 properly register an NTP pool with the NTP pool project though.
3267 * coredumpctl gained a new "--reverse" option for printing the list
3268 of coredumps in reverse order.
3270 * coredumpctl will now show additional information about truncated and
3271 inaccessible coredumps, as well as coredumps that are still being
3272 processed. It also gained a new --quiet switch for suppressing
3273 additional informational message in its output.
3275 * coredumpctl gained support for only showing coredumps newer and/or
3276 older than specific timestamps, using the new --since= and --until=
3277 options, reminiscent of journalctl's options by the same name.
3279 * The systemd-coredump logic has been improved so that it may be reused
3280 to collect backtraces in non-compiled languages, for example in
3281 scripting languages such as Python.
3283 * machinectl will now show the UID shift of local containers, if user
3284 namespacing is enabled for them.
3286 * systemd will now optionally run "environment generator" binaries at
3287 configuration load time. They may be used to add environment
3288 variables to the environment block passed to services invoked. One
3289 user environment generator is shipped by default that sets up
3290 environment variables based on files dropped into /etc/environment.d
3291 and ~/.config/environment.d/.
3293 * systemd-resolved now includes the new, recently published 2017 DNSSEC
3296 * hostnamed has been updated to report a new chassis type of
3297 "convertible" to cover "foldable" laptops that can both act as a
3298 tablet and as a laptop, such as various Lenovo Yoga devices.
3300 Contributions from: Adrián López, Alexander Galanin, Alexander
3301 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
3302 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
3303 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
3304 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
3305 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
3306 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
3307 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
3308 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
3309 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
3310 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
3311 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
3312 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
3313 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
3314 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
3315 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
3316 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
3317 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
3318 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
3319 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
3320 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
3321 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
3322 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
3323 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
3324 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
3325 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
3326 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
3327 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
3330 — Berlin, 2017-03-01
3334 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
3335 RestrictAddressFamilies= enabled. These sandboxing options should
3336 generally be compatible with the various external udev call-out
3337 binaries we are aware of, however there may be exceptions, in
3338 particular when exotic languages for these call-outs are used. In
3339 this case, consider turning off these settings locally.
3341 * The new RemoveIPC= option can be used to remove IPC objects owned by
3342 the user or group of a service when that service exits.
3344 * The new ProtectKernelModules= option can be used to disable explicit
3345 load and unload operations of kernel modules by a service. In
3346 addition access to /usr/lib/modules is removed if this option is set.
3348 * ProtectSystem= option gained a new value "strict", which causes the
3349 whole file system tree with the exception of /dev, /proc, and /sys,
3350 to be remounted read-only for a service.
3352 * The new ProtectKernelTunables= option can be used to disable
3353 modification of configuration files in /sys and /proc by a service.
3354 Various directories and files are remounted read-only, so access is
3355 restricted even if the file permissions would allow it.
3357 * The new ProtectControlGroups= option can be used to disable write
3358 access by a service to /sys/fs/cgroup.
3360 * Various systemd services have been hardened with
3361 ProtectKernelTunables=yes, ProtectControlGroups=yes,
3362 RestrictAddressFamilies=.
3364 * Support for dynamically creating users for the lifetime of a service
3365 has been added. If DynamicUser=yes is specified, user and group IDs
3366 will be allocated from the range 61184..65519 for the lifetime of the
3367 service. They can be resolved using the new nss-systemd.so NSS
3368 module. The module must be enabled in /etc/nsswitch.conf. Services
3369 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
3370 any resources allocated by the service will be cleaned up when the
3371 service exits. They also have ProtectHome=read-only and
3372 ProtectSystem=strict enabled, so they are not able to make any
3373 permanent modifications to the system.
3375 * The nss-systemd module also always resolves root and nobody, making
3376 it possible to have no /etc/passwd or /etc/group files in minimal
3377 container or chroot environments.
3379 * Services may be started with their own user namespace using the new
3380 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
3381 under which the service is running are mapped. All other users are
3384 * Support for the cgroup namespace has been added to systemd-nspawn. If
3385 supported by kernel, the container system started by systemd-nspawn
3386 will have its own view of the cgroup hierarchy. This new behaviour
3387 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
3389 * The new MemorySwapMax= option can be used to limit the maximum swap
3390 usage under the unified cgroup hierarchy.
3392 * Support for the CPU controller in the unified cgroup hierarchy has
3393 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
3394 options. This controller requires out-of-tree patches for the kernel
3395 and the support is provisional.
3397 * Mount and automount units may now be created transiently
3398 (i.e. dynamically at runtime via the bus API, instead of requiring
3399 unit files in the file system).
3401 * systemd-mount is a new tool which may mount file systems – much like
3402 mount(8), optionally pulling in additional dependencies through
3403 transient .mount and .automount units. For example, this tool
3404 automatically runs fsck on a backing block device before mounting,
3405 and allows the automount logic to be used dynamically from the
3406 command line for establishing mount points. This tool is particularly
3407 useful when dealing with removable media, as it will ensure fsck is
3408 run – if necessary – before the first access and that the file system
3409 is quickly unmounted after each access by utilizing the automount
3410 logic. This maximizes the chance that the file system on the
3411 removable media stays in a clean state, and if it isn't in a clean
3412 state is fixed automatically.
3414 * LazyUnmount=yes option for mount units has been added to expose the
3415 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
3418 * /efi will be used as the mount point of the EFI boot partition, if
3419 the directory is present, and the mount point was not configured
3420 through other means (e.g. fstab). If /efi directory does not exist,
3421 /boot will be used as before. This makes it easier to automatically
3422 mount the EFI partition on systems where /boot is used for something
3425 * When operating on GPT disk images for containers, systemd-nspawn will
3426 now mount the ESP to /boot or /efi according to the same rules as PID
3427 1 running on a host. This allows tools like "bootctl" to operate
3428 correctly within such containers, in order to make container images
3429 bootable on physical systems.
3431 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
3433 * Two new user session targets have been added to support running
3434 graphical sessions under the systemd --user instance:
3435 graphical-session.target and graphical-session-pre.target. See
3436 systemd.special(7) for a description of how those targets should be
3439 * The vconsole initialization code has been significantly reworked to
3440 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
3441 support unicode keymaps. Font and keymap configuration will now be
3442 copied to all allocated virtual consoles.
3444 * FreeBSD's bhyve virtualization is now detected.
3446 * Information recorded in the journal for core dumps now includes the
3447 contents of /proc/mountinfo and the command line of the process at
3448 the top of the process hierarchy (which is usually the init process
3451 * systemd-journal-gatewayd learned the --directory= option to serve
3452 files from the specified location.
3454 * journalctl --root=… can be used to peruse the journal in the
3455 /var/log/ directories inside of a container tree. This is similar to
3456 the existing --machine= option, but does not require the container to
3459 * The hardware database has been extended to support
3460 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
3463 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
3464 specify the click rate for mice which include a horizontal wheel with
3465 a click rate that is different than the one for the vertical wheel.
3467 * systemd-run gained a new --wait option that makes service execution
3468 synchronous. (Specifically, the command will not return until the
3469 specified service binary exited.)
3471 * systemctl gained a new --wait option that causes the start command to
3472 wait until the units being started have terminated again.
3474 * A new journal output mode "short-full" has been added which displays
3475 timestamps with abbreviated English day names and adds a timezone
3476 suffix. Those timestamps include more information than the default
3477 "short" output mode, and can be passed directly to journalctl's
3478 --since= and --until= options.
3480 * /etc/resolv.conf will be bind-mounted into containers started by
3481 systemd-nspawn, if possible, so any changes to resolv.conf contents
3482 are automatically propagated to the container.
3484 * The number of instances for socket-activated services originating
3485 from a single IP address can be limited with
3486 MaxConnectionsPerSource=, extending the existing setting of
3489 * systemd-networkd gained support for vcan ("Virtual CAN") interface
3492 * .netdev and .network configuration can now be extended through
3495 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
3496 Segmentation Offload, Generic Receive Offload, Large Receive Offload
3497 can be enabled and disabled using the new UDPSegmentationOffload=,
3498 TCPSegmentationOffload=, GenericSegmentationOffload=,
3499 GenericReceiveOffload=, LargeReceiveOffload= options in the
3500 [Link] section of .link files.
3502 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
3503 Port VLAN ID can be configured for bridge devices using the new STP=,
3504 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
3505 section of .netdev files.
3507 * The route table to which routes received over DHCP or RA should be
3508 added can be configured with the new RouteTable= option in the [DHCP]
3509 and [IPv6AcceptRA] sections of .network files.
3511 * The Address Resolution Protocol can be disabled on links managed by
3512 systemd-networkd using the ARP=no setting in the [Link] section of
3515 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
3516 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
3517 encode information about the result and exit codes of the current
3518 service runtime cycle.
3520 * systemd-sysctl will now configure kernel parameters in the order
3521 they occur in the configuration files. This matches what sysctl
3522 has been traditionally doing.
3524 * kernel-install "plugins" that are executed to perform various
3525 tasks after a new kernel is added and before an old one is removed
3526 can now return a special value to terminate the procedure and
3527 prevent any later plugins from running.
3529 * Journald's SplitMode=login setting has been deprecated. It has been
3530 removed from documentation, and its use is discouraged. In a future
3531 release it will be completely removed, and made equivalent to current
3532 default of SplitMode=uid.
3534 * Storage=both option setting in /etc/systemd/coredump.conf has been
3535 removed. With fast LZ4 compression storing the core dump twice is not
3538 * The --share-system systemd-nspawn option has been replaced with an
3539 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
3540 this functionality is discouraged. In addition the variables
3541 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
3542 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
3543 individual namespaces.
3545 * "machinectl list" now shows the IP address of running containers in
3546 the output, as well as OS release information.
3548 * "loginctl list" now shows the TTY of each session in the output.
3550 * sd-bus gained new API calls sd_bus_track_set_recursive(),
3551 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
3552 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
3553 tracking objects in a "recursive" mode, where a single client can be
3554 counted multiple times, if it takes multiple references.
3556 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
3557 sd_bus_get_exit_on_disconnect(). They may be used to make a
3558 process using sd-bus automatically exit if the bus connection is
3561 * Bus clients of the service manager may now "pin" loaded units into
3562 memory, by taking an explicit reference on them. This is useful to
3563 ensure the client can retrieve runtime data about the service even
3564 after the service completed execution. Taking such a reference is
3565 available only for privileged clients and should be helpful to watch
3566 running services in a race-free manner, and in particular collect
3567 information about exit statuses and results.
3569 * The nss-resolve module has been changed to strictly return UNAVAIL
3570 when communication via D-Bus with resolved failed, and NOTFOUND when
3571 a lookup completed but was negative. This means it is now possible to
3572 neatly configure fallbacks using nsswitch.conf result checking
3573 expressions. Taking benefit of this, the new recommended
3574 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
3576 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
3578 * A new setting CtrlAltDelBurstAction= has been added to
3579 /etc/systemd/system.conf which may be used to configure the precise
3580 behaviour if the user on the console presses Ctrl-Alt-Del more often
3581 than 7 times in 2s. Previously this would unconditionally result in
3582 an expedited, immediate reboot. With this new setting the precise
3583 operation may be configured in more detail, and also turned off
3586 * In .netdev files two new settings RemoteChecksumTx= and
3587 RemoteChecksumRx= are now understood that permit configuring the
3588 remote checksumming logic for VXLAN networks.
3590 * The service manager learnt a new "invocation ID" concept for invoked
3591 services. Each runtime cycle of a service will get a new invocation
3592 ID (a 128bit random UUID) assigned that identifies the current
3593 run of the service uniquely and globally. A new invocation ID
3594 is generated each time a service starts up. The journal will store
3595 the invocation ID of a service along with any logged messages, thus
3596 making the invocation ID useful for matching the online runtime of a
3597 service with the offline log data it generated in a safe way without
3598 relying on synchronized timestamps. In many ways this new service
3599 invocation ID concept is similar to the kernel's boot ID concept that
3600 uniquely and globally identifies the runtime of each boot. The
3601 invocation ID of a service is passed to the service itself via an
3602 environment variable ($INVOCATION_ID). A new bus call
3603 GetUnitByInvocationID() has been added that is similar to GetUnit()
3604 but instead of retrieving the bus path for a unit by its name
3605 retrieves it by its invocation ID. The returned path is valid only as
3606 long as the passed invocation ID is current.
3608 * systemd-resolved gained a new "DNSStubListener" setting in
3609 resolved.conf. It either takes a boolean value or the special values
3610 "udp" and "tcp", and configures whether to enable the stub DNS
3611 listener on 127.0.0.53:53.
3613 * IP addresses configured via networkd may now carry additional
3614 configuration settings supported by the kernel. New options include:
3615 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
3616 PrefixRoute=, AutoJoin=.
3618 * The PAM configuration fragment file for "user@.service" shipped with
3619 systemd (i.e. the --user instance of systemd) has been stripped to
3620 the minimum necessary to make the system boot. Previously, it
3621 contained Fedora-specific stanzas that did not apply to other
3622 distributions. It is expected that downstream distributions add
3623 additional configuration lines, matching their needs to this file,
3624 using it only as rough template of what systemd itself needs. Note
3625 that this reduced fragment does not even include an invocation of
3626 pam_limits which most distributions probably want to add, even though
3627 systemd itself does not need it. (There's also the new build time
3628 option --with-pamconfdir=no to disable installation of the PAM
3631 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
3632 capability is now also dropped from its set (in addition to
3633 CAP_SYS_MKNOD as before).
3635 * In service unit files it is now possible to connect a specific named
3636 file descriptor with stdin/stdout/stdout of an executed service. The
3637 name may be specified in matching .socket units using the
3638 FileDescriptorName= setting.
3640 * A number of journal settings may now be configured on the kernel
3641 command line. Specifically, the following options are now understood:
3642 systemd.journald.max_level_console=,
3643 systemd.journald.max_level_store=,
3644 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
3645 systemd.journald.max_level_wall=.
3647 * "systemctl is-enabled --full" will now show by which symlinks a unit
3648 file is enabled in the unit dependency tree.
3650 * Support for VeraCrypt encrypted partitions has been added to the
3651 "cryptsetup" logic and /etc/crypttab.
3653 * systemd-detect-virt gained support for a new --private-users switch
3654 that checks whether the invoking processes are running inside a user
3655 namespace. Similar, a new special value "private-users" for the
3656 existing ConditionVirtualization= setting has been added, permitting
3657 skipping of specific units in user namespace environments.
3659 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
3660 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
3661 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
3662 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
3663 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
3664 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
3665 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
3666 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
3667 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
3668 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
3669 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
3670 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
3671 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
3672 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
3673 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
3674 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
3675 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
3676 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
3677 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
3678 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
3679 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
3680 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
3681 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
3682 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
3683 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
3684 Jędrzejewski-Szmek, Zeal Jagannatha
3686 — Santa Fe, 2016-11-03
3690 * In service units the various ExecXYZ= settings have been extended
3691 with an additional special character as first argument of the
3692 assigned value: if the character '+' is used the specified command
3693 line it will be run with full privileges, regardless of User=,
3694 Group=, CapabilityBoundingSet= and similar options. The effect is
3695 similar to the existing PermissionsStartOnly= option, but allows
3696 configuration of this concept for each executed command line
3699 * Services may now alter the service watchdog timeout at runtime by
3700 sending a WATCHDOG_USEC= message via sd_notify().
3702 * MemoryLimit= and related unit settings now optionally take percentage
3703 specifications. The percentage is taken relative to the amount of
3704 physical memory in the system (or in case of containers, the assigned
3705 amount of memory). This allows scaling service resources neatly with
3706 the amount of RAM available on the system. Similarly, systemd-logind's
3707 RuntimeDirectorySize= option now also optionally takes percentage
3710 * In similar fashion TasksMax= takes percentage values now, too. The
3711 value is taken relative to the configured maximum number of processes
3712 on the system. The per-service task maximum has been changed to 15%
3713 using this functionality. (Effectively this is an increase of 512 →
3714 4915 for service units, given the kernel's default pid_max setting.)
3716 * Calendar time specifications in .timer units now understand a ".."
3717 syntax for time ranges. Example: "4..7:10" may now be used for
3718 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
3721 * The InaccessableDirectories=, ReadOnlyDirectories= and
3722 ReadWriteDirectories= unit file settings have been renamed to
3723 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
3724 applied to all kinds of file nodes, and not just directories, with
3725 the exception of symlinks. Specifically these settings may now be
3726 used on block and character device nodes, UNIX sockets and FIFOS as
3727 well as regular files. The old names of these settings remain
3728 available for compatibility.
3730 * systemd will now log about all service processes it kills forcibly
3731 (using SIGKILL) because they remained after the clean shutdown phase
3732 of the service completed. This should help identifying services that
3733 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
3734 systemd-logind's configuration a similar log message is generated for
3735 processes killed at the end of each session due to this setting.
3737 * systemd will now set the $JOURNAL_STREAM environment variable for all
3738 services whose stdout/stderr are connected to the Journal (which
3739 effectively means by default: all services). The variable contains
3740 the device and inode number of the file descriptor used for
3741 stdout/stderr. This may be used by invoked programs to detect whether
3742 their stdout/stderr is connected to the Journal, in which case they
3743 can switch over to direct Journal communication, thus being able to
3744 pass extended, structured metadata along with their log messages. As
3745 one example, this is now used by glib's logging primitives.
3747 * When using systemd's default tmp.mount unit for /tmp, the mount point
3748 will now be established with the "nosuid" and "nodev" options. This
3749 avoids privilege escalation attacks that put traps and exploits into
3750 /tmp. However, this might cause problems if you e. g. put container
3751 images or overlays into /tmp; if you need this, override tmp.mount's
3752 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
3755 * systemd now supports the "memory" cgroup controller also on
3758 * The systemd-cgtop tool now optionally takes a control group path as
3759 command line argument. If specified, the control group list shown is
3760 limited to subgroups of that group.
3762 * The SystemCallFilter= unit file setting gained support for
3763 pre-defined, named system call filter sets. For example
3764 SystemCallFilter=@clock is now an effective way to make all clock
3765 changing-related system calls unavailable to a service. A number of
3766 similar pre-defined groups are defined. Writing system call filters
3767 for system services is simplified substantially with this new
3768 concept. Accordingly, all of systemd's own, long-running services now
3769 enable system call filtering based on this, by default.
3771 * A new service setting MemoryDenyWriteExecute= has been added, taking
3772 a boolean value. If turned on, a service may no longer create memory
3773 mappings that are writable and executable at the same time. This
3774 enhances security for services where this is enabled as it becomes
3775 harder to dynamically write and then execute memory in exploited
3776 service processes. This option has been enabled for all of systemd's
3777 own long-running services.
3779 * A new RestrictRealtime= service setting has been added, taking a
3780 boolean argument. If set the service's processes may no longer
3781 acquire realtime scheduling. This improves security as realtime
3782 scheduling may otherwise be used to easily freeze the system.
3784 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
3785 value. This may be used for requesting that the system manager inside
3786 of the container reports start-up completion to nspawn which then
3787 propagates this notification further to the service manager
3788 supervising nspawn itself. A related option NotifyReady= in .nspawn
3789 files has been added too. This functionality allows ordering of the
3790 start-up of multiple containers using the usual systemd ordering
3793 * machinectl gained a new command "stop" that is an alias for
3796 * systemd-resolved gained support for contacting DNS servers on
3797 link-local IPv6 addresses.
3799 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
3800 its caches. A method call for requesting the same operation has been
3801 added to the bus API too, and is made available via "systemd-resolve
3804 * systemd-resolve gained a new --status switch. If passed a brief
3805 summary of the used DNS configuration with per-interface information
3808 * resolved.conf gained a new Cache= boolean option, defaulting to
3809 on. If turned off local DNS caching is disabled. This comes with a
3810 performance penalty in particular when DNSSEC is enabled. Note that
3811 resolved disables its internal caching implicitly anyway, when the
3812 configured DNS server is on a host-local IP address such as ::1 or
3813 127.0.0.1, thus automatically avoiding double local caching.
3815 * systemd-resolved now listens on the local IP address 127.0.0.53:53
3816 for DNS requests. This improves compatibility with local programs
3817 that do not use the libc NSS or systemd-resolved's bus APIs for name
3818 resolution. This minimal DNS service is only available to local
3819 programs and does not implement the full DNS protocol, but enough to
3820 cover local DNS clients. A new, static resolv.conf file, listing just
3821 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
3822 now recommended to make /etc/resolv.conf a symlink to this file in
3823 order to route all DNS lookups to systemd-resolved, regardless if
3824 done via NSS, the bus API or raw DNS packets. Note that this local
3825 DNS service is not as fully featured as the libc NSS or
3826 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
3827 used to deliver link-local address information (as this implies
3828 sending a local interface index along), LLMNR/mDNS support via this
3829 interface is severely restricted. It is thus strongly recommended for
3830 all applications to use the libc NSS API or native systemd-resolved
3833 * systemd-networkd's bridge support learned a new setting
3834 VLANFiltering= for controlling VLAN filtering. Moreover a new section
3835 in .network files has been added for configuring VLAN bridging in
3836 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
3838 * systemd-networkd's IPv6 Router Advertisement code now makes use of
3839 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
3840 now be acquired without relying on DHCPv6. Two new options
3841 UseDomains= and UseDNS= have been added to configure this behaviour.
3843 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
3844 renamed IPv6AcceptRA=, without altering its behaviour. The old
3845 setting name remains available for compatibility reasons.
3847 * The systemd-networkd VTI/VTI6 tunneling support gained new options
3848 Key=, InputKey= and OutputKey=.
3850 * systemd-networkd gained support for VRF ("Virtual Routing Function")
3851 interface configuration.
3853 * "systemctl edit" may now be used to create new unit files by
3854 specifying the --force switch.
3856 * sd-event gained a new function sd_event_get_iteration() for
3857 requesting the current iteration counter of the event loop. It starts
3858 at zero and is increased by one with each event loop iteration.
3860 * A new rpm macro %systemd_ordering is provided by the macros.systemd
3861 file. It can be used in lieu of %systemd_requires in packages which
3862 don't use any systemd functionality and are intended to be installed
3863 in minimal containers without systemd present. This macro provides
3864 ordering dependencies to ensure that if the package is installed in
3865 the same rpm transaction as systemd, systemd will be installed before
3866 the scriptlets for the package are executed, allowing unit presets
3869 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
3870 been added to simplify packaging of generators.
3872 * The os-release file gained VERSION_CODENAME field for the
3873 distribution nickname (e.g. VERSION_CODENAME=woody).
3875 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
3876 can be set to disable parsing of metadata and the creation
3877 of persistent symlinks for that device.
3879 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
3880 to make them available to logged-in users has been reverted.
3882 * Much of the common code of the various systemd components is now
3883 built into an internal shared library libsystemd-shared-231.so
3884 (incorporating the systemd version number in the name, to be updated
3885 with future releases) that the components link to. This should
3886 decrease systemd footprint both in memory during runtime and on
3887 disk. Note that the shared library is not for public use, and is
3888 neither API nor ABI stable, but is likely to change with every new
3889 released update. Packagers need to make sure that binaries
3890 linking to libsystemd-shared.so are updated in step with the
3893 * Configuration for "mkosi" is now part of the systemd
3894 repository. mkosi is a tool to easily build legacy-free OS images,
3895 and is available on github: https://github.com/systemd/mkosi. If
3896 "mkosi" is invoked in the build tree a new raw OS image is generated
3897 incorporating the systemd sources currently being worked on and a
3898 clean, fresh distribution installation. The generated OS image may be
3899 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
3900 UEFI PC. This functionality is particularly useful to easily test
3901 local changes made to systemd in a pristine, defined environment. See
3902 doc/HACKING for details.
3904 * configure learned the --with-support-url= option to specify the
3905 distribution's bugtracker.
3907 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
3908 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
3909 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
3910 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
3911 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
3912 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
3913 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
3914 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
3915 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
3916 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
3917 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
3918 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
3919 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
3920 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
3921 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
3922 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
3923 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
3924 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
3925 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
3927 — Berlin, 2016-07-25
3931 * DNSSEC is now turned on by default in systemd-resolved (in
3932 "allow-downgrade" mode), but may be turned off during compile time by
3933 passing "--with-default-dnssec=no" to "configure" (and of course,
3934 during runtime with DNSSEC= in resolved.conf). We recommend
3935 downstreams to leave this on at least during development cycles and
3936 report any issues with the DNSSEC logic upstream. We are very
3937 interested in collecting feedback about the DNSSEC validator and its
3938 limitations in the wild. Note however, that DNSSEC support is
3939 probably nothing downstreams should turn on in stable distros just
3940 yet, as it might create incompatibilities with a few DNS servers and
3941 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
3942 automatically whenever we detect such incompatible setups, but there
3943 might be systems we do not cover yet. Hence: please help us testing
3944 the DNSSEC code, leave this on where you can, report back, but then
3945 again don't consider turning this on in your stable, LTS or
3946 production release just yet. (Note that you have to enable
3947 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
3948 and its DNSSEC mode for host name resolution from local
3951 * systemd-resolve conveniently resolves DANE records with the --tlsa
3952 option and OPENPGPKEY records with the --openpgp option. It also
3953 supports dumping raw DNS record data via the new --raw= switch.
3955 * systemd-logind will now by default terminate user processes that are
3956 part of the user session scope unit (session-XX.scope) when the user
3957 logs out. This behavior is controlled by the KillUserProcesses=
3958 setting in logind.conf, and the previous default of "no" is now
3959 changed to "yes". This means that user sessions will be properly
3960 cleaned up after, but additional steps are necessary to allow
3961 intentionally long-running processes to survive logout.
3963 While the user is logged in at least once, user@.service is running,
3964 and any service that should survive the end of any individual login
3965 session can be started at a user service or scope using systemd-run.
3966 systemd-run(1) man page has been extended with an example which shows
3967 how to run screen in a scope unit underneath user@.service. The same
3968 command works for tmux.
3970 After the user logs out of all sessions, user@.service will be
3971 terminated too, by default, unless the user has "lingering" enabled.
3972 To effectively allow users to run long-term tasks even if they are
3973 logged out, lingering must be enabled for them. See loginctl(1) for
3974 details. The default polkit policy was modified to allow users to
3975 set lingering for themselves without authentication.
3977 Previous defaults can be restored at compile time by the
3978 --without-kill-user-processes option to "configure".
3980 * systemd-logind gained new configuration settings SessionsMax= and
3981 InhibitorsMax=, both with a default of 8192. It will not register new
3982 user sessions or inhibitors above this limit.
3984 * systemd-logind will now reload configuration on SIGHUP.
3986 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
3987 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
3988 enable. Also, support for the "io" cgroup controller in the unified
3989 hierarchy has been added, so that the "memory", "pids" and "io" are
3990 now the controllers that are supported on the unified hierarchy.
3992 WARNING: it is not possible to use previous systemd versions with
3993 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
3994 is necessary to also update systemd in the initramfs if using the
3995 unified hierarchy. An updated SELinux policy is also required.
3997 * LLDP support has been extended, and both passive (receive-only) and
3998 active (sender) modes are supported. Passive mode ("routers-only") is
3999 enabled by default in systemd-networkd. Active LLDP mode is enabled
4000 by default for containers on the internal network. The "networkctl
4001 lldp" command may be used to list information gathered. "networkctl
4002 status" will also show basic LLDP information on connected peers now.
4004 * The IAID and DUID unique identifier sent in DHCP requests may now be
4005 configured for the system and each .network file managed by
4006 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
4008 * systemd-networkd gained support for configuring proxy ARP support for
4009 each interface, via the ProxyArp= setting in .network files. It also
4010 gained support for configuring the multicast querier feature of
4011 bridge devices, via the new MulticastQuerier= setting in .netdev
4012 files. Similarly, snooping on the IGMP traffic can be controlled
4013 via the new setting MulticastSnooping=.
4015 A new setting PreferredLifetime= has been added for addresses
4016 configured in .network file to configure the lifetime intended for an
4019 The systemd-networkd DHCP server gained the option EmitRouter=, which
4020 defaults to yes, to configure whether the DHCP Option 3 (Router)
4023 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
4024 systemd-socket-activate and installed into /usr/bin. It is now fully
4027 * systemd-journald now uses separate threads to flush changes to disk
4028 when closing journal files, thus reducing impact of slow disk I/O on
4029 logging performance.
4031 * The sd-journal API gained two new calls
4032 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
4033 can be used to open journal files using file descriptors instead of
4034 file or directory paths. sd_journal_open_container() has been
4035 deprecated, sd_journal_open_directory_fd() should be used instead
4036 with the flag SD_JOURNAL_OS_ROOT.
4038 * journalctl learned a new output mode "-o short-unix" that outputs log
4039 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
4040 UTC). It also gained support for a new --no-hostname setting to
4041 suppress the hostname column in the family of "short" output modes.
4043 * systemd-ask-password now optionally skips printing of the password to
4044 stdout with --no-output which can be useful in scripts.
4046 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
4047 (devices tagged with ID_MAKER_TOOL) are now tagged with
4048 "uaccess" and are available to logged in users.
4050 * The DeviceAllow= unit setting now supports specifiers (with "%").
4052 * "systemctl show" gained a new --value switch, which allows print a
4053 only the contents of a specific unit property, without also printing
4054 the property's name. Similar support was added to "show*" verbs
4055 of loginctl and machinectl that output "key=value" lists.
4057 * A new unit type "generated" was added for files dynamically generated
4058 by generator tools. Similarly, a new unit type "transient" is used
4059 for unit files created using the runtime API. "systemctl enable" will
4060 refuse to operate on such files.
4062 * A new command "systemctl revert" has been added that may be used to
4063 revert to the vendor version of a unit file, in case local changes
4064 have been made by adding drop-ins or overriding the unit file.
4066 * "machinectl clean" gained a new verb to automatically remove all or
4067 just hidden container images.
4069 * systemd-tmpfiles gained support for a new line type "e" for emptying
4070 directories, if they exist, without creating them if they don't.
4072 * systemd-nspawn gained support for automatically patching the UID/GIDs
4073 of the owners and the ACLs of all files and directories in a
4074 container tree to match the UID/GID user namespacing range selected
4075 for the container invocation. This mode is enabled via the new
4076 --private-users-chown switch. It also gained support for
4077 automatically choosing a free, previously unused UID/GID range when
4078 starting a container, via the new --private-users=pick setting (which
4079 implies --private-users-chown). Together, these options for the first
4080 time make user namespacing for nspawn containers fully automatic and
4081 thus deployable. The systemd-nspawn@.service template unit file has
4082 been changed to use this functionality by default.
4084 * systemd-nspawn gained a new --network-zone= switch, that allows
4085 creating ad-hoc virtual Ethernet links between multiple containers,
4086 that only exist as long as at least one container referencing them is
4087 running. This allows easy connecting of multiple containers with a
4088 common link that implements an Ethernet broadcast domain. Each of
4089 these network "zones" may be named relatively freely by the user, and
4090 may be referenced by any number of containers, but each container may
4091 only reference one of these "zones". On the lower level, this is
4092 implemented by an automatically managed bridge network interface for
4093 each zone, that is created when the first container referencing its
4094 zone is created and removed when the last one referencing its zone
4097 * The default start timeout may now be configured on the kernel command
4098 line via systemd.default_timeout_start_sec=. It was already
4099 configurable via the DefaultTimeoutStartSec= option in
4100 /etc/systemd/system.conf.
4102 * Socket units gained a new TriggerLimitIntervalSec= and
4103 TriggerLimitBurst= setting to configure a limit on the activation
4104 rate of the socket unit.
4106 * The LimitNICE= setting now optionally takes normal UNIX nice values
4107 in addition to the raw integer limit value. If the specified
4108 parameter is prefixed with "+" or "-" and is in the range -20..19 the
4109 value is understood as UNIX nice value. If not prefixed like this it
4110 is understood as raw RLIMIT_NICE limit.
4112 * Note that the effect of the PrivateDevices= unit file setting changed
4113 slightly with this release: the per-device /dev file system will be
4114 mounted read-only from this version on, and will have "noexec"
4115 set. This (minor) change of behavior might cause some (exceptional)
4116 legacy software to break, when PrivateDevices=yes is set for its
4117 service. Please leave PrivateDevices= off if you run into problems
4120 * systemd-bootchart has been split out to a separate repository:
4121 https://github.com/systemd/systemd-bootchart
4123 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
4124 merged into the kernel in its current form.
4126 * The compatibility libraries libsystemd-daemon.so,
4127 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
4128 which have been deprecated since systemd-209 have been removed along
4129 with the corresponding pkg-config files. All symbols provided by
4130 those libraries are provided by libsystemd.so.
4132 * The Capabilities= unit file setting has been removed (it is ignored
4133 for backwards compatibility). AmbientCapabilities= and
4134 CapabilityBoundingSet= should be used instead.
4136 * A new special target has been added, initrd-root-device.target,
4137 which creates a synchronization point for dependencies of the root
4138 device in early userspace. Initramfs builders must ensure that this
4139 target is now included in early userspace.
4141 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
4142 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
4143 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
4144 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
4145 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
4146 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
4147 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
4148 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
4149 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
4150 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
4151 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
4152 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
4153 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
4154 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
4155 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
4156 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
4157 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
4158 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
4159 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
4160 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
4161 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
4162 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
4163 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
4164 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
4165 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
4168 — Fairfax, 2016-05-21
4172 * The systemd-resolved DNS resolver service has gained a substantial
4173 set of new features, most prominently it may now act as a DNSSEC
4174 validating stub resolver. DNSSEC mode is currently turned off by
4175 default, but is expected to be turned on by default in one of the
4176 next releases. For now, we invite everybody to test the DNSSEC logic
4177 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
4178 service also gained a full set of D-Bus interfaces, including calls
4179 to configure DNS and DNSSEC settings per link (for use by external
4180 network management software). systemd-resolved and systemd-networkd
4181 now distinguish between "search" and "routing" domains. The former
4182 are used to qualify single-label names, the latter are used purely
4183 for routing lookups within certain domains to specific links.
4184 resolved now also synthesizes RRs for all entries from /etc/hosts.
4186 * The systemd-resolve tool (which is a client utility for
4187 systemd-resolved) has been improved considerably and is now fully
4188 supported and documented. Hence it has moved from /usr/lib/systemd to
4191 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
4194 * The coredump collection logic has been reworked: when a coredump is
4195 collected it is now written to disk, compressed and processed
4196 (including stacktrace extraction) from a new instantiated service
4197 systemd-coredump@.service, instead of directly from the
4198 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
4199 processing large coredumps can take up a substantial amount of
4200 resources and time, and this previously happened entirely outside of
4201 systemd's service supervision. With the new logic the core_pattern
4202 hook only does minimal metadata collection before passing off control
4203 to the new instantiated service, which is configured with a time
4204 limit, a nice level and other settings to minimize negative impact on
4205 the rest of the system. Also note that the new logic will honour the
4206 RLIMIT_CORE setting of the crashed process, which now allows users
4207 and processes to turn off coredumping for their processes by setting
4210 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
4211 and all forked processes by default. Previously, PID 1 would leave
4212 the setting at "0" for all processes, as set by the kernel. Note that
4213 the resource limit traditionally has no effect on the generated
4214 coredumps on the system if the /proc/sys/kernel/core_pattern hook
4215 logic is used. Since the limit is now honoured (see above) its
4216 default has been changed so that the coredumping logic is enabled by
4217 default for all processes, while allowing specific opt-out.
4219 * When the stacktrace is extracted from processes of system users, this
4220 is now done as "systemd-coredump" user, in order to sandbox this
4221 potentially security sensitive parsing operation. (Note that when
4222 processing coredumps of normal users this is done under the user ID
4223 of process that crashed, as before.) Packagers should take notice
4224 that it is now necessary to create the "systemd-coredump" system user
4225 and group at package installation time.
4227 * The systemd-activate socket activation testing tool gained support
4228 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
4229 and --seqpacket switches. It also has been extended to support both
4230 new-style and inetd-style file descriptor passing. Use the new
4231 --inetd switch to request inetd-style file descriptor passing.
4233 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
4234 variable, which takes a boolean value. If set to false, ANSI color
4235 output is disabled in the tools even when run on a terminal that
4238 * The VXLAN support in networkd now supports two new settings
4239 DestinationPort= and PortRange=.
4241 * A new systemd.machine_id= kernel command line switch has been added,
4242 that may be used to set the machine ID in /etc/machine-id if it is
4243 not initialized yet. This command line option has no effect if the
4244 file is already initialized.
4246 * systemd-nspawn gained a new --as-pid2 switch that invokes any
4247 specified command line as PID 2 rather than PID 1 in the
4248 container. In this mode PID 1 is a minimal stub init process that
4249 implements the special POSIX and Linux semantics of PID 1 regarding
4250 signal and child process management. Note that this stub init process
4251 is implemented in nspawn itself and requires no support from the
4252 container image. This new logic is useful to support running
4253 arbitrary commands in the container, as normal processes are
4254 generally not prepared to run as PID 1.
4256 * systemd-nspawn gained a new --chdir= switch for setting the current
4257 working directory for the process started in the container.
4259 * "journalctl /dev/sda" will now output all kernel log messages for
4260 specified device from the current boot, in addition to all devices
4261 that are parents of it. This should make log output about devices
4262 pretty useful, as long as kernel drivers attach enough metadata to
4263 the log messages. (The usual SATA drivers do.)
4265 * The sd-journal API gained two new calls
4266 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
4267 that report whether log data from /run or /var has been found.
4269 * journalctl gained a new switch "--fields" that prints all journal
4270 record field names currently in use in the journal. This is backed
4271 by two new sd-journal API calls sd_journal_enumerate_fields() and
4272 sd_journal_restart_fields().
4274 * Most configurable timeouts in systemd now expect an argument of
4275 "infinity" to turn them off, instead of "0" as before. The semantics
4276 from now on is that a timeout of "0" means "now", and "infinity"
4277 means "never". To maintain backwards compatibility, "0" continues to
4278 turn off previously existing timeout settings.
4280 * "systemctl reload-or-try-restart" has been renamed to "systemctl
4281 try-reload-or-restart" to clarify what it actually does: the "try"
4282 logic applies to both reloading and restarting, not just restarting.
4283 The old name continues to be accepted for compatibility.
4285 * On boot-up, when PID 1 detects that the system clock is behind the
4286 release date of the systemd version in use, the clock is now set
4287 to the latter. Previously, this was already done in timesyncd, in order
4288 to avoid running with clocks set to the various clock epochs such as
4289 1902, 1938 or 1970. With this change the logic is now done in PID 1
4290 in addition to timesyncd during early boot-up, so that it is enforced
4291 before the first process is spawned by systemd. Note that the logic
4292 in timesyncd remains, as it is more comprehensive and ensures
4293 clock monotonicity by maintaining a persistent timestamp file in
4294 /var. Since /var is generally not available in earliest boot or the
4295 initrd, this part of the logic remains in timesyncd, and is not done
4298 * Support for tweaking details in net_cls.class_id through the
4299 NetClass= configuration directive has been removed, as the kernel
4300 people have decided to deprecate that controller in cgroup v2.
4301 Userspace tools such as nftables are moving over to setting rules
4302 that are specific to the full cgroup path of a task, which obsoletes
4303 these controllers anyway. The NetClass= directive is kept around for
4304 legacy compatibility reasons. For a more in-depth description of the
4305 kernel change, please refer to the respective upstream commit:
4307 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
4309 * A new service setting RuntimeMaxSec= has been added that may be used
4310 to specify a maximum runtime for a service. If the timeout is hit, the
4311 service is terminated and put into a failure state.
4313 * A new service setting AmbientCapabilities= has been added. It allows
4314 configuration of additional Linux process capabilities that are
4315 passed to the activated processes. This is only available on very
4318 * The process resource limit settings in service units may now be used
4319 to configure hard and soft limits individually.
4321 * The various libsystemd APIs such as sd-bus or sd-event now publicly
4322 expose support for gcc's __attribute__((cleanup())) C extension.
4323 Specifically, for many object destructor functions alternative
4324 versions have been added that have names suffixed with "p" and take a
4325 pointer to a pointer to the object to destroy, instead of just a
4326 pointer to the object itself. This is useful because these destructor
4327 functions may be used directly as parameters to the cleanup
4328 construct. Internally, systemd has been a heavy user of this GCC
4329 extension for a long time, and with this change similar support is
4330 now available to consumers of the library outside of systemd. Note
4331 that by using this extension in your sources compatibility with old
4332 and strictly ANSI compatible C compilers is lost. However, all gcc or
4333 LLVM versions of recent years support this extension.
4335 * Timer units gained support for a new setting RandomizedDelaySec= that
4336 allows configuring some additional randomized delay to the configured
4337 time. This is useful to spread out timer events to avoid load peaks in
4338 clusters or larger setups.
4340 * Calendar time specifications now support sub-second accuracy.
4342 * Socket units now support listening on SCTP and UDP-lite protocol
4345 * The sd-event API now comes with a full set of man pages.
4347 * Older versions of systemd contained experimental support for
4348 compressing journal files and coredumps with the LZ4 compressor that
4349 was not compatible with the lz4 binary (due to API limitations of the
4350 lz4 library). This support has been removed; only support for files
4351 compatible with the lz4 binary remains. This LZ4 logic is now
4352 officially supported and no longer considered experimental.
4354 * The dkr image import logic has been removed again from importd. dkr's
4355 micro-services focus doesn't fit into the machine image focus of
4356 importd, and quickly got out of date with the upstream dkr API.
4358 * Creation of the /run/lock/lockdev/ directory was dropped from
4359 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
4360 been available for many years. If you still need this, you need to
4361 create your own tmpfiles.d config file with:
4363 d /run/lock/lockdev 0775 root lock -
4365 * The settings StartLimitBurst=, StartLimitInterval=, StartLimitAction=
4366 and RebootArgument= have been moved from the [Service] section of
4367 unit files to [Unit], and they are now supported on all unit types,
4368 not just service units. Of course, systemd will continue to
4369 understand these settings also at the old location, in order to
4370 maintain compatibility.
4372 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
4373 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
4374 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
4375 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
4376 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
4377 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
4378 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
4379 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
4380 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
4381 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
4382 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
4383 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
4384 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
4385 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
4386 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
4387 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
4388 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
4389 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
4390 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
4392 — Berlin, 2016-02-11
4396 * A number of properties previously only settable in unit
4397 files are now also available as properties to set when
4398 creating transient units programmatically via the bus, as it
4399 is exposed with systemd-run's --property=
4400 setting. Specifically, these are: SyslogIdentifier=,
4401 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
4402 EnvironmentFile=, ReadWriteDirectories=,
4403 ReadOnlyDirectories=, InaccessibleDirectories=,
4404 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
4406 * When creating transient services via the bus API it is now
4407 possible to pass in a set of file descriptors to use as
4408 STDIN/STDOUT/STDERR for the invoked process.
4410 * Slice units may now be created transiently via the bus APIs,
4411 similar to the way service and scope units may already be
4412 created transiently.
4414 * Wherever systemd expects a calendar timestamp specification
4415 (like in journalctl's --since= and --until= switches) UTC
4416 timestamps are now supported. Timestamps suffixed with "UTC"
4417 are now considered to be in Universal Time Coordinated
4418 instead of the local timezone. Also, timestamps may now
4419 optionally be specified with sub-second accuracy. Both of
4420 these additions also apply to recurring calendar event
4421 specification, such as OnCalendar= in timer units.
4423 * journalctl gained a new "--sync" switch that asks the
4424 journal daemon to write all so far unwritten log messages to
4425 disk and sync the files, before returning.
4427 * systemd-tmpfiles learned two new line types "q" and "Q" that
4428 operate like "v", but also set up a basic btrfs quota
4429 hierarchy when used on a btrfs file system with quota
4432 * tmpfiles' "v", "q" and "Q" will now create a plain directory
4433 instead of a subvolume (even on a btrfs file system) if the
4434 root directory is a plain directory, and not a
4435 subvolume. This should simplify things with certain chroot()
4436 environments which are not aware of the concept of btrfs
4439 * systemd-detect-virt gained a new --chroot switch to detect
4440 whether execution takes place in a chroot() environment.
4442 * CPUAffinity= now takes CPU index ranges in addition to
4445 * The various memory-related resource limit settings (such as
4446 LimitAS=) now understand the usual K, M, G, ... suffixes to
4447 the base of 1024 (IEC). Similar, the time-related resource
4448 limit settings understand the usual min, h, day, ...
4451 * There's a new system.conf setting DefaultTasksMax= to
4452 control the default TasksMax= setting for services and
4453 scopes running on the system. (TasksMax= is the primary
4454 setting that exposes the "pids" cgroup controller on systemd
4455 and was introduced in the previous systemd release.) The
4456 setting now defaults to 512, which means services that are
4457 not explicitly configured otherwise will only be able to
4458 create 512 processes or threads at maximum, from this
4459 version on. Note that this means that thread- or
4460 process-heavy services might need to be reconfigured to set
4461 TasksMax= to a higher value. It is sufficient to set
4462 TasksMax= in these specific unit files to a higher value, or
4463 even "infinity". Similar, there's now a logind.conf setting
4464 UserTasksMax= that defaults to 4096 and limits the total
4465 number of processes or tasks each user may own
4466 concurrently. nspawn containers also have the TasksMax=
4467 value set by default now, to 8192. Note that all of this
4468 only has an effect if the "pids" cgroup controller is
4469 enabled in the kernel. The general benefit of these changes
4470 should be a more robust and safer system, that provides a
4471 certain amount of per-service fork() bomb protection.
4473 * systemd-nspawn gained the new --network-veth-extra= switch
4474 to define additional and arbitrarily-named virtual Ethernet
4475 links between the host and the container.
4477 * A new service execution setting PassEnvironment= has been
4478 added that allows importing select environment variables
4479 from PID1's environment block into the environment block of
4482 * Timer units gained support for a new RemainAfterElapse=
4483 setting which takes a boolean argument. It defaults to on,
4484 exposing behaviour unchanged to previous releases. If set to
4485 off, timer units are unloaded after they elapsed if they
4486 cannot elapse again. This is particularly useful for
4487 transient timer units, which shall not stay around longer
4488 than until they first elapse.
4490 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
4491 default now (the kernel default is 16). This is beneficial
4492 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
4493 allows substantially larger numbers of queued
4494 datagrams. This should increase the capability of systemd to
4495 parallelize boot-up, as logging and sd_notify() are unlikely
4496 to stall execution anymore. If you need to change the value
4497 from the new defaults, use the usual sysctl.d/ snippets.
4499 * The compression framing format used by the journal or
4500 coredump processing has changed to be in line with what the
4501 official LZ4 tools generate. LZ4 compression support in
4502 systemd was considered unsupported previously, as the format
4503 was not compatible with the normal tools. With this release
4504 this has changed now, and it is hence safe for downstream
4505 distributions to turn it on. While not compressing as well
4506 as the XZ, LZ4 is substantially faster, which makes
4507 it a good default choice for the compression logic in the
4508 journal and in coredump handling.
4510 * Any reference to /etc/mtab has been dropped from
4511 systemd. The file has been obsolete since a while, but
4512 systemd refused to work on systems where it was incorrectly
4513 set up (it should be a symlink or non-existent). Please make
4514 sure to update to util-linux 2.27.1 or newer in conjunction
4515 with this systemd release, which also drops any reference to
4516 /etc/mtab. If you maintain a distribution make sure that no
4517 software you package still references it, as this is a
4518 likely source of bugs. There's also a glibc bug pending,
4519 asking for removal of any reference to this obsolete file:
4521 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
4523 Note that only util-linux versions built with
4524 --enable-libmount-force-mountinfo are supported.
4526 * Support for the ".snapshot" unit type has been removed. This
4527 feature turned out to be little useful and little used, and
4528 has now been removed from the core and from systemctl.
4530 * The dependency types RequiresOverridable= and
4531 RequisiteOverridable= have been removed from systemd. They
4532 have been used only very sparingly to our knowledge and
4533 other options that provide a similar effect (such as
4534 systemctl --mode=ignore-dependencies) are much more useful
4535 and commonly used. Moreover, they were only half-way
4536 implemented as the option to control behaviour regarding
4537 these dependencies was never added to systemctl. By removing
4538 these dependency types the execution engine becomes a bit
4539 simpler. Unit files that use these dependencies should be
4540 changed to use the non-Overridable dependency types
4541 instead. In fact, when parsing unit files with these
4542 options, that's what systemd will automatically convert them
4543 too, but it will also warn, asking users to fix the unit
4544 files accordingly. Removal of these dependency types should
4545 only affect a negligible number of unit files in the wild.
4547 * Behaviour of networkd's IPForward= option changed
4548 (again). It will no longer maintain a per-interface setting,
4549 but propagate one way from interfaces where this is enabled
4550 to the global kernel setting. The global setting will be
4551 enabled when requested by a network that is set up, but
4552 never be disabled again. This change was made to make sure
4553 IPv4 and IPv6 behaviour regarding packet forwarding is
4554 similar (as the Linux IPv6 stack does not support
4555 per-interface control of this setting) and to minimize
4558 * In unit files the behaviour of %u, %U, %h, %s has
4559 changed. These specifiers will now unconditionally resolve
4560 to the various user database fields of the user that the
4561 systemd instance is running as, instead of the user
4562 configured in the specific unit via User=. Note that this
4563 effectively doesn't change much, as resolving of these
4564 specifiers was already turned off in the --system instance
4565 of systemd, as we cannot do NSS lookups from PID 1. In the
4566 --user instance of systemd these specifiers where correctly
4567 resolved, but hardly made any sense, since the user instance
4568 lacks privileges to do user switches anyway, and User= is
4569 hence useless. Moreover, even in the --user instance of
4570 systemd behaviour was awkward as it would only take settings
4571 from User= assignment placed before the specifier into
4572 account. In order to unify and simplify the logic around
4573 this the specifiers will now always resolve to the
4574 credentials of the user invoking the manager (which in case
4575 of PID 1 is the root user).
4577 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
4578 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
4579 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
4580 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
4581 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
4582 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
4583 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
4584 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
4585 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
4586 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
4587 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
4588 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
4589 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
4590 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
4593 — Berlin, 2015-11-18
4597 * systemd now depends on util-linux v2.27. More specifically,
4598 the newly added mount monitor feature in libmount now
4599 replaces systemd's former own implementation.
4601 * libmount mandates /etc/mtab not to be regular file, and
4602 systemd now enforces this condition at early boot.
4603 /etc/mtab has been deprecated and warned about for a very
4604 long time, so systems running systemd should already have
4605 stopped having this file around as anything else than a
4606 symlink to /proc/self/mounts.
4608 * Support for the "pids" cgroup controller has been added. It
4609 allows accounting the number of tasks in a cgroup and
4610 enforcing limits on it. This adds two new setting
4611 TasksAccounting= and TasksMax= to each unit, as well as a
4612 global option DefaultTasksAccounting=.
4614 * Support for the "net_cls" cgroup controller has been added.
4615 It allows assigning a net class ID to each task in the
4616 cgroup, which can then be used in firewall rules and traffic
4617 shaping configurations. Note that the kernel netfilter net
4618 class code does not currently work reliably for ingress
4619 packets on unestablished sockets.
4621 This adds a new config directive called NetClass= to CGroup
4622 enabled units. Allowed values are positive numbers for fixed
4623 assignments and "auto" for picking a free value
4626 * 'systemctl is-system-running' now returns 'offline' if the
4627 system is not booted with systemd. This command can now be
4628 used as a substitute for 'systemd-notify --booted'.
4630 * Watchdog timeouts have been increased to 3 minutes for all
4631 in-tree service files. Apparently, disk IO issues are more
4632 frequent than we hoped, and user reported >1 minute waiting
4635 * 'machine-id-commit' functionality has been merged into
4636 'machine-id-setup --commit'. The separate binary has been
4639 * The WorkingDirectory= directive in unit files may now be set
4640 to the special value '~'. In this case, the working
4641 directory is set to the home directory of the user
4642 configured in User=.
4644 * "machinectl shell" will now open the shell in the home
4645 directory of the selected user by default.
4647 * The CrashChVT= configuration file setting is renamed to
4648 CrashChangeVT=, following our usual logic of not
4649 abbreviating unnecessarily. The old directive is still
4650 supported for compat reasons. Also, this directive now takes
4651 an integer value between 1 and 63, or a boolean value. The
4652 formerly supported '-1' value for disabling stays around for
4655 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
4656 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
4657 RootDirectory= properties can now be set for transient
4660 * The systemd-analyze tool gained a new "set-log-target" verb
4661 to change the logging target the system manager logs to
4662 dynamically during runtime. This is similar to how
4663 "systemd-analyze set-log-level" already changes the log
4666 * In nspawn /sys is now mounted as tmpfs, with only a selected
4667 set of subdirectories mounted in from the real sysfs. This
4668 enhances security slightly, and is useful for ensuring user
4669 namespaces work correctly.
4671 * Support for USB FunctionFS activation has been added. This
4672 allows implementation of USB gadget services that are
4673 activated as soon as they are requested, so that they don't
4674 have to run continuously, similar to classic socket
4677 * The "systemctl exit" command now optionally takes an
4678 additional parameter that sets the exit code to return from
4679 the systemd manager when exiting. This is only relevant when
4680 running the systemd user instance, or when running the
4681 system instance in a container.
4683 * sd-bus gained the new API calls sd_bus_path_encode_many()
4684 and sd_bus_path_decode_many() that allow easy encoding and
4685 decoding of multiple identifier strings inside a D-Bus
4686 object path. Another new call sd_bus_default_flush_close()
4687 has been added to flush and close per-thread default
4690 * systemd-cgtop gained support for a -M/--machine= switch to
4691 show the control groups within a certain container only.
4693 * "systemctl kill" gained support for an optional --fail
4694 switch. If specified the requested operation will fail of no
4695 processes have been killed, because the unit had no
4696 processes attached, or similar.
4698 * A new systemd.crash_reboot=1 kernel command line option has
4699 been added that triggers a reboot after crashing. This can
4700 also be set through CrashReboot= in systemd.conf.
4702 * The RuntimeDirectory= setting now understands unit
4703 specifiers like %i or %f.
4705 * A new (still internal) library API sd-ipv4acd has been added,
4706 that implements address conflict detection for IPv4. It's
4707 based on code from sd-ipv4ll, and will be useful for
4708 detecting DHCP address conflicts.
4710 * File descriptors passed during socket activation may now be
4711 named. A new API sd_listen_fds_with_names() is added to
4712 access the names. The default names may be overridden,
4713 either in the .socket file using the FileDescriptorName=
4714 parameter, or by passing FDNAME= when storing the file
4715 descriptors using sd_notify().
4717 * systemd-networkd gained support for:
4719 - Setting the IPv6 Router Advertisement settings via
4720 IPv6AcceptRouterAdvertisements= in .network files.
4722 - Configuring the HelloTimeSec=, MaxAgeSec= and
4723 ForwardDelaySec= bridge parameters in .netdev files.
4725 - Configuring PreferredSource= for static routes in
4728 * The "ask-password" framework used to query for LUKS harddisk
4729 passwords or SSL passwords during boot gained support for
4730 caching passwords in the kernel keyring, if it is
4731 available. This makes sure that the user only has to type in
4732 a passphrase once if there are multiple objects to unlock
4733 with the same one. Previously, such password caching was
4734 available only when Plymouth was used; this moves the
4735 caching logic into the systemd codebase itself. The
4736 "systemd-ask-password" utility gained a new --keyname=
4737 switch to control which kernel keyring key to use for
4738 caching a password in. This functionality is also useful for
4739 enabling display managers such as gdm to automatically
4740 unlock the user's GNOME keyring if its passphrase, the
4741 user's password and the harddisk password are the same, if
4742 gdm-autologin is used.
4744 * When downloading tar or raw images using "machinectl
4745 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
4746 file is now also downloaded, if it is available and stored
4747 next to the image file.
4749 * Units of type ".socket" gained a new boolean setting
4750 Writable= which is only useful in conjunction with
4751 ListenSpecial=. If true, enables opening the specified
4752 special file in O_RDWR mode rather than O_RDONLY mode.
4754 * systemd-rfkill has been reworked to become a singleton
4755 service that is activated through /dev/rfkill on each rfkill
4756 state change and saves the settings to disk. This way,
4757 systemd-rfkill is now compatible with devices that exist
4758 only intermittendly, and even restores state if the previous
4759 system shutdown was abrupt rather than clean.
4761 * The journal daemon gained support for vacuuming old journal
4762 files controlled by the number of files that shall remain,
4763 in addition to the already existing control by size and by
4764 date. This is useful as journal interleaving performance
4765 degrades with too many separate journal files, and allows
4766 putting an effective limit on them. The new setting defaults
4767 to 100, but this may be changed by setting SystemMaxFiles=
4768 and RuntimeMaxFiles= in journald.conf. Also, the
4769 "journalctl" tool gained the new --vacuum-files= switch to
4770 manually vacuum journal files to leave only the specified
4771 number of files in place.
4773 * udev will now create /dev/disk/by-path links for ATA devices
4774 on kernels where that is supported.
4776 * Galician, Serbian, Turkish and Korean translations were added.
4778 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
4779 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
4780 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
4781 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
4782 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
4783 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
4784 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
4785 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
4786 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
4787 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
4788 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
4789 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
4790 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
4791 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
4792 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
4793 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
4794 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
4795 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
4797 — Berlin, 2015-10-07
4801 * The DHCP implementation of systemd-networkd gained a set of
4804 - The DHCP server now supports emitting DNS and NTP
4805 information. It may be enabled and configured via
4806 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
4807 and NTP information is enabled, but no servers are
4808 configured, the corresponding uplink information (if there
4809 is any) is propagated.
4811 - Server and client now support transmission and reception
4812 of timezone information. It can be configured via the
4813 newly introduced network options UseTimezone=,
4814 EmitTimezone=, and Timezone=. Transmission of timezone
4815 information is enabled between host and containers by
4816 default now: the container will change its local timezone
4817 to what the host has set.
4819 - Lease timeouts can now be configured via
4820 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
4822 - The DHCP server improved on the stability of
4823 leases. Clients are more likely to get the same lease
4824 information back, even if the server loses state.
4826 - The DHCP server supports two new configuration options to
4827 control the lease address pool metrics, PoolOffset= and
4830 * The encapsulation limit of tunnels in systemd-networkd may
4831 now be configured via 'EncapsulationLimit='. It allows
4832 modifying the maximum additional levels of encapsulation
4833 that are permitted to be prepended to a packet.
4835 * systemd now supports the concept of user buses replacing
4836 session buses, if used with dbus-1.10 (and enabled via dbus
4837 --enable-user-session). It previously only supported this on
4838 kdbus-enabled systems, and this release expands this to
4839 'dbus-daemon' systems.
4841 * systemd-networkd now supports predictable interface names
4844 * systemd now optionally supports the new Linux kernel
4845 "unified" control group hierarchy. If enabled via the kernel
4846 command-line option 'systemd.unified_cgroup_hierarchy=1',
4847 systemd will try to mount the unified cgroup hierarchy
4848 directly on /sys/fs/cgroup. If not enabled, or not
4849 available, systemd will fall back to the legacy cgroup
4850 hierarchy setup, as before. Host system and containers can
4851 mix and match legacy and unified hierarchies as they
4852 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
4853 environment variable to individually select the hierarchy to
4854 use for executed containers. By default, nspawn will use the
4855 unified hierarchy for the containers if the host uses the
4856 unified hierarchy, and the legacy hierarchy otherwise.
4857 Please note that at this point the unified hierarchy is an
4858 experimental kernel feature and is likely to change in one
4859 of the next kernel releases. Therefore, it should not be
4860 enabled by default in downstream distributions yet. The
4861 minimum required kernel version for the unified hierarchy to
4862 work is 4.2. Note that when the unified hierarchy is used
4863 for the first time delegated access to controllers is
4864 safe. Because of this systemd-nspawn containers will get
4865 access to controllers now, as will systemd user
4866 sessions. This means containers and user sessions may now
4867 manage their own resources, partitioning up what the system
4870 * A new special scope unit "init.scope" has been introduced
4871 that encapsulates PID 1 of the system. It may be used to
4872 determine resource usage and enforce resource limits on PID
4873 1 itself. PID 1 hence moved out of the root of the control
4876 * The cgtop tool gained support for filtering out kernel
4877 threads when counting tasks in a control group. Also, the
4878 count of processes is now recursively summed up by
4879 default. Two options -k and --recursive= have been added to
4880 revert to old behaviour. The tool has also been updated to
4881 work correctly in containers now.
4883 * systemd-nspawn's --bind= and --bind-ro= options have been
4884 extended to allow creation of non-recursive bind mounts.
4886 * libsystemd gained two new calls sd_pid_get_cgroup() and
4887 sd_peer_get_cgroup() which return the control group path of
4888 a process or peer of a connected AF_UNIX socket. This
4889 function call is particularly useful when implementing
4890 delegated subtrees support in the control group hierarchy.
4892 * The "sd-event" event loop API of libsystemd now supports
4893 correct dequeuing of real-time signals, without losing
4896 * When systemd requests a polkit decision when managing units it
4897 will now add additional fields to the request, including unit
4898 name and desired operation. This enables more powerful polkit
4899 policies, that make decisions depending on these parameters.
4901 * nspawn learnt support for .nspawn settings files, that may
4902 accompany the image files or directories of containers, and
4903 may contain additional settings for the container. This is
4904 an alternative to configuring container parameters via the
4905 nspawn command line.
4907 Contributions from: Cristian Rodríguez, Daniel Mack, David
4908 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
4909 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
4910 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
4911 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
4912 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
4913 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
4914 Andersen, Tom Gundersen, Torstein Husebø
4916 — Berlin, 2015-09-08
4920 * machinectl gained a new verb 'shell' which opens a fresh
4921 shell on the target container or the host. It is similar to
4922 the existing 'login' command of machinectl, but spawns the
4923 shell directly without prompting for username or
4924 password. The pseudo machine '.host' now refers to the local
4925 host and is used by default. Hence, 'machinectl shell' can
4926 be used as replacement for 'su -' which spawns a session as
4927 a fresh systemd unit in a way that is fully isolated from
4928 the originating session.
4930 * systemd-networkd learned to cope with private-zone DHCP
4931 options and allows other programs to query the values.
4933 * SELinux access control when enabling/disabling units is no
4934 longer enforced with this release. The previous implementation
4935 was incorrect, and a new corrected implementation is not yet
4936 available. As unit file operations are still protected via
4937 polkit and D-Bus policy this is not a security problem. Yet,
4938 distributions which care about optimal SELinux support should
4939 probably not stabilize on this release.
4941 * sd-bus gained support for matches of type "arg0has=", that
4942 test for membership of strings in string arrays sent in bus
4945 * systemd-resolved now dumps the contents of its DNS and LLMNR
4946 caches to the logs on reception of the SIGUSR1 signal. This
4947 is useful to debug DNS behaviour.
4949 * The coredumpctl tool gained a new --directory= option to
4950 operate on journal files in a specific directory.
4952 * "systemctl reboot" and related commands gained a new
4953 "--message=" option which may be used to set a free-text
4954 wall message when shutting down or rebooting the
4955 system. This message is also logged, which is useful for
4956 figuring out the reason for a reboot or shutdown a
4959 * The "systemd-resolve-host" tool's -i switch now takes
4960 network interface numbers as alternative to interface names.
4962 * A new unit file setting for services has been introduced:
4963 UtmpMode= allows configuration of how precisely systemd
4964 handles utmp and wtmp entries for the service if this is
4965 enabled. This allows writing services that appear similar to
4966 user sessions in the output of the "w", "who", "last" and
4969 * systemd-resolved will now locally synthesize DNS resource
4970 records for the "localhost" and "gateway" domains as well as
4971 the local hostname. This should ensure that clients querying
4972 RRs via resolved will get similar results as those going via
4973 NSS, if nss-myhostname is enabled.
4975 Contributions from: Alastair Hughes, Alex Crawford, Daniel
4976 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
4977 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
4978 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
4979 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
4980 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
4981 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
4982 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
4983 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
4984 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
4985 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
4986 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
4988 — Berlin, 2015-08-27
4992 * The systemd-efi-boot-generator functionality was merged into
4993 systemd-gpt-auto-generator.
4995 * systemd-networkd now supports Group Policy for vxlan
4996 devices. It can be enabled via the new boolean configuration
4997 option called 'GroupPolicyExtension='.
4999 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
5000 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
5001 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
5003 — Berlin, 2015-07-31
5007 * The python-systemd code has been removed from the systemd repository.
5008 A new repository has been created which accommodates the code from
5009 now on, and we kindly ask distributions to create a separate package
5010 for this: https://github.com/systemd/python-systemd
5012 * The systemd daemon will now reload its main configuration
5013 (/etc/systemd/system.conf) on daemon-reload.
5015 * sd-dhcp now exposes vendor specific extensions via
5016 sd_dhcp_lease_get_vendor_specific().
5018 * systemd-networkd gained a number of new configuration options.
5020 - A new boolean configuration option for TAP devices called
5021 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
5022 device, thus allowing to send and receive GSO packets.
5024 - A new tunnel configuration option called 'CopyDSCP='.
5025 If enabled, the DSCP field of ip6 tunnels is copied into the
5026 decapsulated packet.
5028 - A set of boolean bridge configuration options were added.
5029 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
5030 and 'UnicastFlood=' are now parsed by networkd and applied to the
5031 respective bridge link device via the respective IFLA_BRPORT_*
5034 - A new string configuration option to override the hostname sent
5035 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
5036 is true, networkd will use the configured hostname instead of the
5037 system hostname when sending DHCP requests.
5039 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
5040 networkd will configure the IPv6 flow-label of the tunnel device
5041 according to RFC2460.
5043 - The 'macvtap' virtual network devices are now supported, similar to
5044 the already supported 'macvlan' devices.
5046 * systemd-resolved now implements RFC5452 to improve resilience against
5047 cache poisoning. Additionally, source port randomization is enabled
5048 by default to further protect against DNS spoofing attacks.
5050 * nss-mymachines now supports translating UIDs and GIDs of running
5051 containers with user-namespaces enabled. If a container 'foo'
5052 translates a host uid 'UID' to the container uid 'TUID', then
5053 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
5054 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
5055 mapped as 'vg-foo-TGID'.
5057 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
5058 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
5059 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
5060 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
5061 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
5062 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
5063 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
5064 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
5065 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
5066 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
5068 — Berlin, 2015-07-29
5072 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
5073 There are no known issues with current sysfs, and udev does not need
5074 or should be used to work around such bugs.
5076 * udev does no longer enable USB HID power management. Several reports
5077 indicate, that some devices cannot handle that setting.
5079 * The udev accelerometer helper was removed. The functionality
5080 is now fully included in iio-sensor-proxy. But this means,
5081 older iio-sensor-proxy versions will no longer provide
5082 accelerometer/orientation data with this systemd version.
5083 Please upgrade iio-sensor-proxy to version 1.0.
5085 * networkd gained a new configuration option IPv6PrivacyExtensions=
5086 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
5087 for Stateless Address") on selected networks.
5089 * For the sake of fewer build-time dependencies and less code in the
5090 main repository, the python bindings are about to be removed in the
5091 next release. A new repository has been created which accommodates
5092 the code from now on, and we kindly ask distributions to create a
5093 separate package for this. The removal will take place in v223.
5095 https://github.com/systemd/python-systemd
5097 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
5098 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
5099 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
5100 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
5101 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
5102 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
5103 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
5104 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
5105 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
5106 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
5108 — Berlin, 2015-07-07
5112 * The sd-bus.h and sd-event.h APIs have now been declared
5113 stable and have been added to the official interface of
5114 libsystemd.so. sd-bus implements an alternative D-Bus client
5115 library, that is relatively easy to use, very efficient and
5116 supports both classic D-Bus as well as kdbus as transport
5117 backend. sd-event is a generic event loop abstraction that
5118 is built around Linux epoll, but adds features such as event
5119 prioritization or efficient timer handling. Both APIs are good
5120 choices for C programs looking for a bus and/or event loop
5121 implementation that is minimal and does not have to be
5122 portable to other kernels.
5124 * kdbus support is no longer compile-time optional. It is now
5125 always built-in. However, it can still be disabled at
5126 runtime using the kdbus=0 kernel command line setting, and
5127 that setting may be changed to default to off, by specifying
5128 --disable-kdbus at build-time. Note though that the kernel
5129 command line setting has no effect if the kdbus.ko kernel
5130 module is not installed, in which case kdbus is (obviously)
5131 also disabled. We encourage all downstream distributions to
5132 begin testing kdbus by adding it to the kernel images in the
5133 development distributions, and leaving kdbus support in
5136 * The minimal required util-linux version has been bumped to
5139 * Support for chkconfig (--enable-chkconfig) was removed in
5140 favor of calling an abstraction tool
5141 /lib/systemd/systemd-sysv-install. This needs to be
5142 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
5143 in README for details.
5145 * If there's a systemd unit and a SysV init script for the
5146 same service name, and the user executes "systemctl enable"
5147 for it (or a related call), then this will now enable both
5148 (or execute the related operation on both), not just the
5151 * The libudev API documentation has been converted from gtkdoc
5154 * gudev has been removed from the systemd tree, it is now an
5157 * The systemd-cgtop tool learnt a new --raw switch to generate
5158 "raw" (machine parsable) output.
5160 * networkd's IPForwarding= .network file setting learnt the
5161 new setting "kernel", which ensures that networkd does not
5162 change the IP forwarding sysctl from the default kernel
5165 * The systemd-logind bus API now exposes a new boolean
5166 property "Docked" that reports whether logind considers the
5167 system "docked", i.e. connected to a docking station or not.
5169 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
5170 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
5171 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
5172 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
5173 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
5174 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
5175 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
5176 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
5177 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
5178 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
5179 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
5180 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
5181 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
5182 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
5183 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
5184 Fink, Zbigniew Jędrzejewski-Szmek
5186 — Berlin, 2015-06-19
5190 * The gudev library has been extracted into a separate repository
5191 available at: https://git.gnome.org/browse/libgudev/
5192 It is now managed as part of the Gnome project. Distributions
5193 are recommended to pass --disable-gudev to systemd and use
5194 gudev from the Gnome project instead. gudev is still included
5195 in systemd, for now. It will be removed soon, though. Please
5196 also see the announcement-thread on systemd-devel:
5197 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
5199 * systemd now exposes a CPUUsageNSec= property for each
5200 service unit on the bus, that contains the overall consumed
5201 CPU time of a service (the sum of what each process of the
5202 service consumed). This value is only available if
5203 CPUAccounting= is turned on for a service, and is then shown
5204 in the "systemctl status" output.
5206 * Support for configuring alternative mappings of the old SysV
5207 runlevels to systemd targets has been removed. They are now
5208 hardcoded in a way that runlevels 2, 3, 4 all map to
5209 multi-user.target and 5 to graphical.target (which
5210 previously was already the default behaviour).
5212 * The auto-mounter logic gained support for mount point
5213 expiry, using a new TimeoutIdleSec= setting in .automount
5214 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
5216 * The EFI System Partition (ESP) as mounted to /boot by
5217 systemd-efi-boot-generator will now be unmounted
5218 automatically after 2 minutes of not being used. This should
5219 minimize the risk of ESP corruptions.
5221 * New /etc/fstab options x-systemd.requires= and
5222 x-systemd.requires-mounts-for= are now supported to express
5223 additional dependencies for mounts. This is useful for
5224 journalling file systems that support external journal
5225 devices or overlay file systems that require underlying file
5226 systems to be mounted.
5228 * systemd does not support direct live-upgrades (via systemctl
5229 daemon-reexec) from versions older than v44 anymore. As no
5230 distribution we are aware of shipped such old versions in a
5231 stable release this should not be problematic.
5233 * When systemd forks off a new per-connection service instance
5234 it will now set the $REMOTE_ADDR environment variable to the
5235 remote IP address, and $REMOTE_PORT environment variable to
5236 the remote IP port. This behaviour is similar to the
5237 corresponding environment variables defined by CGI.
5239 * systemd-networkd gained support for uplink failure
5240 detection. The BindCarrier= option allows binding interface
5241 configuration dynamically to the link sense of other
5242 interfaces. This is useful to achieve behaviour like in
5245 * systemd-networkd gained support for configuring the DHCP
5246 client identifier to use when requesting leases.
5248 * systemd-networkd now has a per-network UseNTP= option to
5249 configure whether NTP server information acquired via DHCP
5250 is passed on to services like systemd-timesyncd.
5252 * systemd-networkd gained support for vti6 tunnels.
5254 * Note that systemd-networkd manages the sysctl variable
5255 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
5256 it is configured for since v219. The variable controls IP
5257 forwarding, and is a per-interface alternative to the global
5258 /proc/sys/net/ipv[46]/ip_forward. This setting is
5259 configurable in the IPForward= option, which defaults to
5260 "no". This means if networkd is used for an interface it is
5261 no longer sufficient to set the global sysctl option to turn
5262 on IP forwarding! Instead, the .network file option
5263 IPForward= needs to be turned on! Note that the
5264 implementation of this behaviour was broken in v219 and has
5267 * Many bonding and vxlan options are now configurable in
5270 * systemd-nspawn gained a new --property= setting to set unit
5271 properties for the container scope. This is useful for
5272 setting resource parameters (e.g. "CPUShares=500") on
5273 containers started from the command line.
5275 * systemd-nspawn gained a new --private-users= switch to make
5276 use of user namespacing available on recent Linux kernels.
5278 * systemd-nspawn may now be called as part of a shell pipeline
5279 in which case the pipes used for stdin and stdout are passed
5280 directly to the process invoked in the container, without
5281 indirection via a pseudo tty.
5283 * systemd-nspawn gained a new switch to control the UNIX
5284 signal to use when killing the init process of the container
5287 * systemd-nspawn gained a new --overlay= switch for mounting
5288 overlay file systems into the container using the new kernel
5291 * When a container image is imported via systemd-importd and
5292 the host file system is not btrfs, a loopback block device
5293 file is created in /var/lib/machines.raw with a btrfs file
5294 system inside. It is then mounted to /var/lib/machines to
5295 enable btrfs features for container management. The loopback
5296 file and btrfs file system is grown as needed when container
5297 images are imported via systemd-importd.
5299 * systemd-machined/systemd-importd gained support for btrfs
5300 quota, to enforce container disk space limits on disk. This
5301 is exposed in "machinectl set-limit".
5303 * systemd-importd now can import containers from local .tar,
5304 .raw and .qcow2 images, and export them to .tar and .raw. It
5305 can also import dkr v2 images now from the network (on top
5308 * systemd-importd gained support for verifying downloaded
5309 images with gpg2 (previously only gpg1 was supported).
5311 * systemd-machined, systemd-logind, systemd: most bus calls are
5312 now accessible to unprivileged processes via polkit. Also,
5313 systemd-logind will now allow users to kill their own sessions
5314 without further privileges or authorization.
5316 * systemd-shutdownd has been removed. This service was
5317 previously responsible for implementing scheduled shutdowns
5318 as exposed in /usr/bin/shutdown's time parameter. This
5319 functionality has now been moved into systemd-logind and is
5320 accessible via a bus interface.
5322 * "systemctl reboot" gained a new switch --firmware-setup that
5323 can be used to reboot into the EFI firmware setup, if that
5324 is available. systemd-logind now exposes an API on the bus
5325 to trigger such reboots, in case graphical desktop UIs want
5326 to cover this functionality.
5328 * "systemctl enable", "systemctl disable" and "systemctl mask"
5329 now support a new "--now" switch. If specified the units
5330 that are enabled will also be started, and the ones
5331 disabled/masked also stopped.
5333 * The Gummiboot EFI boot loader tool has been merged into
5334 systemd, and renamed to "systemd-boot". The bootctl tool has been
5335 updated to support systemd-boot.
5337 * An EFI kernel stub has been added that may be used to create
5338 kernel EFI binaries that contain not only the actual kernel,
5339 but also an initrd, boot splash, command line and OS release
5340 information. This combined binary can then be signed as a
5341 single image, so that the firmware can verify it all in one
5342 step. systemd-boot has special support for EFI binaries created
5343 like this and can extract OS release information from them
5344 and show them in the boot menu. This functionality is useful
5345 to implement cryptographically verified boot schemes.
5347 * Optional support has been added to systemd-fsck to pass
5348 fsck's progress report to an AF_UNIX socket in the file
5351 * udev will no longer create device symlinks for all block
5352 devices by default. A blacklist for excluding special block
5353 devices from this logic has been turned into a whitelist
5354 that requires picking block devices explicitly that require
5357 * A new (currently still internal) API sd-device.h has been
5358 added to libsystemd. This modernized API is supposed to
5359 replace libudev eventually. In fact, already much of libudev
5360 is now just a wrapper around sd-device.h.
5362 * A new hwdb database for storing metadata about pointing
5363 stick devices has been added.
5365 * systemd-tmpfiles gained support for setting file attributes
5366 similar to the "chattr" tool with new 'h' and 'H' lines.
5368 * systemd-journald will no longer unconditionally set the
5369 btrfs NOCOW flag on new journal files. This is instead done
5370 with tmpfiles snippet using the new 'h' line type. This
5371 allows easy disabling of this logic, by masking the
5372 journal-nocow.conf tmpfiles file.
5374 * systemd-journald will now translate audit message types to
5375 human readable identifiers when writing them to the
5376 journal. This should improve readability of audit messages.
5378 * The LUKS logic gained support for the offset= and skip=
5379 options in /etc/crypttab, as previously implemented by
5382 * /usr/lib/os-release gained a new optional field VARIANT= for
5383 distributions that support multiple variants (such as a
5384 desktop edition, a server edition, ...)
5386 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
5387 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
5388 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
5389 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
5390 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
5391 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
5392 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
5393 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
5394 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
5395 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
5396 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
5397 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
5398 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
5399 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
5400 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
5401 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
5402 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
5403 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
5404 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
5405 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
5406 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
5407 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
5408 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
5409 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
5410 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
5411 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
5412 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
5414 — Berlin, 2015-05-22
5418 * Introduce a new API "sd-hwdb.h" for querying the hardware
5419 metadata database. With this minimal interface one can query
5420 and enumerate the udev hwdb, decoupled from the old libudev
5421 library. libudev's interface for this is now only a wrapper
5422 around sd-hwdb. A new tool systemd-hwdb has been added to
5423 interface with and update the database.
5425 * When any of systemd's tools copies files (for example due to
5426 tmpfiles' C lines) a btrfs reflink will attempted first,
5427 before bytewise copying is done.
5429 * systemd-nspawn gained a new --ephemeral switch. When
5430 specified a btrfs snapshot is taken of the container's root
5431 directory, and immediately removed when the container
5432 terminates again. Thus, a container can be started whose
5433 changes never alter the container's root directory, and are
5434 lost on container termination. This switch can also be used
5435 for starting a container off the root file system of the
5436 host without affecting the host OS. This switch is only
5437 available on btrfs file systems.
5439 * systemd-nspawn gained a new --template= switch. It takes the
5440 path to a container tree to use as template for the tree
5441 specified via --directory=, should that directory be
5442 missing. This allows instantiating containers dynamically,
5443 on first run. This switch is only available on btrfs file
5446 * When a .mount unit refers to a mount point on which multiple
5447 mounts are stacked, and the .mount unit is stopped all of
5448 the stacked mount points will now be unmounted until no
5449 mount point remains.
5451 * systemd now has an explicit notion of supported and
5452 unsupported unit types. Jobs enqueued for unsupported unit
5453 types will now fail with an "unsupported" error code. More
5454 specifically .swap, .automount and .device units are not
5455 supported in containers, .busname units are not supported on
5456 non-kdbus systems. .swap and .automount are also not
5457 supported if their respective kernel compile time options
5460 * machinectl gained support for two new "copy-from" and
5461 "copy-to" commands for copying files from a running
5462 container to the host or vice versa.
5464 * machinectl gained support for a new "bind" command to bind
5465 mount host directories into local containers. This is
5466 currently only supported for nspawn containers.
5468 * networkd gained support for configuring bridge forwarding
5469 database entries (fdb) from .network files.
5471 * A new tiny daemon "systemd-importd" has been added that can
5472 download container images in tar, raw, qcow2 or dkr formats,
5473 and make them available locally in /var/lib/machines, so
5474 that they can run as nspawn containers. The daemon can GPG
5475 verify the downloads (not supported for dkr, since it has no
5476 provisions for verifying downloads). It will transparently
5477 decompress bz2, xz, gzip compressed downloads if necessary,
5478 and restore sparse files on disk. The daemon uses privilege
5479 separation to ensure the actual download logic runs with
5480 fewer privileges than the daemon itself. machinectl has
5481 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
5482 make the functionality of importd available to the
5483 user. With this in place the Fedora and Ubuntu "Cloud"
5484 images can be downloaded and booted as containers unmodified
5485 (the Fedora images lack the appropriate GPG signature files
5486 currently, so they cannot be verified, but this will change
5487 soon, hopefully). Note that downloading images is currently
5488 only fully supported on btrfs.
5490 * machinectl is now able to list container images found in
5491 /var/lib/machines, along with some metadata about sizes of
5492 disk and similar. If the directory is located on btrfs and
5493 quota is enabled, this includes quota display. A new command
5494 "image-status" has been added that shows additional
5495 information about images.
5497 * machinectl is now able to clone container images
5498 efficiently, if the underlying file system (btrfs) supports
5499 it, with the new "machinectl clone" command. It also
5500 gained commands for renaming and removing images, as well as
5501 marking them read-only or read-write (supported also on
5502 legacy file systems).
5504 * networkd gained support for collecting LLDP network
5505 announcements, from hardware that supports this. This is
5506 shown in networkctl output.
5508 * systemd-run gained support for a new -t (--pty) switch for
5509 invoking a binary on a pty whose input and output is
5510 connected to the invoking terminal. This allows executing
5511 processes as system services while interactively
5512 communicating with them via the terminal. Most interestingly
5513 this is supported across container boundaries. Invoking
5514 "systemd-run -t /bin/bash" is an alternative to running a
5515 full login session, the difference being that the former
5516 will not register a session, nor go through the PAM session
5519 * tmpfiles gained support for a new "v" line type for creating
5520 btrfs subvolumes. If the underlying file system is a legacy
5521 file system, this automatically degrades to creating a
5522 normal directory. Among others /var/lib/machines is now
5523 created like this at boot, should it be missing.
5525 * The directory /var/lib/containers/ has been deprecated and
5526 been replaced by /var/lib/machines. The term "machines" has
5527 been used in the systemd context as generic term for both
5528 VMs and containers, and hence appears more appropriate for
5529 this, as the directory can also contain raw images bootable
5532 * systemd-nspawn when invoked with -M but without --directory=
5533 or --image= is now capable of searching for the container
5534 root directory, subvolume or disk image automatically, in
5535 /var/lib/machines. systemd-nspawn@.service has been updated
5536 to make use of this, thus allowing it to be used for raw
5539 * A new machines.target unit has been introduced that is
5540 supposed to group all containers/VMs invoked as services on
5541 the system. systemd-nspawn@.service has been updated to
5542 integrate with that.
5544 * machinectl gained a new "start" command, for invoking a
5545 container as a service. "machinectl start foo" is mostly
5546 equivalent to "systemctl start systemd-nspawn@foo.service",
5547 but handles escaping in a nicer way.
5549 * systemd-nspawn will now mount most of the cgroupfs tree
5550 read-only into each container, with the exception of the
5551 container's own subtree in the name=systemd hierarchy.
5553 * journald now sets the special FS_NOCOW file flag for its
5554 journal files. This should improve performance on btrfs, by
5555 avoiding heavy fragmentation when journald's write-pattern
5556 is used on COW file systems. It degrades btrfs' data
5557 integrity guarantees for the files to the same levels as for
5558 ext3/ext4 however. This should be OK though as journald does
5559 its own data integrity checks and all its objects are
5560 checksummed on disk. Also, journald should handle btrfs disk
5561 full events a lot more gracefully now, by processing SIGBUS
5562 errors, and not relying on fallocate() anymore.
5564 * When journald detects that journal files it is writing to
5565 have been deleted it will immediately start new journal
5568 * systemd now provides a way to store file descriptors
5569 per-service in PID 1. This is useful for daemons to ensure
5570 that fds they require are not lost during a daemon
5571 restart. The fds are passed to the daemon on the next
5572 invocation in the same way socket activation fds are
5573 passed. This is now used by journald to ensure that the
5574 various sockets connected to all the system's stdout/stderr
5575 are not lost when journald is restarted. File descriptors
5576 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
5577 an extension to sd_notify(). Note that a limit is enforced
5578 on the number of fds a service can store in PID 1, and it
5579 defaults to 0, so that no fds may be stored, unless this is
5580 explicitly turned on.
5582 * The default TERM variable to use for units connected to a
5583 terminal, when no other value is explicitly is set is now
5584 vt220 rather than vt102. This should be fairly safe still,
5585 but allows PgUp/PgDn work.
5587 * The /etc/crypttab option header= as known from Debian is now
5590 * "loginctl user-status" and "loginctl session-status" will
5591 now show the last 10 lines of log messages of the
5592 user/session following the status output. Similar,
5593 "machinectl status" will show the last 10 log lines
5594 associated with a virtual machine or container
5595 service. (Note that this is usually not the log messages
5596 done in the VM/container itself, but simply what the
5597 container manager logs. For nspawn this includes all console
5600 * "loginctl session-status" without further argument will now
5601 show the status of the session of the caller. Similar,
5602 "lock-session", "unlock-session", "activate",
5603 "enable-linger", "disable-linger" may now be called without
5604 session/user parameter in which case they apply to the
5605 caller's session/user.
5607 * An X11 session scriptlet is now shipped that uploads
5608 $DISPLAY and $XAUTHORITY into the environment of the systemd
5609 --user daemon if a session begins. This should improve
5610 compatibility with X11 enabled applications run as systemd
5613 * Generators are now subject to masking via /etc and /run, the
5614 same way as unit files.
5616 * networkd .network files gained support for configuring
5617 per-link IPv4/IPv6 packet forwarding as well as IPv4
5618 masquerading. This is by default turned on for veth links to
5619 containers, as registered by systemd-nspawn. This means that
5620 nspawn containers run with --network-veth will now get
5621 automatic routed access to the host's networks without any
5622 further configuration or setup, as long as networkd runs on
5625 * systemd-nspawn gained the --port= (-p) switch to expose TCP
5626 or UDP posts of a container on the host. With this in place
5627 it is possible to run containers with private veth links
5628 (--network-veth), and have their functionality exposed on
5629 the host as if their services were running directly on the
5632 * systemd-nspawn's --network-veth switch now gained a short
5633 version "-n", since with the changes above it is now truly
5634 useful out-of-the-box. The systemd-nspawn@.service has been
5635 updated to make use of it too by default.
5637 * systemd-nspawn will now maintain a per-image R/W lock, to
5638 ensure that the same image is not started more than once
5639 writable. (It's OK to run an image multiple times
5640 simultaneously in read-only mode.)
5642 * systemd-nspawn's --image= option is now capable of
5643 dissecting and booting MBR and GPT disk images that contain
5644 only a single active Linux partition. Previously it
5645 supported only GPT disk images with proper GPT type
5646 IDs. This allows running cloud images from major
5647 distributions directly with systemd-nspawn, without
5650 * In addition to collecting mouse dpi data in the udev
5651 hardware database, there's now support for collecting angle
5652 information for mouse scroll wheels. The database is
5653 supposed to guarantee similar scrolling behavior on mice
5654 that it knows about. There's also support for collecting
5655 information about Touchpad types.
5657 * udev's input_id built-in will now also collect touch screen
5658 dimension data and attach it to probed devices.
5660 * /etc/os-release gained support for a Distribution Privacy
5663 * networkd gained support for creating "ipvlan", "gretap",
5664 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
5666 * systemd-tmpfiles gained support for "a" lines for setting
5669 * systemd-nspawn will now mount /tmp in the container to
5670 tmpfs, automatically.
5672 * systemd now exposes the memory.usage_in_bytes cgroup
5673 attribute and shows it for each service in the "systemctl
5674 status" output, if available.
5676 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
5677 immediate reboot is triggered. This useful if shutdown is
5678 hung and is unable to complete, to expedite the
5679 operation. Note that this kind of reboot will still unmount
5680 all file systems, and hence should not result in fsck being
5683 * A .device unit for an optical block device will now be
5684 considered active only when a medium is in the drive. Also,
5685 mount units are now bound to their backing devices thus
5686 triggering automatic unmounting when devices become
5687 unavailable. With this in place systemd will now
5688 automatically unmount left-over mounts when a CD-ROM is
5689 ejected or an USB stick is yanked from the system.
5691 * networkd-wait-online now has support for waiting for
5692 specific interfaces only (with globbing), and for giving up
5693 after a configurable timeout.
5695 * networkd now exits when idle. It will be automatically
5696 restarted as soon as interfaces show up, are removed or
5697 change state. networkd will stay around as long as there is
5698 at least one DHCP state machine or similar around, that keep
5701 * networkd may now configure IPv6 link-local addressing in
5702 addition to IPv4 link-local addressing.
5704 * The IPv6 "token" for use in SLAAC may now be configured for
5705 each .network interface in networkd.
5707 * Routes configured with networkd may now be assigned a scope
5710 * networkd's [Match] sections now support globbing and lists
5711 of multiple space-separated matches per item.
5713 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
5714 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
5715 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
5716 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
5717 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
5718 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
5719 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
5720 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
5721 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
5722 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
5723 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
5724 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
5725 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
5726 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
5727 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
5728 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
5729 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
5730 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
5731 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
5732 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
5733 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
5734 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
5735 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
5736 Hoffmann, Zbigniew Jędrzejewski-Szmek
5738 — Berlin, 2015-02-16
5742 * When querying unit file enablement status (for example via
5743 "systemctl is-enabled"), a new state "indirect" is now known
5744 which indicates that a unit might not be enabled itself, but
5745 another unit listed in its Also= setting might be.
5747 * Similar to the various existing ConditionXYZ= settings for
5748 units, there are now matching AssertXYZ= settings. While
5749 failing conditions cause a unit to be skipped, but its job
5750 to succeed, failing assertions declared like this will cause
5751 a unit start operation and its job to fail.
5753 * hostnamed now knows a new chassis type "embedded".
5755 * systemctl gained a new "edit" command. When used on a unit
5756 file, this allows extending unit files with .d/ drop-in
5757 configuration snippets or editing the full file (after
5758 copying it from /usr/lib to /etc). This will invoke the
5759 user's editor (as configured with $EDITOR), and reload the
5760 modified configuration after editing.
5762 * "systemctl status" now shows the suggested enablement state
5763 for a unit, as declared in the (usually vendor-supplied)
5764 system preset files.
5766 * nss-myhostname will now resolve the single-label host name
5767 "gateway" to the locally configured default IP routing
5768 gateways, ordered by their metrics. This assigns a stable
5769 name to the used gateways, regardless which ones are
5770 currently configured. Note that the name will only be
5771 resolved after all other name sources (if nss-myhostname is
5772 configured properly) and should hence not negatively impact
5773 systems that use the single-label host name "gateway" in
5776 * systemd-inhibit now allows filtering by mode when listing
5779 * Scope and service units gained a new "Delegate" boolean
5780 property, which, when set, allows processes running inside the
5781 unit to further partition resources. This is primarily
5782 useful for systemd user instances as well as container
5785 * journald will now pick up audit messages directly from
5786 the kernel, and log them like any other log message. The
5787 audit fields are split up and fully indexed. This means that
5788 journalctl in many ways is now a (nicer!) alternative to
5789 ausearch, the traditional audit client. Note that this
5790 implements only a minimal audit client. If you want the
5791 special audit modes like reboot-on-log-overflow, please use
5792 the traditional auditd instead, which can be used in
5793 parallel to journald.
5795 * The ConditionSecurity= unit file option now understands the
5796 special string "audit" to check whether auditing is
5799 * journalctl gained two new commands --vacuum-size= and
5800 --vacuum-time= to delete old journal files until the
5801 remaining ones take up no more than the specified size on disk,
5802 or are not older than the specified time.
5804 * A new, native PPPoE library has been added to sd-network,
5805 systemd's library of light-weight networking protocols. This
5806 library will be used in a future version of networkd to
5807 enable PPPoE communication without an external pppd daemon.
5809 * The busctl tool now understands a new "capture" verb that
5810 works similar to "monitor", but writes a packet capture
5811 trace to STDOUT that can be redirected to a file which is
5812 compatible with libcap's capture file format. This can then
5813 be loaded in Wireshark and similar tools to inspect bus
5816 * The busctl tool now understands a new "tree" verb that shows
5817 the object trees of a specific service on the bus, or of all
5820 * The busctl tool now understands a new "introspect" verb that
5821 shows all interfaces and members of objects on the bus,
5822 including their signature and values. This is particularly
5823 useful to get more information about bus objects shown by
5824 the new "busctl tree" command.
5826 * The busctl tool now understands new verbs "call",
5827 "set-property" and "get-property" for invoking bus method
5828 calls, setting and getting bus object properties in a
5831 * busctl gained a new --augment-creds= argument that controls
5832 whether the tool shall augment credential information it
5833 gets from the bus with data from /proc, in a possibly
5836 * nspawn's --link-journal= switch gained two new values
5837 "try-guest" and "try-host" that work like "guest" and
5838 "host", but do not fail if the host has no persistent
5839 journalling enabled. -j is now equivalent to
5840 --link-journal=try-guest.
5842 * macvlan network devices created by nspawn will now have
5843 stable MAC addresses.
5845 * A new SmackProcessLabel= unit setting has been added, which
5846 controls the SMACK security label processes forked off by
5847 the respective unit shall use.
5849 * If compiled with --enable-xkbcommon, systemd-localed will
5850 verify x11 keymap settings by compiling the given keymap. It
5851 will spew out warnings if the compilation fails. This
5852 requires libxkbcommon to be installed.
5854 * When a coredump is collected, a larger number of metadata
5855 fields is now collected and included in the journal records
5856 created for it. More specifically, control group membership,
5857 environment variables, memory maps, working directory,
5858 chroot directory, /proc/$PID/status, and a list of open file
5859 descriptors is now stored in the log entry.
5861 * The udev hwdb now contains DPI information for mice. For
5864 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
5866 * All systemd programs that read standalone configuration
5867 files in /etc now also support a corresponding series of
5868 .conf.d configuration directories in /etc/, /run/,
5869 /usr/local/lib/, /usr/lib/, and (if configured with
5870 --enable-split-usr) /lib/. In particular, the following
5871 configuration files now have corresponding configuration
5872 directories: system.conf user.conf, logind.conf,
5873 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
5874 resolved.conf, timesyncd.conf, journal-remote.conf, and
5875 journal-upload.conf. Note that distributions should use the
5876 configuration directories in /usr/lib/; the directories in
5877 /etc/ are reserved for the system administrator.
5879 * systemd-rfkill will no longer take the rfkill device name
5880 into account when storing rfkill state on disk, as the name
5881 might be dynamically assigned and not stable. Instead, the
5882 ID_PATH udev variable combined with the rfkill type (wlan,
5883 bluetooth, ...) is used.
5885 * A new service systemd-machine-id-commit.service has been
5886 added. When used on systems where /etc is read-only during
5887 boot, and /etc/machine-id is not initialized (but an empty
5888 file), this service will copy the temporary machine ID
5889 created as replacement into /etc after the system is fully
5890 booted up. This is useful for systems that are freshly
5891 installed with a non-initialized machine ID, but should get
5892 a fixed machine ID for subsequent boots.
5894 * networkd's .netdev files now provide a large set of
5895 configuration parameters for VXLAN devices. Similarly, the
5896 bridge port cost parameter is now configurable in .network
5897 files. There's also new support for configuring IP source
5898 routing. networkd .link files gained support for a new
5899 OriginalName= match that is useful to match against the
5900 original interface name the kernel assigned. .network files
5901 may include MTU= and MACAddress= fields for altering the MTU
5902 and MAC address while being connected to a specific network
5905 * The LUKS logic gained supported for configuring
5906 UUID-specific key files. There's also new support for naming
5907 LUKS device from the kernel command line, using the new
5908 luks.name= argument.
5910 * Timer units may now be transiently created via the bus API
5911 (this was previously already available for scope and service
5912 units). In addition it is now possible to create multiple
5913 transient units at the same time with a single bus call. The
5914 "systemd-run" tool has been updated to make use of this for
5915 running commands on a specified time, in at(1)-style.
5917 * tmpfiles gained support for "t" lines, for assigning
5918 extended attributes to files. Among other uses this may be
5919 used to assign SMACK labels to files.
5921 Contributions from: Alin Rauta, Alison Chaiken, Andrej
5922 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
5923 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
5924 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
5925 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
5926 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
5927 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
5928 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
5929 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
5930 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
5931 Michael Biebl, Michael Chapman, Michael Marineau, Michal
5932 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
5933 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
5934 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
5935 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
5936 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
5937 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
5938 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
5940 — Berlin, 2014-12-10
5944 * journalctl gained the new options -t/--identifier= to match
5945 on the syslog identifier (aka "tag"), as well as --utc to
5946 show log timestamps in the UTC timezone. journalctl now also
5947 accepts -n/--lines=all to disable line capping in a pager.
5949 * journalctl gained a new switch, --flush, that synchronously
5950 flushes logs from /run/log/journal to /var/log/journal if
5951 persistent storage is enabled. systemd-journal-flush.service
5952 now waits until the operation is complete.
5954 * Services can notify the manager before they start a reload
5955 (by sending RELOADING=1) or shutdown (by sending
5956 STOPPING=1). This allows the manager to track and show the
5957 internal state of daemons and closes a race condition when
5958 the process is still running but has closed its D-Bus
5961 * Services with Type=oneshot do not have to have any ExecStart
5964 * User units are now loaded also from
5965 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
5966 /run/systemd/user directory that was already previously
5967 supported, but is under the control of the user.
5969 * Job timeouts (i.e. timeouts on the time a job that is
5970 queued stays in the run queue) can now optionally result in
5971 immediate reboot or power-off actions (JobTimeoutAction= and
5972 JobTimeoutRebootArgument=). This is useful on ".target"
5973 units, to limit the maximum time a target remains
5974 undispatched in the run queue, and to trigger an emergency
5975 operation in such a case. This is now used by default to
5976 turn off the system if boot-up (as defined by everything in
5977 basic.target) hangs and does not complete for at least
5978 15min. Also, if power-off or reboot hang for at least 30min
5979 an immediate power-off/reboot operation is triggered. This
5980 functionality is particularly useful to increase reliability
5981 on embedded devices, but also on laptops which might
5982 accidentally get powered on when carried in a backpack and
5983 whose boot stays stuck in a hard disk encryption passphrase
5986 * systemd-logind can be configured to also handle lid switch
5987 events even when the machine is docked or multiple displays
5988 are attached (HandleLidSwitchDocked= option).
5990 * A helper binary and a service have been added which can be
5991 used to resume from hibernation in the initramfs. A
5992 generator will parse the resume= option on the kernel
5993 command line to trigger resume.
5995 * A user console daemon systemd-consoled has been
5996 added. Currently, it is a preview, and will so far open a
5997 single terminal on each session of the user marked as
5998 Desktop=systemd-console.
6000 * Route metrics can be specified for DHCP routes added by
6003 * The SELinux context of socket-activated services can be set
6004 from the information provided by the networking stack
6005 (SELinuxContextFromNet= option).
6007 * Userspace firmware loading support has been removed and
6008 the minimum supported kernel version is thus bumped to 3.7.
6010 * Timeout for udev workers has been increased from 1 to 3
6011 minutes, but a warning will be printed after 1 minute to
6012 help diagnose kernel modules that take a long time to load.
6014 * Udev rules can now remove tags on devices with TAG-="foobar".
6016 * systemd's readahead implementation has been removed. In many
6017 circumstances it didn't give expected benefits even for
6018 rotational disk drives and was becoming less relevant in the
6019 age of SSDs. As none of the developers has been using
6020 rotating media anymore, and nobody stepped up to actively
6021 maintain this component of systemd it has now been removed.
6023 * Swap units can use Options= to specify discard options.
6024 Discard options specified for swaps in /etc/fstab are now
6027 * Docker containers are now detected as a separate type of
6030 * The Password Agent protocol gained support for queries where
6031 the user input is shown, useful e.g. for user names.
6032 systemd-ask-password gained a new --echo option to turn that
6035 * The default sysctl.d/ snippets will now set:
6037 net.core.default_qdisc = fq_codel
6039 This selects Fair Queuing Controlled Delay as the default
6040 queuing discipline for network interfaces. fq_codel helps
6041 fight the network bufferbloat problem. It is believed to be
6042 a good default with no tuning required for most workloads.
6043 Downstream distributions may override this choice. On 10Gbit
6044 servers that do not do forwarding, "fq" may perform better.
6045 Systems without a good clocksource should use "pfifo_fast".
6047 * If kdbus is enabled during build a new option BusPolicy= is
6048 available for service units, that allows locking all service
6049 processes into a stricter bus policy, in order to limit
6050 access to various bus services, or even hide most of them
6051 from the service's view entirely.
6053 * networkctl will now show the .network and .link file
6054 networkd has applied to a specific interface.
6056 * sd-login gained a new API call sd_session_get_desktop() to
6057 query which desktop environment has been selected for a
6060 * UNIX utmp support is now compile-time optional to support
6061 legacy-free systems.
6063 * systemctl gained two new commands "add-wants" and
6064 "add-requires" for pulling in units from specific targets
6067 * If the word "rescue" is specified on the kernel command line
6068 the system will now boot into rescue mode (aka
6069 rescue.target), which was previously available only by
6070 specifying "1" or "systemd.unit=rescue.target" on the kernel
6071 command line. This new kernel command line option nicely
6072 mirrors the already existing "emergency" kernel command line
6075 * New kernel command line options mount.usr=, mount.usrflags=,
6076 mount.usrfstype= have been added that match root=, rootflags=,
6077 rootfstype= but allow mounting a specific file system to
6080 * The $NOTIFY_SOCKET is now also passed to control processes of
6081 services, not only the main process.
6083 * This version reenables support for fsck's -l switch. This
6084 means at least version v2.25 of util-linux is required for
6085 operation, otherwise dead-locks on device nodes may
6086 occur. Again: you need to update util-linux to at least
6087 v2.25 when updating systemd to v217.
6089 * The "multi-seat-x" tool has been removed from systemd, as
6090 its functionality has been integrated into X servers 1.16,
6091 and the tool is hence redundant. It is recommended to update
6092 display managers invoking this tool to simply invoke X
6093 directly from now on, again.
6095 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
6096 message flag has been added for all of systemd's polkit
6097 authenticated method calls has been added. In particular this
6098 now allows optional interactive authorization via polkit for
6099 many of PID1's privileged operations such as unit file
6100 enabling and disabling.
6102 * "udevadm hwdb --update" learnt a new switch "--usr" for
6103 placing the rebuilt hardware database in /usr instead of
6104 /etc. When used only hardware database entries stored in
6105 /usr will be used, and any user database entries in /etc are
6106 ignored. This functionality is useful for vendors to ship a
6107 pre-built database on systems where local configuration is
6108 unnecessary or unlikely.
6110 * Calendar time specifications in .timer units now also
6111 understand the strings "semi-annually", "quarterly" and
6112 "minutely" as shortcuts (in addition to the preexisting
6113 "anually", "hourly", ...).
6115 * systemd-tmpfiles will now correctly create files in /dev
6116 at boot which are marked for creation only at boot. It is
6117 recommended to always create static device nodes with 'c!'
6118 and 'b!', so that they are created only at boot and not
6119 overwritten at runtime.
6121 * When the watchdog logic is used for a service (WatchdogSec=)
6122 and the watchdog timeout is hit the service will now be
6123 terminated with SIGABRT (instead of just SIGTERM), in order
6124 to make sure a proper coredump and backtrace is
6125 generated. This ensures that hanging services will result in
6126 similar coredump/backtrace behaviour as services that hit a
6129 Contributions from: Andreas Henriksson, Andrei Borzenkov,
6130 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
6131 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
6132 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
6133 Herrmann, David Sommerseth, David Strauss, Emil Renner
6134 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
6135 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
6136 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
6137 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
6138 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
6139 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
6140 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
6141 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
6142 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
6143 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
6144 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
6145 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
6146 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
6147 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
6148 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
6149 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
6152 — Berlin, 2014-10-28
6156 * timedated no longer reads NTP implementation unit names from
6157 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
6158 implementations should add a
6160 Conflicts=systemd-timesyncd.service
6162 to their unit files to take over and replace systemd's NTP
6163 default functionality.
6165 * systemd-sysusers gained a new line type "r" for configuring
6166 which UID/GID ranges to allocate system users/groups
6167 from. Lines of type "u" may now add an additional column
6168 that specifies the home directory for the system user to be
6169 created. Also, systemd-sysusers may now optionally read user
6170 information from STDIN instead of a file. This is useful for
6171 invoking it from RPM preinst scriptlets that need to create
6172 users before the first RPM file is installed since these
6173 files might need to be owned by them. A new
6174 %sysusers_create_inline RPM macro has been introduced to do
6175 just that. systemd-sysusers now updates the shadow files as
6176 well as the user/group databases, which should enhance
6177 compatibility with certain tools like grpck.
6179 * A number of bus APIs of PID 1 now optionally consult polkit to
6180 permit access for otherwise unprivileged clients under certain
6181 conditions. Note that this currently doesn't support
6182 interactive authentication yet, but this is expected to be
6183 added eventually, too.
6185 * /etc/machine-info now has new fields for configuring the
6186 deployment environment of the machine, as well as the
6187 location of the machine. hostnamectl has been updated with
6188 new command to update these fields.
6190 * systemd-timesyncd has been updated to automatically acquire
6191 NTP server information from systemd-networkd, which might
6192 have been discovered via DHCP.
6194 * systemd-resolved now includes a caching DNS stub resolver
6195 and a complete LLMNR name resolution implementation. A new
6196 NSS module "nss-resolve" has been added which can be used
6197 instead of glibc's own "nss-dns" to resolve hostnames via
6198 systemd-resolved. Hostnames, addresses and arbitrary RRs may
6199 be resolved via systemd-resolved D-Bus APIs. In contrast to
6200 the glibc internal resolver systemd-resolved is aware of
6201 multi-homed system, and keeps DNS server and caches separate
6202 and per-interface. Queries are sent simultaneously on all
6203 interfaces that have DNS servers configured, in order to
6204 properly handle VPNs and local LANs which might resolve
6205 separate sets of domain names. systemd-resolved may acquire
6206 DNS server information from systemd-networkd automatically,
6207 which in turn might have discovered them via DHCP. A tool
6208 "systemd-resolve-host" has been added that may be used to
6209 query the DNS logic in resolved. systemd-resolved implements
6210 IDNA and automatically uses IDNA or UTF-8 encoding depending
6211 on whether classic DNS or LLMNR is used as transport. In the
6212 next releases we intend to add a DNSSEC and mDNS/DNS-SD
6213 implementation to systemd-resolved.
6215 * A new NSS module nss-mymachines has been added, that
6216 automatically resolves the names of all local registered
6217 containers to their respective IP addresses.
6219 * A new client tool "networkctl" for systemd-networkd has been
6220 added. It currently is entirely passive and will query
6221 networking configuration from udev, rtnetlink and networkd,
6222 and present it to the user in a very friendly
6223 way. Eventually, we hope to extend it to become a full
6224 control utility for networkd.
6226 * .socket units gained a new DeferAcceptSec= setting that
6227 controls the kernels' TCP_DEFER_ACCEPT sockopt for
6228 TCP. Similarly, support for controlling TCP keep-alive
6229 settings has been added (KeepAliveTimeSec=,
6230 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
6231 turning off Nagle's algorithm on TCP has been added
6234 * logind learned a new session type "web", for use in projects
6235 like Cockpit which register web clients as PAM sessions.
6237 * timer units with at least one OnCalendar= setting will now
6238 be started only after time-sync.target has been
6239 reached. This way they will not elapse before the system
6240 clock has been corrected by a local NTP client or
6241 similar. This is particular useful on RTC-less embedded
6242 machines, that come up with an invalid system clock.
6244 * systemd-nspawn's --network-veth= switch should now result in
6245 stable MAC addresses for both the outer and the inner side
6248 * systemd-nspawn gained a new --volatile= switch for running
6249 container instances with /etc or /var unpopulated.
6251 * The kdbus client code has been updated to use the new Linux
6252 3.17 memfd subsystem instead of the old kdbus-specific one.
6254 * systemd-networkd's DHCP client and server now support
6255 FORCERENEW. There are also new configuration options to
6256 configure the vendor client identifier and broadcast mode
6259 * systemd will no longer inform the kernel about the current
6260 timezone, as this is necessarily incorrect and racy as the
6261 kernel has no understanding of DST and similar
6262 concepts. This hence means FAT timestamps will be always
6263 considered UTC, similar to what Android is already
6264 doing. Also, when the RTC is configured to the local time
6265 (rather than UTC) systemd will never synchronize back to it,
6266 as this might confuse Windows at a later boot.
6268 * systemd-analyze gained a new command "verify" for offline
6269 validation of unit files.
6271 * systemd-networkd gained support for a couple of additional
6272 settings for bonding networking setups. Also, the metric for
6273 statically configured routes may now be configured. For
6274 network interfaces where this is appropriate the peer IP
6275 address may now be configured.
6277 * systemd-networkd's DHCP client will no longer request
6278 broadcasting by default, as this tripped up some networks.
6279 For hardware where broadcast is required the feature should
6280 be switched back on using RequestBroadcast=yes.
6282 * systemd-networkd will now set up IPv4LL addresses (when
6283 enabled) even if DHCP is configured successfully.
6285 * udev will now default to respect network device names given
6286 by the kernel when the kernel indicates that these are
6287 predictable. This behavior can be tweaked by changing
6288 NamePolicy= in the relevant .link file.
6290 * A new library systemd-terminal has been added that
6291 implements full TTY stream parsing and rendering. This
6292 library is supposed to be used later on for implementing a
6293 full userspace VT subsystem, replacing the current kernel
6296 * A new tool systemd-journal-upload has been added to push
6297 journal data to a remote system running
6298 systemd-journal-remote.
6300 * journald will no longer forward all local data to another
6301 running syslog daemon. This change has been made because
6302 rsyslog (which appears to be the most commonly used syslog
6303 implementation these days) no longer makes use of this, and
6304 instead pulls the data out of the journal on its own. Since
6305 forwarding the messages to a non-existent syslog server is
6306 more expensive than we assumed we have now turned this
6307 off. If you run a syslog server that is not a recent rsyslog
6308 version, you have to turn this option on again
6309 (ForwardToSyslog= in journald.conf).
6311 * journald now optionally supports the LZ4 compressor for
6312 larger journal fields. This compressor should perform much
6313 better than XZ which was the previous default.
6315 * machinectl now shows the IP addresses of local containers,
6316 if it knows them, plus the interface name of the container.
6318 * A new tool "systemd-escape" has been added that makes it
6319 easy to escape strings to build unit names and similar.
6321 * sd_notify() messages may now include a new ERRNO= field
6322 which is parsed and collected by systemd and shown among the
6323 "systemctl status" output for a service.
6325 * A new component "systemd-firstboot" has been added that
6326 queries the most basic systemd information (timezone,
6327 hostname, root password) interactively on first
6328 boot. Alternatively it may also be used to provision these
6329 things offline on OS images installed into directories.
6331 * The default sysctl.d/ snippets will now set
6333 net.ipv4.conf.default.promote_secondaries=1
6335 This has the benefit of no flushing secondary IP addresses
6336 when primary addresses are removed.
6338 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
6339 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
6340 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
6341 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
6342 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
6343 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
6344 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
6345 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
6346 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
6347 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
6348 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
6349 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
6350 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
6351 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
6352 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
6354 — Berlin, 2014-08-19
6358 * A new tool systemd-sysusers has been added. This tool
6359 creates system users and groups in /etc/passwd and
6360 /etc/group, based on static declarative system user/group
6361 definitions in /usr/lib/sysusers.d/. This is useful to
6362 enable factory resets and volatile systems that boot up with
6363 an empty /etc directory, and thus need system users and
6364 groups created during early boot. systemd now also ships
6365 with two default sysusers.d/ files for the most basic
6366 users and groups systemd and the core operating system
6369 * A new tmpfiles snippet has been added that rebuilds the
6370 essential files in /etc on boot, should they be missing.
6372 * A directive for ensuring automatic clean-up of
6373 /var/cache/man/ has been removed from the default
6374 configuration. This line should now be shipped by the man
6375 implementation. The necessary change has been made to the
6376 man-db implementation. Note that you need to update your man
6377 implementation to one that ships this line, otherwise no
6378 automatic clean-up of /var/cache/man will take place.
6380 * A new condition ConditionNeedsUpdate= has been added that
6381 may conditionalize services to only run when /etc or /var
6382 are "older" than the vendor operating system resources in
6383 /usr. This is useful for reconstructing or updating /etc
6384 after an offline update of /usr or a factory reset, on the
6385 next reboot. Services that want to run once after such an
6386 update or reset should use this condition and order
6387 themselves before the new systemd-update-done.service, which
6388 will mark the two directories as fully updated. A number of
6389 service files have been added making use of this, to rebuild
6390 the udev hardware database, the journald message catalog and
6391 dynamic loader cache (ldconfig). The systemd-sysusers tool
6392 described above also makes use of this now. With this in
6393 place it is now possible to start up a minimal operating
6394 system with /etc empty cleanly. For more information on the
6395 concepts involved see this recent blog story:
6397 http://0pointer.de/blog/projects/stateless.html
6399 * A new system group "input" has been introduced, and all
6400 input device nodes get this group assigned. This is useful
6401 for system-level software to get access to input devices. It
6402 complements what is already done for "audio" and "video".
6404 * systemd-networkd learnt minimal DHCPv4 server support in
6405 addition to the existing DHCPv4 client support. It also
6406 learnt DHCPv6 client and IPv6 Router Solicitation client
6407 support. The DHCPv4 client gained support for static routes
6408 passed in from the server. Note that the [DHCPv4] section
6409 known in older systemd-networkd versions has been renamed to
6410 [DHCP] and is now also used by the DHCPv6 client. Existing
6411 .network files using settings of this section should be
6412 updated, though compatibility is maintained. Optionally, the
6413 client hostname may now be sent to the DHCP server.
6415 * networkd gained support for vxlan virtual networks as well
6416 as tun/tap and dummy devices.
6418 * networkd gained support for automatic allocation of address
6419 ranges for interfaces from a system-wide pool of
6420 addresses. This is useful for dynamically managing a large
6421 number of interfaces with a single network configuration
6422 file. In particular this is useful to easily assign
6423 appropriate IP addresses to the veth links of a large number
6424 of nspawn instances.
6426 * RPM macros for processing sysusers, sysctl and binfmt
6427 drop-in snippets at package installation time have been
6430 * The /etc/os-release file should now be placed in
6431 /usr/lib/os-release. The old location is automatically
6432 created as symlink. /usr/lib is the more appropriate
6433 location of this file, since it shall actually describe the
6434 vendor operating system shipped in /usr, and not the
6435 configuration stored in /etc.
6437 * .mount units gained a new boolean SloppyOptions= setting
6438 that maps to mount(8)'s -s option which enables permissive
6439 parsing of unknown mount options.
6441 * tmpfiles learnt a new "L+" directive which creates a symlink
6442 but (unlike "L") deletes a pre-existing file first, should
6443 it already exist and not already be the correct
6444 symlink. Similarly, "b+", "c+" and "p+" directives have been
6445 added as well, which create block and character devices, as
6446 well as fifos in the filesystem, possibly removing any
6447 pre-existing files of different types.
6449 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
6450 'argument' field (which so far specified the source to
6451 symlink/copy the files from) is now optional. If omitted the
6452 same file os copied from /usr/share/factory/ suffixed by the
6453 full destination path. This is useful for populating /etc
6454 with essential files, by copying them from vendor defaults
6455 shipped in /usr/share/factory/etc.
6457 * A new command "systemctl preset-all" has been added that
6458 applies the service preset settings to all installed unit
6459 files. A new switch --preset-mode= has been added that
6460 controls whether only enable or only disable operations
6463 * A new command "systemctl is-system-running" has been added
6464 that allows checking the overall state of the system, for
6465 example whether it is fully up and running.
6467 * When the system boots up with an empty /etc, the equivalent
6468 to "systemctl preset-all" is executed during early boot, to
6469 make sure all default services are enabled after a factory
6472 * systemd now contains a minimal preset file that enables the
6473 most basic services systemd ships by default.
6475 * Unit files' [Install] section gained a new DefaultInstance=
6476 field for defining the default instance to create if a
6477 template unit is enabled with no instance specified.
6479 * A new passive target cryptsetup-pre.target has been added
6480 that may be used by services that need to make they run and
6481 finish before the first LUKS cryptographic device is set up.
6483 * The /dev/loop-control and /dev/btrfs-control device nodes
6484 are now owned by the "disk" group by default, opening up
6485 access to this group.
6487 * systemd-coredump will now automatically generate a
6488 stack trace of all core dumps taking place on the system,
6489 based on elfutils' libdw library. This stack trace is logged
6492 * systemd-coredump may now optionally store coredumps directly
6493 on disk (in /var/lib/systemd/coredump, possibly compressed),
6494 instead of storing them unconditionally in the journal. This
6495 mode is the new default. A new configuration file
6496 /etc/systemd/coredump.conf has been added to configure this
6497 and other parameters of systemd-coredump.
6499 * coredumpctl gained a new "info" verb to show details about a
6500 specific coredump. A new switch "-1" has also been added
6501 that makes sure to only show information about the most
6502 recent entry instead of all entries. Also, as the tool is
6503 generally useful now the "systemd-" prefix of the binary
6504 name has been removed. Distributions that want to maintain
6505 compatibility with the old name should add a symlink from
6506 the old name to the new name.
6508 * journald's SplitMode= now defaults to "uid". This makes sure
6509 that unprivileged users can access their own coredumps with
6510 coredumpctl without restrictions.
6512 * New kernel command line options "systemd.wants=" (for
6513 pulling an additional unit during boot), "systemd.mask="
6514 (for masking a specific unit for the boot), and
6515 "systemd.debug-shell" (for enabling the debug shell on tty9)
6516 have been added. This is implemented in the new generator
6517 "systemd-debug-generator".
6519 * systemd-nspawn will now by default filter a couple of
6520 syscalls for containers, among them those required for
6521 kernel module loading, direct x86 IO port access, swap
6522 management, and kexec. Most importantly though
6523 open_by_handle_at() is now prohibited for containers,
6524 closing a hole similar to a recently discussed vulnerability
6525 in docker regarding access to files on file hierarchies the
6526 container should normally not have access to. Note that, for
6527 nspawn, we generally make no security claims anyway (and
6528 this is explicitly documented in the man page), so this is
6529 just a fix for one of the most obvious problems.
6531 * A new man page file-hierarchy(7) has been added that
6532 contains a minimized, modernized version of the file system
6533 layout systemd expects, similar in style to the FHS
6534 specification or hier(5). A new tool systemd-path(1) has
6535 been added to query many of these paths for the local
6538 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
6539 longer done. Since the directory now has a per-user size
6540 limit, and is cleaned on logout this appears unnecessary,
6541 in particular since this now brings the lifecycle of this
6542 directory closer in line with how IPC objects are handled.
6544 * systemd.pc now exports a number of additional directories,
6545 including $libdir (which is useful to identify the library
6546 path for the primary architecture of the system), and a
6547 couple of drop-in directories.
6549 * udev's predictable network interface names now use the dev_port
6550 sysfs attribute, introduced in linux 3.15 instead of dev_id to
6551 distinguish between ports of the same PCI function. dev_id should
6552 only be used for ports using the same HW address, hence the need
6555 * machined has been updated to export the OS version of a
6556 container (read from /etc/os-release and
6557 /usr/lib/os-release) on the bus. This is now shown in
6558 "machinectl status" for a machine.
6560 * A new service setting RestartForceExitStatus= has been
6561 added. If configured to a set of exit signals or process
6562 return values, the service will be restarted when the main
6563 daemon process exits with any of them, regardless of the
6566 * systemctl's -H switch for connecting to remote systemd
6567 machines has been extended so that it may be used to
6568 directly connect to a specific container on the
6569 host. "systemctl -H root@foobar:waldi" will now connect as
6570 user "root" to host "foobar", and then proceed directly to
6571 the container named "waldi". Note that currently you have to
6572 authenticate as user "root" for this to work, as entering
6573 containers is a privileged operation.
6575 Contributions from: Andreas Henriksson, Benjamin Steinwender,
6576 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
6577 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
6578 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
6579 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
6580 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
6581 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
6582 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
6583 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
6584 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
6585 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
6586 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
6588 — Berlin, 2014-07-03
6592 * As an experimental feature, udev now tries to lock the
6593 disk device node (flock(LOCK_SH|LOCK_NB)) while it
6594 executes events for the disk or any of its partitions.
6595 Applications like partitioning programs can lock the
6596 disk device node (flock(LOCK_EX)) and claim temporary
6597 device ownership that way; udev will entirely skip all event
6598 handling for this disk and its partitions. If the disk
6599 was opened for writing, the close will trigger a partition
6600 table rescan in udev's "watch" facility, and if needed
6601 synthesize "change" events for the disk and all its partitions.
6602 This is now unconditionally enabled, and if it turns out to
6603 cause major problems, we might turn it on only for specific
6604 devices, or might need to disable it entirely. Device Mapper
6605 devices are excluded from this logic.
6607 * We temporarily dropped the "-l" switch for fsck invocations,
6608 since they collide with the flock() logic above. util-linux
6609 upstream has been changed already to avoid this conflict,
6610 and we will re-add "-l" as soon as util-linux with this
6611 change has been released.
6613 * The dependency on libattr has been removed. Since a long
6614 time, the extended attribute calls have moved to glibc, and
6615 libattr is thus unnecessary.
6617 * Virtualization detection works without privileges now. This
6618 means the systemd-detect-virt binary no longer requires
6619 CAP_SYS_PTRACE file capabilities, and our daemons can run
6620 with fewer privileges.
6622 * systemd-networkd now runs under its own "systemd-network"
6623 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
6624 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
6625 loses the ability to write to files owned by root this way.
6627 * Similarly, systemd-resolved now runs under its own
6628 "systemd-resolve" user with no capabilities remaining.
6630 * Similarly, systemd-bus-proxyd now runs under its own
6631 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
6633 * systemd-networkd gained support for setting up "veth"
6634 virtual Ethernet devices for container connectivity, as well
6635 as GRE and VTI tunnels.
6637 * systemd-networkd will no longer automatically attempt to
6638 manually load kernel modules necessary for certain tunnel
6639 transports. Instead, it is assumed the kernel loads them
6640 automatically when required. This only works correctly on
6641 very new kernels. On older kernels, please consider adding
6642 the kernel modules to /etc/modules-load.d/ as a work-around.
6644 * The resolv.conf file systemd-resolved generates has been
6645 moved to /run/systemd/resolve/. If you have a symlink from
6646 /etc/resolv.conf, it might be necessary to correct it.
6648 * Two new service settings, ProtectHome= and ProtectSystem=,
6649 have been added. When enabled, they will make the user data
6650 (such as /home) inaccessible or read-only and the system
6651 (such as /usr) read-only, for specific services. This allows
6652 very light-weight per-service sandboxing to avoid
6653 modifications of user data or system files from
6654 services. These two new switches have been enabled for all
6655 of systemd's long-running services, where appropriate.
6657 * Socket units gained new SocketUser= and SocketGroup=
6658 settings to set the owner user and group of AF_UNIX sockets
6659 and FIFOs in the file system.
6661 * Socket units gained a new RemoveOnStop= setting. If enabled,
6662 all FIFOS and sockets in the file system will be removed
6663 when the specific socket unit is stopped.
6665 * Socket units gained a new Symlinks= setting. It takes a list
6666 of symlinks to create to file system sockets or FIFOs
6667 created by the specific Unix sockets. This is useful to
6668 manage symlinks to socket nodes with the same lifecycle as
6671 * The /dev/log socket and /dev/initctl FIFO have been moved to
6672 /run, and have been replaced by symlinks. This allows
6673 connecting to these facilities even if PrivateDevices=yes is
6674 used for a service (which makes /dev/log itself unavailable,
6675 but /run is left). This also has the benefit of ensuring
6676 that /dev only contains device nodes, directories and
6677 symlinks, and nothing else.
6679 * sd-daemon gained two new calls sd_pid_notify() and
6680 sd_pid_notifyf(). They are similar to sd_notify() and
6681 sd_notifyf(), but allow overriding of the source PID of
6682 notification messages if permissions permit this. This is
6683 useful to send notify messages on behalf of a different
6684 process (for example, the parent process). The
6685 systemd-notify tool has been updated to make use of this
6686 when sending messages (so that notification messages now
6687 originate from the shell script invoking systemd-notify and
6688 not the systemd-notify process itself. This should minimize
6689 a race where systemd fails to associate notification
6690 messages to services when the originating process already
6693 * A new "on-abnormal" setting for Restart= has been added. If
6694 set, it will result in automatic restarts on all "abnormal"
6695 reasons for a process to exit, which includes unclean
6696 signals, core dumps, timeouts and watchdog timeouts, but
6697 does not include clean and unclean exit codes or clean
6698 signals. Restart=on-abnormal is an alternative for
6699 Restart=on-failure for services that shall be able to
6700 terminate and avoid restarts on certain errors, by
6701 indicating so with an unclean exit code. Restart=on-failure
6702 or Restart=on-abnormal is now the recommended setting for
6703 all long-running services.
6705 * If the InaccessibleDirectories= service setting points to a
6706 mount point (or if there are any submounts contained within
6707 it), it is now attempted to completely unmount it, to make
6708 the file systems truly unavailable for the respective
6711 * The ReadOnlyDirectories= service setting and
6712 systemd-nspawn's --read-only parameter are now recursively
6713 applied to all submounts, too.
6715 * Mount units may now be created transiently via the bus APIs.
6717 * The support for SysV and LSB init scripts has been removed
6718 from the systemd daemon itself. Instead, it is now
6719 implemented as a generator that creates native systemd units
6720 from these scripts when needed. This enables us to remove a
6721 substantial amount of legacy code from PID 1, following the
6722 fact that many distributions only ship a very small number
6723 of LSB/SysV init scripts nowadays.
6725 * Privileged Xen (dom0) domains are not considered
6726 virtualization anymore by the virtualization detection
6727 logic. After all, they generally have unrestricted access to
6728 the hardware and usually are used to manage the unprivileged
6731 * systemd-tmpfiles gained a new "C" line type, for copying
6732 files or entire directories.
6734 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
6735 lines. So far, they have been non-globbing versions of the
6736 latter, and have thus been redundant. In future, it is
6737 recommended to only use "z". "m" has hence been removed
6738 from the documentation, even though it stays supported.
6740 * A tmpfiles snippet to recreate the most basic structure in
6741 /var has been added. This is enough to create the /var/run →
6742 /run symlink and create a couple of structural
6743 directories. This allows systems to boot up with an empty or
6744 volatile /var. Of course, while with this change, the core OS
6745 now is capable with dealing with a volatile /var, not all
6746 user services are ready for it. However, we hope that sooner
6747 or later, many service daemons will be changed upstream so
6748 that they are able to automatically create their necessary
6749 directories in /var at boot, should they be missing. This is
6750 the first step to allow state-less systems that only require
6751 the vendor image for /usr to boot.
6753 * systemd-nspawn has gained a new --tmpfs= switch to mount an
6754 empty tmpfs instance to a specific directory. This is
6755 particularly useful for making use of the automatic
6756 reconstruction of /var (see above), by passing --tmpfs=/var.
6758 * Access modes specified in tmpfiles snippets may now be
6759 prefixed with "~", which indicates that they shall be masked
6760 by whether the existing file or directory is currently
6761 writable, readable or executable at all. Also, if specified,
6762 the sgid/suid/sticky bits will be masked for all
6765 * A new passive target unit "network-pre.target" has been
6766 added which is useful for services that shall run before any
6767 network is configured, for example firewall scripts.
6769 * The "floppy" group that previously owned the /dev/fd*
6770 devices is no longer used. The "disk" group is now used
6771 instead. Distributions should probably deprecate usage of
6774 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
6775 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
6776 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
6777 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
6778 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
6779 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
6782 — Berlin, 2014-06-11
6786 * A new "systemd-timesyncd" daemon has been added for
6787 synchronizing the system clock across the network. It
6788 implements an SNTP client. In contrast to NTP
6789 implementations such as chrony or the NTP reference server,
6790 this only implements a client side, and does not bother with
6791 the full NTP complexity, focusing only on querying time from
6792 one remote server and synchronizing the local clock to
6793 it. Unless you intend to serve NTP to networked clients or
6794 want to connect to local hardware clocks, this simple NTP
6795 client should be more than appropriate for most
6796 installations. The daemon runs with minimal privileges, and
6797 has been hooked up with networkd to only operate when
6798 network connectivity is available. The daemon saves the
6799 current clock to disk every time a new NTP sync has been
6800 acquired, and uses this to possibly correct the system clock
6801 early at bootup, in order to accommodate for systems that
6802 lack an RTC such as the Raspberry Pi and embedded devices,
6803 and to make sure that time monotonically progresses on these
6804 systems, even if it is not always correct. To make use of
6805 this daemon, a new system user and group "systemd-timesync"
6806 needs to be created on installation of systemd.
6808 * The queue "seqnum" interface of libudev has been disabled, as
6809 it was generally incompatible with device namespacing as
6810 sequence numbers of devices go "missing" if the devices are
6811 part of a different namespace.
6813 * "systemctl list-timers" and "systemctl list-sockets" gained
6814 a --recursive switch for showing units of these types also
6815 for all local containers, similar in style to the already
6816 supported --recursive switch for "systemctl list-units".
6818 * A new RebootArgument= setting has been added for service
6819 units, which may be used to specify a kernel reboot argument
6820 to use when triggering reboots with StartLimitAction=.
6822 * A new FailureAction= setting has been added for service
6823 units which may be used to specify an operation to trigger
6824 when a service fails. This works similarly to
6825 StartLimitAction=, but unlike it, controls what is done
6826 immediately rather than only after several attempts to
6827 restart the service in question.
6829 * hostnamed got updated to also expose the kernel name,
6830 release, and version on the bus. This is useful for
6831 executing commands like hostnamectl with the -H switch.
6832 systemd-analyze makes use of this to properly display
6833 details when running non-locally.
6835 * The bootchart tool can now show cgroup information in the
6836 graphs it generates.
6838 * The CFS CPU quota cgroup attribute is now exposed for
6839 services. The new CPUQuota= switch has been added for this
6840 which takes a percentage value. Setting this will have the
6841 result that a service may never get more CPU time than the
6842 specified percentage, even if the machine is otherwise idle.
6844 * systemd-networkd learned IPIP and SIT tunnel support.
6846 * LSB init scripts exposing a dependency on $network will now
6847 get a dependency on network-online.target rather than simply
6848 network.target. This should bring LSB handling closer to
6849 what it was on SysV systems.
6851 * A new fsck.repair= kernel option has been added to control
6852 how fsck shall deal with unclean file systems at boot.
6854 * The (.ini) configuration file parser will now silently
6855 ignore sections whose name begins with "X-". This may be
6856 used to maintain application-specific extension sections in unit
6859 * machined gained a new API to query the IP addresses of
6860 registered containers. "machinectl status" has been updated
6861 to show these addresses in its output.
6863 * A new call sd_uid_get_display() has been added to the
6864 sd-login APIs for querying the "primary" session of a
6865 user. The "primary" session of the user is elected from the
6866 user's sessions and generally a graphical session is
6867 preferred over a text one.
6869 * A minimal systemd-resolved daemon has been added. It
6870 currently simply acts as a companion to systemd-networkd and
6871 manages resolv.conf based on per-interface DNS
6872 configuration, possibly supplied via DHCP. In the long run
6873 we hope to extend this into a local DNSSEC enabled DNS and
6876 * The systemd-networkd-wait-online tool is now enabled by
6877 default. It will delay network-online.target until a network
6878 connection has been configured. The tool primarily integrates
6879 with networkd, but will also make a best effort to make sense
6880 of network configuration performed in some other way.
6882 * Two new service options StartupCPUShares= and
6883 StartupBlockIOWeight= have been added that work similarly to
6884 CPUShares= and BlockIOWeight= however only apply during
6885 system startup. This is useful to prioritize certain services
6886 differently during bootup than during normal runtime.
6888 * hostnamed has been changed to prefer the statically
6889 configured hostname in /etc/hostname (unless set to
6890 'localhost' or empty) over any dynamic one supplied by
6891 dhcp. With this change, the rules for picking the hostname
6892 match more closely the rules of other configuration settings
6893 where the local administrator's configuration in /etc always
6894 overrides any other settings.
6896 Contributions from: Ali H. Caliskan, Alison Chaiken, Bas van
6897 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
6898 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
6899 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
6900 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
6901 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
6902 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
6903 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
6904 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
6905 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
6906 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
6907 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
6908 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
6909 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
6910 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
6911 Lindskog, WaLyong Cho, Will Woods, Zbigniew
6914 — Beijing, 2014-05-28
6918 * When restoring the screen brightness at boot, stay away from
6919 the darkest setting or from the lowest 5% of the available
6920 range, depending on which is the larger value of both. This
6921 should effectively protect the user from rebooting into a
6922 black screen, should the brightness have been set to minimum
6925 * sd-login gained a new sd_machine_get_class() call to
6926 determine the class ("vm" or "container") of a machine
6927 registered with machined.
6929 * sd-login gained new calls
6930 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
6931 to query the identity of the peer of a local AF_UNIX
6932 connection. They operate similarly to their sd_pid_get_xyz()
6935 * PID 1 will now maintain a system-wide system state engine
6936 with the states "starting", "running", "degraded",
6937 "maintenance", "stopping". These states are bound to system
6938 startup, normal runtime, runtime with at least one failed
6939 service, rescue/emergency mode and system shutdown. This
6940 state is shown in the "systemctl status" output when no unit
6941 name is passed. It is useful to determine system state, in
6942 particularly when doing so for many systems or containers at
6945 * A new command "list-machines" has been added to "systemctl"
6946 that lists all local OS containers and shows their system
6947 state (see above), if systemd runs inside of them.
6949 * systemctl gained a new "-r" switch to recursively enumerate
6950 units on all local containers, when used with the
6951 "list-unit" command (which is the default one that is
6952 executed when no parameters are specified).
6954 * The GPT automatic partition discovery logic will now honour
6955 two GPT partition flags: one may be set on a partition to
6956 cause it to be mounted read-only, and the other may be set
6957 on a partition to ignore it during automatic discovery.
6959 * Two new GPT type UUIDs have been added for automatic root
6960 partition discovery, for 32-bit and 64-bit ARM. This is not
6961 particularly useful for discovering the root directory on
6962 these architectures during bare-metal boots (since UEFI is
6963 not common there), but still very useful to allow booting of
6964 ARM disk images in nspawn with the -i option.
6966 * MAC addresses of interfaces created with nspawn's
6967 --network-interface= switch will now be generated from the
6968 machine name, and thus be stable between multiple invocations
6971 * logind will now automatically remove all IPC objects owned
6972 by a user if she or he fully logs out. This makes sure that
6973 users who are logged out cannot continue to consume IPC
6974 resources. This covers SysV memory, semaphores and message
6975 queues as well as POSIX shared memory and message
6976 queues. Traditionally, SysV and POSIX IPC had no lifecycle
6977 limits. With this functionality, that is corrected. This may
6978 be turned off by using the RemoveIPC= switch of logind.conf.
6980 * The systemd-machine-id-setup and tmpfiles tools gained a
6981 --root= switch to operate on a specific root directory,
6984 * journald can now forward logged messages to the TTYs of all
6985 logged in users ("wall"). This is the default for all
6986 emergency messages now.
6988 * A new tool systemd-journal-remote has been added to stream
6989 journal log messages across the network.
6991 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
6992 controller trees are mounted into it. Note that the
6993 directories mounted beneath it are not read-only. This is a
6994 security measure and is particularly useful because glibc
6995 actually includes a search logic to pick any tmpfs it can
6996 find to implement shm_open() if /dev/shm is not available
6997 (which it might very well be in namespaced setups).
6999 * machinectl gained a new "poweroff" command to cleanly power
7000 down a local OS container.
7002 * The PrivateDevices= unit file setting will now also drop the
7003 CAP_MKNOD capability from the capability bound set, and
7004 imply DevicePolicy=closed.
7006 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
7007 comprehensively on all long-running systemd services where
7008 this is appropriate.
7010 * systemd-udevd will now run in a disassociated mount
7011 namespace. To mount directories from udev rules, make sure to
7012 pull in mount units via SYSTEMD_WANTS properties.
7014 * The kdbus support gained support for uploading policy into
7015 the kernel. sd-bus gained support for creating "monitoring"
7016 connections that can eavesdrop into all bus communication
7017 for debugging purposes.
7019 * Timestamps may now be specified in seconds since the UNIX
7020 epoch Jan 1st, 1970 by specifying "@" followed by the value
7023 * Native tcpwrap support in systemd has been removed. tcpwrap
7024 is old code, not really maintained anymore and has serious
7025 shortcomings, and better options such as firewalls
7026 exist. For setups that require tcpwrap usage, please
7027 consider invoking your socket-activated service via tcpd,
7028 like on traditional inetd.
7030 * A new system.conf configuration option
7031 DefaultTimerAccuracySec= has been added that controls the
7032 default AccuracySec= setting of .timer units.
7034 * Timer units gained a new WakeSystem= switch. If enabled,
7035 timers configured this way will cause the system to resume
7036 from system suspend (if the system supports that, which most
7039 * Timer units gained a new Persistent= switch. If enabled,
7040 timers configured this way will save to disk when they have
7041 been last triggered. This information is then used on next
7042 reboot to possible execute overdue timer events, that
7043 could not take place because the system was powered off.
7044 This enables simple anacron-like behaviour for timer units.
7046 * systemctl's "list-timers" will now also list the time a
7047 timer unit was last triggered in addition to the next time
7048 it will be triggered.
7050 * systemd-networkd will now assign predictable IPv4LL
7051 addresses to its local interfaces.
7053 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
7054 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
7055 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
7056 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
7057 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
7058 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
7059 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
7060 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
7063 — Berlin, 2014-03-25
7067 * A new unit file setting RestrictAddressFamilies= has been
7068 added to restrict which socket address families unit
7069 processes gain access to. This takes address family names
7070 like "AF_INET" or "AF_UNIX", and is useful to minimize the
7071 attack surface of services via exotic protocol stacks. This
7072 is built on seccomp system call filters.
7074 * Two new unit file settings RuntimeDirectory= and
7075 RuntimeDirectoryMode= have been added that may be used to
7076 manage a per-daemon runtime directories below /run. This is
7077 an alternative for setting up directory permissions with
7078 tmpfiles snippets, and has the advantage that the runtime
7079 directory's lifetime is bound to the daemon runtime and that
7080 the daemon starts up with an empty directory each time. This
7081 is particularly useful when writing services that drop
7082 privileges using the User= or Group= setting.
7084 * The DeviceAllow= unit setting now supports globbing for
7085 matching against device group names.
7087 * The systemd configuration file system.conf gained new
7088 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
7089 DefaultMemoryAccounting= to globally turn on/off accounting
7090 for specific resources (cgroups) for all units. These
7091 settings may still be overridden individually in each unit
7094 * systemd-gpt-auto-generator is now able to discover /srv and
7095 root partitions in addition to /home and swap partitions. It
7096 also supports LUKS-encrypted partitions now. With this in
7097 place, automatic discovery of partitions to mount following
7098 the Discoverable Partitions Specification
7099 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
7100 is now a lot more complete. This allows booting without
7101 /etc/fstab and without root= on the kernel command line on
7102 systems prepared appropriately.
7104 * systemd-nspawn gained a new --image= switch which allows
7105 booting up disk images and Linux installations on any block
7106 device that follow the Discoverable Partitions Specification
7107 (see above). This means that installations made with
7108 appropriately updated installers may now be started and
7109 deployed using container managers, completely
7110 unmodified. (We hope that libvirt-lxc will add support for
7111 this feature soon, too.)
7113 * systemd-nspawn gained a new --network-macvlan= setting to
7114 set up a private macvlan interface for the
7115 container. Similarly, systemd-networkd gained a new
7116 Kind=macvlan setting in .netdev files.
7118 * systemd-networkd now supports configuring local addresses
7121 * A new tool systemd-network-wait-online has been added to
7122 synchronously wait for network connectivity using
7125 * The sd-bus.h bus API gained a new sd_bus_track object for
7126 tracking the lifecycle of bus peers. Note that sd-bus.h is
7127 still not a public API though (unless you specify
7128 --enable-kdbus on the configure command line, which however
7129 voids your warranty and you get no API stability guarantee).
7131 * The $XDG_RUNTIME_DIR runtime directories for each user are
7132 now individual tmpfs instances, which has the benefit of
7133 introducing separate pools for each user, with individual
7134 size limits, and thus making sure that unprivileged clients
7135 can no longer negatively impact the system or other users by
7136 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
7137 RuntimeDirectorySize= has been introduced that allows
7138 controlling the default size limit for all users. It
7139 defaults to 10% of the available physical memory. This is no
7140 replacement for quotas on tmpfs though (which the kernel
7141 still does not support), as /dev/shm and /tmp are still
7142 shared resources used by both the system and unprivileged
7145 * logind will now automatically turn off automatic suspending
7146 on laptop lid close when more than one display is
7147 connected. This was previously expected to be implemented
7148 individually in desktop environments (such as GNOME),
7149 however has been added to logind now, in order to fix a
7150 boot-time race where a desktop environment might not have
7151 been started yet and thus not been able to take an inhibitor
7152 lock at the time where logind already suspends the system
7153 due to a closed lid.
7155 * logind will now wait at least 30s after each system
7156 suspend/resume cycle, and 3min after system boot before
7157 suspending the system due to a closed laptop lid. This
7158 should give USB docking stations and similar enough time to
7159 be probed and configured after system resume and boot in
7160 order to then act as suspend blocker.
7162 * systemd-run gained a new --property= setting which allows
7163 initialization of resource control properties (and others)
7164 for the created scope or service unit. Example: "systemd-run
7165 --property=BlockIOWeight=10 updatedb" may be used to run
7166 updatedb at a low block IO scheduling weight.
7168 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
7169 now also work in --scope mode.
7171 * When systemd is compiled with kdbus support, basic support
7172 for enforced policies is now in place. (Note that enabling
7173 kdbus still voids your warranty and no API compatibility
7176 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
7177 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
7178 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
7179 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
7180 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
7181 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
7182 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
7183 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
7184 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
7185 Zbigniew Jędrzejewski-Szmek
7187 — Berlin, 2014-03-12
7191 * systemd will now relabel /dev after loading the SMACK policy
7192 according to SMACK rules.
7194 * A new unit file option AppArmorProfile= has been added to
7195 set the AppArmor profile for the processes of a unit.
7197 * A new condition check ConditionArchitecture= has been added
7198 to conditionalize units based on the system architecture, as
7199 reported by uname()'s "machine" field.
7201 * systemd-networkd now supports matching on the system
7202 virtualization, architecture, kernel command line, host name
7205 * logind is now a lot more aggressive when suspending the
7206 machine due to a closed laptop lid. Instead of acting only
7207 on the lid close action, it will continuously watch the lid
7208 status and act on it. This is useful for laptops where the
7209 power button is on the outside of the chassis so that it can
7210 be reached without opening the lid (such as the Lenovo
7211 Yoga). On those machines, logind will now immediately
7212 re-suspend the machine if the power button has been
7213 accidentally pressed while the laptop was suspended and in a
7214 backpack or similar.
7216 * logind will now watch SW_DOCK switches and inhibit reaction
7217 to the lid switch if it is pressed. This means that logind
7218 will not suspend the machine anymore if the lid is closed
7219 and the system is docked, if the laptop supports SW_DOCK
7220 notifications via the input layer. Note that ACPI docking
7221 stations do not generate this currently. Also note that this
7222 logic is usually not fully sufficient and Desktop
7223 Environments should take a lid switch inhibitor lock when an
7224 external display is connected, as systemd will not watch
7227 * nspawn will now make use of the devices cgroup controller by
7228 default, and only permit creation of and access to the usual
7229 API device nodes like /dev/null or /dev/random, as well as
7230 access to (but not creation of) the pty devices.
7232 * We will now ship a default .network file for
7233 systemd-networkd that automatically configures DHCP for
7234 network interfaces created by nspawn's --network-veth or
7235 --network-bridge= switches.
7237 * systemd will now understand the usual M, K, G, T suffixes
7238 according to SI conventions (i.e. to the base 1000) when
7239 referring to throughput and hardware metrics. It will stay
7240 with IEC conventions (i.e. to the base 1024) for software
7241 metrics, according to what is customary according to
7242 Wikipedia. We explicitly document which base applies for
7243 each configuration option.
7245 * The DeviceAllow= setting in unit files now supports a syntax
7246 to whitelist an entire group of devices node majors at once,
7247 based on the /proc/devices listing. For example, with the
7248 string "char-pts", it is now possible to whitelist all
7249 current and future pseudo-TTYs at once.
7251 * sd-event learned a new "post" event source. Event sources of
7252 this type are triggered by the dispatching of any event
7253 source of a type that is not "post". This is useful for
7254 implementing clean-up and check event sources that are
7255 triggered by other work being done in the program.
7257 * systemd-networkd is no longer statically enabled, but uses
7258 the usual [Install] sections so that it can be
7259 enabled/disabled using systemctl. It still is enabled by
7262 * When creating a veth interface pair with systemd-nspawn, the
7263 host side will now be prefixed with "vb-" if
7264 --network-bridge= is used, and with "ve-" if --network-veth
7265 is used. This way, it is easy to distinguish these cases on
7266 the host, for example to apply different configuration to
7267 them with systemd-networkd.
7269 * The compatibility libraries for libsystemd-journal.so,
7270 libsystem-id128.so, libsystemd-login.so and
7271 libsystemd-daemon.so do not make use of IFUNC
7272 anymore. Instead, we now build libsystemd.so multiple times
7273 under these alternative names. This means that the footprint
7274 is drastically increased, but given that these are
7275 transitional compatibility libraries, this should not matter
7276 much. This change has been made necessary to support the ARM
7277 platform for these compatibility libraries, as the ARM
7278 toolchain is not really at the same level as the toolchain
7279 for other architectures like x86 and does not support
7280 IFUNC. Please make sure to use --enable-compat-libs only
7281 during a transitional period!
7283 * The .include syntax has been deprecated and is not documented
7284 anymore. Drop-in files in .d directories should be used instead.
7286 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
7287 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
7288 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
7289 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
7290 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
7291 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
7292 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
7293 Zbigniew Jędrzejewski-Szmek
7295 — Berlin, 2014-02-24
7299 * A new component "systemd-networkd" has been added that can
7300 be used to configure local network interfaces statically or
7301 via DHCP. It is capable of bringing up bridges, VLANs, and
7302 bonding. Currently, no hook-ups for interactive network
7303 configuration are provided. Use this for your initrd,
7304 container, embedded, or server setup if you need a simple,
7305 yet powerful, network configuration solution. This
7306 configuration subsystem is quite nifty, as it allows wildcard
7307 hotplug matching in interfaces. For example, with a single
7308 configuration snippet, you can configure that all Ethernet
7309 interfaces showing up are automatically added to a bridge,
7310 or similar. It supports link-sensing and more.
7312 * A new tool "systemd-socket-proxyd" has been added which can
7313 act as a bidirectional proxy for TCP sockets. This is
7314 useful for adding socket activation support to services that
7315 do not actually support socket activation, including virtual
7316 machines and the like.
7318 * Add a new tool to save/restore rfkill state on
7321 * Save/restore state of keyboard backlights in addition to
7322 display backlights on shutdown/boot.
7324 * udev learned a new SECLABEL{} construct to label device
7325 nodes with a specific security label when they appear. For
7326 now, only SECLABEL{selinux} is supported, but the syntax is
7327 prepared for additional security frameworks.
7329 * udev gained a new scheme to configure link-level attributes
7330 from files in /etc/systemd/network/*.link. These files can
7331 match against MAC address, device path, driver name and type,
7332 and will apply attributes like the naming policy, link speed,
7333 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
7334 address assignment policy (randomized, ...).
7336 * The configuration of network interface naming rules for
7337 "permanent interface names" has changed: a new NamePolicy=
7338 setting in the [Link] section of .link files determines the
7339 priority of possible naming schemes (onboard, slot, MAC,
7340 path). The default value of this setting is determined by
7341 /usr/lib/net/links/99-default.link. Old
7342 80-net-name-slot.rules udev configuration file has been
7343 removed, so local configuration overriding this file should
7344 be adapted to override 99-default.link instead.
7346 * When the User= switch is used in a unit file, also
7347 initialize $SHELL= based on the user database entry.
7349 * systemd no longer depends on libdbus. All communication is
7350 now done with sd-bus, systemd's low-level bus library
7353 * kdbus support has been added to PID 1 itself. When kdbus is
7354 enabled, this causes PID 1 to set up the system bus and
7355 enable support for a new ".busname" unit type that
7356 encapsulates bus name activation on kdbus. It works a little
7357 bit like ".socket" units, except for bus names. A new
7358 generator has been added that converts classic dbus1 service
7359 activation files automatically into native systemd .busname
7362 * sd-bus: add a light-weight vtable implementation that allows
7363 defining objects on the bus with a simple static const
7364 vtable array of its methods, signals and properties.
7366 * systemd will not generate or install static dbus
7367 introspection data anymore to /usr/share/dbus-1/interfaces,
7368 as the precise format of these files is unclear, and
7369 nothing makes use of it.
7371 * A proxy daemon is now provided to proxy clients connecting
7372 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
7373 compatibility with classic D-Bus.
7375 * A bus driver implementation has been added that supports the
7376 classic D-Bus bus driver calls on kdbus, also for
7377 compatibility purposes.
7379 * A new API "sd-event.h" has been added that implements a
7380 minimal event loop API built around epoll. It provides a
7381 couple of features that direct epoll usage is lacking:
7382 prioritization of events, scales to large numbers of timer
7383 events, per-event timer slack (accuracy), system-wide
7384 coalescing of timer events, exit handlers, watchdog
7385 supervision support using systemd's sd_notify() API, child
7388 * A new API "sd-rntl.h" has been added that provides an API
7389 around the route netlink interface of the kernel, similar in
7390 style to "sd-bus.h".
7392 * A new API "sd-dhcp-client.h" has been added that provides a
7393 small DHCPv4 client-side implementation. This is used by
7396 * There is a new kernel command line option
7397 "systemd.restore_state=0|1". When set to "0", none of the
7398 systemd tools will restore saved runtime state to hardware
7399 devices. More specifically, the rfkill and backlight states
7402 * The FsckPassNo= compatibility option in mount/service units
7403 has been removed. The fstab generator will now add the
7404 necessary dependencies automatically, and does not require
7405 PID1's support for that anymore.
7407 * journalctl gained a new switch, --list-boots, that lists
7408 recent boots with their times and boot IDs.
7410 * The various tools like systemctl, loginctl, timedatectl,
7411 busctl, systemd-run, ... have gained a new switch "-M" to
7412 connect to a specific, local OS container (as direct
7413 connection, without requiring SSH). This works on any
7414 container that is registered with machined, such as those
7415 created by libvirt-lxc or nspawn.
7417 * systemd-run and systemd-analyze also gained support for "-H"
7418 to connect to remote hosts via SSH. This is particularly
7419 useful for systemd-run because it enables queuing of jobs
7420 onto remote systems.
7422 * machinectl gained a new command "login" to open a getty
7423 login in any local container. This works with any container
7424 that is registered with machined (such as those created by
7425 libvirt-lxc or nspawn), and which runs systemd inside.
7427 * machinectl gained a new "reboot" command that may be used to
7428 trigger a reboot on a specific container that is registered
7429 with machined. This works on any container that runs an init
7430 system of some kind.
7432 * systemctl gained a new "list-timers" command to print a nice
7433 listing of installed timer units with the times they elapse
7436 * Alternative reboot() parameters may now be specified on the
7437 "systemctl reboot" command line and are passed to the
7438 reboot() system call.
7440 * systemctl gained a new --job-mode= switch to configure the
7441 mode to queue a job with. This is a more generic version of
7442 --fail, --irreversible, and --ignore-dependencies, which are
7443 still available but not advertised anymore.
7445 * /etc/systemd/system.conf gained new settings to configure
7446 various default timeouts of units, as well as the default
7447 start limit interval and burst. These may still be overridden
7450 * PID1 will now export on the bus profile data of the security
7451 policy upload process (such as the SELinux policy upload to
7454 * journald: when forwarding logs to the console, include
7455 timestamps (following the setting in
7456 /sys/module/printk/parameters/time).
7458 * OnCalendar= in timer units now understands the special
7459 strings "yearly" and "annually". (Both are equivalent)
7461 * The accuracy of timer units is now configurable with the new
7462 AccuracySec= setting. It defaults to 1min.
7464 * A new dependency type JoinsNamespaceOf= has been added that
7465 allows running two services within the same /tmp and network
7466 namespace, if PrivateNetwork= or PrivateTmp= are used.
7468 * A new command "cat" has been added to systemctl. It outputs
7469 the original unit file of a unit, and concatenates the
7470 contents of additional "drop-in" unit file snippets, so that
7471 the full configuration is shown.
7473 * systemctl now supports globbing on the various "list-xyz"
7474 commands, like "list-units" or "list-sockets", as well as on
7475 those commands which take multiple unit names.
7477 * journalctl's --unit= switch gained support for globbing.
7479 * All systemd daemons now make use of the watchdog logic so
7480 that systemd automatically notices when they hang.
7482 * If the $container_ttys environment variable is set,
7483 getty-generator will automatically spawn a getty for each
7484 listed tty. This is useful for container managers to request
7485 login gettys to be spawned on as many ttys as needed.
7487 * %h, %s, %U specifier support is not available anymore when
7488 used in unit files for PID 1. This is because NSS calls are
7489 not safe from PID 1. They stay available for --user
7490 instances of systemd, and as special case for the root user.
7492 * loginctl gained a new "--no-legend" switch to turn off output
7495 * The "sd-login.h" API gained three new calls:
7496 sd_session_is_remote(), sd_session_get_remote_user(),
7497 sd_session_get_remote_host() to query information about
7500 * The udev hardware database now also carries vendor/product
7501 information of SDIO devices.
7503 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
7504 determine whether watchdog notifications are requested by
7507 * Socket-activated per-connection services now include a
7508 short description of the connection parameters in the
7511 * tmpfiles gained a new "--boot" option. When this is not used,
7512 only lines where the command character is not suffixed with
7513 "!" are executed. When this option is specified, those
7514 options are executed too. This partitions tmpfiles
7515 directives into those that can be safely executed at any
7516 time, and those which should be run only at boot (for
7517 example, a line that creates /run/nologin).
7519 * A new API "sd-resolve.h" has been added which provides a simple
7520 asynchronous wrapper around glibc NSS host name resolution
7521 calls, such as getaddrinfo(). In contrast to glibc's
7522 getaddrinfo_a(), it does not use signals. In contrast to most
7523 other asynchronous name resolution libraries, this one does
7524 not reimplement DNS, but reuses NSS, so that alternate
7525 host name resolution systems continue to work, such as mDNS,
7526 LDAP, etc. This API is based on libasyncns, but it has been
7527 cleaned up for inclusion in systemd.
7529 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
7530 "sd-daemon.h" are no longer found in individual libraries
7531 libsystemd-journal.so, libsystemd-login.so,
7532 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
7533 merged them into a single library, libsystemd.so, which
7534 provides all symbols. The reason for this is cyclic
7535 dependencies, as these libraries tend to use each other's
7536 symbols. So far, we have managed to workaround that by linking
7537 a copy of a good part of our code into each of these
7538 libraries again and again, which, however, makes certain
7539 things hard to do, like sharing static variables. Also, it
7540 substantially increases footprint. With this change, there
7541 is only one library for the basic APIs systemd
7542 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
7543 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
7544 library as well, however are subject to the --enable-kdbus
7545 switch (see below). Note that "sd-dhcp-client.h" is not part
7546 of this library (this is because it only consumes, never
7547 provides, services of/to other APIs). To make the transition
7548 easy from the separate libraries to the unified one, we
7549 provide the --enable-compat-libs compile-time switch which
7550 will generate stub libraries that are compatible with the
7551 old ones but redirect all calls to the new one.
7553 * All of the kdbus logic and the new APIs "sd-bus.h",
7554 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
7555 and "sd-utf8.h" are compile-time optional via the
7556 "--enable-kdbus" switch, and they are not compiled in by
7557 default. To make use of kdbus, you have to explicitly enable
7558 the switch. Note however, that neither the kernel nor the
7559 userspace API for all of this is considered stable yet. We
7560 want to maintain the freedom to still change the APIs for
7561 now. By specifying this build-time switch, you acknowledge
7562 that you are aware of the instability of the current
7565 * Also, note that while kdbus is pretty much complete,
7566 it lacks one thing: proper policy support. This means you
7567 can build a fully working system with all features; however,
7568 it will be highly insecure. Policy support will be added in
7569 one of the next releases, at the same time that we will
7570 declare the APIs stable.
7572 * When the kernel command line argument "kdbus" is specified,
7573 systemd will automatically load the kdbus.ko kernel module. At
7574 this stage of development, it is only useful for testing kdbus
7575 and should not be used in production. Note: if "--enable-kdbus"
7576 is specified, and the kdbus.ko kernel module is available, and
7577 "kdbus" is added to the kernel command line, the entire system
7578 runs with kdbus instead of dbus-daemon, with the above mentioned
7579 problem of missing the system policy enforcement. Also a future
7580 version of kdbus.ko or a newer systemd will not be compatible with
7581 each other, and will unlikely be able to boot the machine if only
7582 one of them is updated.
7584 * systemctl gained a new "import-environment" command which
7585 uploads the caller's environment (or parts thereof) into the
7586 service manager so that it is inherited by services started
7587 by the manager. This is useful to upload variables like
7588 $DISPLAY into the user service manager.
7590 * A new PrivateDevices= switch has been added to service units
7591 which allows running a service with a namespaced /dev
7592 directory that does not contain any device nodes for
7593 physical devices. More specifically, it only includes devices
7594 such as /dev/null, /dev/urandom, and /dev/zero which are API
7597 * logind has been extended to support behaviour like VT
7598 switching on seats that do not support a VT. This makes
7599 multi-session available on seats that are not the first seat
7600 (seat0), and on systems where kernel support for VTs has
7601 been disabled at compile-time.
7603 * If a process holds a delay lock for system sleep or shutdown
7604 and fails to release it in time, we will now log its
7605 identity. This makes it easier to identify processes that
7606 cause slow suspends or power-offs.
7608 * When parsing /etc/crypttab, support for a new key-slot=
7609 option as supported by Debian is added. It allows indicating
7610 which LUKS slot to use on disk, speeding up key loading.
7612 * The sd_journal_sendv() API call has been checked and
7613 officially declared to be async-signal-safe so that it may
7614 be invoked from signal handlers for logging purposes.
7616 * Boot-time status output is now enabled automatically after a
7617 short timeout if boot does not progress, in order to give
7618 the user an indication what she or he is waiting for.
7620 * The boot-time output has been improved to show how much time
7621 remains until jobs expire.
7623 * The KillMode= switch in service units gained a new possible
7624 value "mixed". If set, and the unit is shut down, then the
7625 initial SIGTERM signal is sent only to the main daemon
7626 process, while the following SIGKILL signal is sent to
7627 all remaining processes of the service.
7629 * When a scope unit is registered, a new property "Controller"
7630 may be set. If set to a valid bus name, systemd will send a
7631 RequestStop() signal to this name when it would like to shut
7632 down the scope. This may be used to hook manager logic into
7633 the shutdown logic of scope units. Also, scope units may now
7634 be put in a special "abandoned" state, in which case the
7635 manager process which created them takes no further
7636 responsibilities for it.
7638 * When reading unit files, systemd will now verify
7639 the access mode of these files, and warn about certain
7640 suspicious combinations. This has been added to make it
7641 easier to track down packaging bugs where unit files are
7642 marked executable or world-writable.
7644 * systemd-nspawn gained a new "--setenv=" switch to set
7645 container-wide environment variables. The similar option in
7646 systemd-activate was renamed from "--environment=" to
7647 "--setenv=" for consistency.
7649 * systemd-nspawn has been updated to create a new kdbus domain
7650 for each container that is invoked, thus allowing each
7651 container to have its own set of system and user buses,
7652 independent of the host.
7654 * systemd-nspawn gained a new --drop-capability= switch to run
7655 the container with less capabilities than the default. Both
7656 --drop-capability= and --capability= now take the special
7657 string "all" for dropping or keeping all capabilities.
7659 * systemd-nspawn gained new switches for executing containers
7660 with specific SELinux labels set.
7662 * systemd-nspawn gained a new --quiet switch to not generate
7663 any additional output but the container's own console
7666 * systemd-nspawn gained a new --share-system switch to run a
7667 container without PID namespacing enabled.
7669 * systemd-nspawn gained a new --register= switch to control
7670 whether the container is registered with systemd-machined or
7671 not. This is useful for containers that do not run full
7672 OS images, but only specific apps.
7674 * systemd-nspawn gained a new --keep-unit which may be used
7675 when invoked as the only program from a service unit, and
7676 results in registration of the unit service itself in
7677 systemd-machined, instead of a newly opened scope unit.
7679 * systemd-nspawn gained a new --network-interface= switch for
7680 moving arbitrary interfaces to the container. The new
7681 --network-veth switch creates a virtual Ethernet connection
7682 between host and container. The new --network-bridge=
7683 switch then allows assigning the host side of this virtual
7684 Ethernet connection to a bridge device.
7686 * systemd-nspawn gained a new --personality= switch for
7687 setting the kernel personality for the container. This is
7688 useful when running a 32-bit container on a 64-bit host. A
7689 similar option Personality= is now also available for service
7692 * logind will now also track a "Desktop" identifier for each
7693 session which encodes the desktop environment of it. This is
7694 useful for desktop environments that want to identify
7695 multiple running sessions of itself easily.
7697 * A new SELinuxContext= setting for service units has been
7698 added that allows setting a specific SELinux execution
7699 context for a service.
7701 * Most systemd client tools will now honour $SYSTEMD_LESS for
7702 settings of the "less" pager. By default, these tools will
7703 override $LESS to allow certain operations to work, such as
7704 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
7705 influence this logic.
7707 * systemd's "seccomp" hook-up has been changed to make use of
7708 the libseccomp library instead of using its own
7709 implementation. This has benefits for portability among
7712 * For usage together with SystemCallFilter=, a new
7713 SystemCallErrorNumber= setting has been introduced that
7714 allows configuration of a system error number to be returned
7715 on filtered system calls, instead of immediately killing the
7716 process. Also, SystemCallArchitectures= has been added to
7717 limit access to system calls of a particular architecture
7718 (in order to turn off support for unused secondary
7719 architectures). There is also a global
7720 SystemCallArchitectures= setting in system.conf now to turn
7721 off support for non-native system calls system-wide.
7723 * systemd requires a kernel with a working name_to_handle_at(),
7724 please see the kernel config requirements in the README file.
7726 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
7727 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
7728 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
7729 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
7730 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
7731 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
7732 Elia Pinto, Florian Weimer, George McCollister, Goffredo
7733 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
7734 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
7735 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
7736 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
7737 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
7738 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
7739 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
7740 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
7741 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
7742 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
7743 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
7744 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
7745 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
7746 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
7747 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
7748 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
7749 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
7751 — Berlin, 2014-02-20
7755 * logind has gained support for facilitating privileged input
7756 and drm device access for unprivileged clients. This work is
7757 useful to allow Wayland display servers (and similar
7758 programs, such as kmscon) to run under the user's ID and
7759 access input and drm devices which are normally
7760 protected. When this is used (and the kernel is new enough)
7761 logind will "mute" IO on the file descriptors passed to
7762 Wayland as long as it is in the background and "unmute" it
7763 if it returns into the foreground. This allows secure
7764 session switching without allowing background sessions to
7765 eavesdrop on input and display data. This also introduces
7766 session switching support if VT support is turned off in the
7767 kernel, and on seats that are not seat0.
7769 * A new kernel command line option luks.options= is understood
7770 now which allows specifying LUKS options for usage for LUKS
7771 encrypted partitions specified with luks.uuid=.
7773 * tmpfiles.d(5) snippets may now use specifier expansion in
7774 path names. More specifically %m, %b, %H, %v, are now
7775 replaced by the local machine id, boot id, hostname, and
7776 kernel version number.
7778 * A new tmpfiles.d(5) command "m" has been introduced which
7779 may be used to change the owner/group/access mode of a file
7780 or directory if it exists, but do nothing if it does not.
7782 * This release removes high-level support for the
7783 MemorySoftLimit= cgroup setting. The underlying kernel
7784 cgroup attribute memory.soft_limit= is currently badly
7785 designed and likely to be removed from the kernel API in its
7786 current form, hence we should not expose it for now.
7788 * The memory.use_hierarchy cgroup attribute is now enabled for
7789 all cgroups systemd creates in the memory cgroup
7790 hierarchy. This option is likely to be come the built-in
7791 default in the kernel anyway, and the non-hierarchical mode
7792 never made much sense in the intrinsically hierarchical
7795 * A new field _SYSTEMD_SLICE= is logged along with all journal
7796 messages containing the slice a message was generated
7797 from. This is useful to allow easy per-customer filtering of
7798 logs among other things.
7800 * systemd-journald will no longer adjust the group of journal
7801 files it creates to the "systemd-journal" group. Instead we
7802 rely on the journal directory to be owned by the
7803 "systemd-journal" group, and its setgid bit set, so that the
7804 kernel file system layer will automatically enforce that
7805 journal files inherit this group assignment. The reason for
7806 this change is that we cannot allow NSS look-ups from
7807 journald which would be necessary to resolve
7808 "systemd-journal" to a numeric GID, because this might
7809 create deadlocks if NSS involves synchronous queries to
7810 other daemons (such as nscd, or sssd) which in turn are
7811 logging clients of journald and might block on it, which
7812 would then dead lock. A tmpfiles.d(5) snippet included in
7813 systemd will make sure the setgid bit and group are
7814 properly set on the journal directory if it exists on every
7815 boot. However, we recommend adjusting it manually after
7816 upgrades too (or from RPM scriptlets), so that the change is
7817 not delayed until next reboot.
7819 * Backlight and random seed files in /var/lib/ have moved into
7820 the /var/lib/systemd/ directory, in order to centralize all
7821 systemd generated files in one directory.
7823 * Boot time performance measurements (as displayed by
7824 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
7825 performance information if that's available to determine how
7826 much time BIOS and boot loader initialization required. With
7827 a sufficiently new BIOS you hence no longer need to boot
7828 with Gummiboot to get access to such information.
7830 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
7831 Cristian Rodríguez, Dave Reisner, David Herrmann, David
7832 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
7833 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
7834 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
7835 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
7836 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
7838 — Berlin, 2013-10-02
7842 * The Restart= option for services now understands a new
7843 on-watchdog setting, which will restart the service
7844 automatically if the service stops sending out watchdog keep
7845 alive messages (as configured with WatchdogSec=).
7847 * The getty generator (which is responsible for bringing up a
7848 getty on configured serial consoles) will no longer only
7849 start a getty on the primary kernel console but on all
7850 others, too. This makes the order in which console= is
7851 specified on the kernel command line less important.
7853 * libsystemd-logind gained a new sd_session_get_vt() call to
7854 retrieve the VT number of a session.
7856 * If the option "tries=0" is set for an entry of /etc/crypttab
7857 its passphrase is queried indefinitely instead of any
7858 maximum number of tries.
7860 * If a service with a configure PID file terminates its PID
7861 file will now be removed automatically if it still exists
7862 afterwards. This should put an end to stale PID files.
7864 * systemd-run will now also take relative binary path names
7865 for execution and no longer insists on absolute paths.
7867 * InaccessibleDirectories= and ReadOnlyDirectories= now take
7868 paths that are optionally prefixed with "-" to indicate that
7869 it should not be considered a failure if they do not exist.
7871 * journalctl -o (and similar commands) now understands a new
7872 output mode "short-precise", it is similar to "short" but
7873 shows timestamps with usec accuracy.
7875 * The option "discard" (as known from Debian) is now
7876 synonymous to "allow-discards" in /etc/crypttab. In fact,
7877 "discard" is preferred now (since it is easier to remember
7880 * Some licensing clean-ups were made, so that more code is now
7881 LGPL-2.1 licensed than before.
7883 * A minimal tool to save/restore the display backlight
7884 brightness across reboots has been added. It will store the
7885 backlight setting as late as possible at shutdown, and
7886 restore it as early as possible during reboot.
7888 * A logic to automatically discover and enable home and swap
7889 partitions on GPT disks has been added. With this in place
7890 /etc/fstab becomes optional for many setups as systemd can
7891 discover certain partitions located on the root disk
7892 automatically. Home partitions are recognized under their
7893 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
7894 partitions are recognized under their GPT type ID
7895 0657fd6da4ab43c484e50933c84b4f4f.
7897 * systemd will no longer pass any environment from the kernel
7898 or initrd to system services. If you want to set an
7899 environment for all services, do so via the kernel command
7900 line systemd.setenv= assignment.
7902 * The systemd-sysctl tool no longer natively reads the file
7903 /etc/sysctl.conf. If desired, the file should be symlinked
7904 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
7905 legacy support by a symlink rather than built-in code, it
7906 also makes the otherwise hidden order of application of the
7907 different files visible. (Note that this partly reverts to a
7908 pre-198 application order of sysctl knobs!)
7910 * The "systemctl set-log-level" and "systemctl dump" commands
7911 have been moved to systemd-analyze.
7913 * systemd-run learned the new --remain-after-exit switch,
7914 which causes the scope unit not to be cleaned up
7915 automatically after the process terminated.
7917 * tmpfiles learned a new --exclude-prefix= switch to exclude
7918 certain paths from operation.
7920 * journald will now automatically flush all messages to disk
7921 as soon as a message at the log level CRIT, ALERT or EMERG
7924 Contributions from: Andrew Cook, Brandon Philips, Christian
7925 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
7926 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
7927 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
7928 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
7929 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
7930 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
7931 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
7932 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
7933 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
7934 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
7935 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
7936 William Giokas, Zbigniew Jędrzejewski-Szmek
7938 — Berlin, 2013-09-13
7942 * The documentation has been updated to cover the various new
7943 concepts introduced with 205.
7945 * Unit files now understand the new %v specifier which
7946 resolves to the kernel version string as returned by "uname
7949 * systemctl now supports filtering the unit list output by
7950 load state, active state and sub state, using the new
7953 * "systemctl status" will now show the results of the
7954 condition checks (like ConditionPathExists= and similar) of
7955 the last start attempts of the unit. They are also logged to
7958 * "journalctl -b" may now be used to look for boot output of a
7959 specific boot. Try "journalctl -b -1" for the previous boot,
7960 but the syntax is substantially more powerful.
7962 * "journalctl --show-cursor" has been added which prints the
7963 cursor string the last shown log line. This may then be used
7964 with the new "journalctl --after-cursor=" switch to continue
7965 browsing logs from that point on.
7967 * "journalctl --force" may now be used to force regeneration
7970 * Creation of "dead" device nodes has been moved from udev
7971 into kmod and tmpfiles. Previously, udev would read the kmod
7972 databases to pre-generate dead device nodes based on meta
7973 information contained in kernel modules, so that these would
7974 be auto-loaded on access rather then at boot. As this
7975 does not really have much to do with the exposing actual
7976 kernel devices to userspace this has always been slightly
7977 alien in the udev codebase. Following the new scheme kmod
7978 will now generate a runtime snippet for tmpfiles from the
7979 module meta information and it now is tmpfiles' job to the
7980 create the nodes. This also allows overriding access and
7981 other parameters for the nodes using the usual tmpfiles
7982 facilities. As side effect this allows us to remove the
7983 CAP_SYS_MKNOD capability bit from udevd entirely.
7985 * logind's device ACLs may now be applied to these "dead"
7986 devices nodes too, thus finally allowing managed access to
7987 devices such as /dev/snd/sequencer without loading the
7988 backing module right-away.
7990 * A new RPM macro has been added that may be used to apply
7991 tmpfiles configuration during package installation.
7993 * systemd-detect-virt and ConditionVirtualization= now can
7994 detect User-Mode-Linux machines (UML).
7996 * journald will now implicitly log the effective capabilities
7997 set of processes in the message metadata.
7999 * systemd-cryptsetup has gained support for TrueCrypt volumes.
8001 * The initrd interface has been simplified (more specifically,
8002 support for passing performance data via environment
8003 variables and fsck results via files in /run has been
8004 removed). These features were non-essential, and are
8005 nowadays available in a much nicer way by having systemd in
8006 the initrd serialize its state and have the hosts systemd
8007 deserialize it again.
8009 * The udev "keymap" data files and tools to apply keyboard
8010 specific mappings of scan to key codes, and force-release
8011 scan code lists have been entirely replaced by a udev
8012 "keyboard" builtin and a hwdb data file.
8014 * systemd will now honour the kernel's "quiet" command line
8015 argument also during late shutdown, resulting in a
8016 completely silent shutdown when used.
8018 * There's now an option to control the SO_REUSEPORT socket
8019 option in .socket units.
8021 * Instance units will now automatically get a per-template
8022 subslice of system.slice unless something else is explicitly
8023 configured. For example, instances of sshd@.service will now
8024 implicitly be placed in system-sshd.slice rather than
8025 system.slice as before.
8027 * Test coverage support may now be enabled at build time.
8029 Contributions from: Dave Reisner, Frederic Crozat, Harald
8030 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
8031 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
8032 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
8033 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
8034 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
8035 Giokas, Zbigniew Jędrzejewski-Szmek
8037 — Berlin, 2013-07-23
8041 * Two new unit types have been introduced:
8043 Scope units are very similar to service units, however, are
8044 created out of pre-existing processes — instead of PID 1
8045 forking off the processes. By using scope units it is
8046 possible for system services and applications to group their
8047 own child processes (worker processes) in a powerful way
8048 which then maybe used to organize them, or kill them
8049 together, or apply resource limits on them.
8051 Slice units may be used to partition system resources in an
8052 hierarchical fashion and then assign other units to them. By
8053 default there are now three slices: system.slice (for all
8054 system services), user.slice (for all user sessions),
8055 machine.slice (for VMs and containers).
8057 Slices and scopes have been introduced primarily in
8058 context of the work to move cgroup handling to a
8059 single-writer scheme, where only PID 1
8060 creates/removes/manages cgroups.
8062 * There's a new concept of "transient" units. In contrast to
8063 normal units these units are created via an API at runtime,
8064 not from configuration from disk. More specifically this
8065 means it is now possible to run arbitrary programs as
8066 independent services, with all execution parameters passed
8067 in via bus APIs rather than read from disk. Transient units
8068 make systemd substantially more dynamic then it ever was,
8069 and useful as a general batch manager.
8071 * logind has been updated to make use of scope and slice units
8072 for managing user sessions. As a user logs in he will get
8073 his own private slice unit, to which all sessions are added
8074 as scope units. We also added support for automatically
8075 adding an instance of user@.service for the user into the
8076 slice. Effectively logind will no longer create cgroup
8077 hierarchies on its own now, it will defer entirely to PID 1
8078 for this by means of scope, service and slice units. Since
8079 user sessions this way become entities managed by PID 1
8080 the output of "systemctl" is now a lot more comprehensive.
8082 * A new mini-daemon "systemd-machined" has been added which
8083 may be used by virtualization managers to register local
8084 VMs/containers. nspawn has been updated accordingly, and
8085 libvirt will be updated shortly. machined will collect a bit
8086 of meta information about the VMs/containers, and assign
8087 them their own scope unit (see above). The collected
8088 meta-data is then made available via the "machinectl" tool,
8089 and exposed in "ps" and similar tools. machined/machinectl
8090 is compile-time optional.
8092 * As discussed earlier, the low-level cgroup configuration
8093 options ControlGroup=, ControlGroupModify=,
8094 ControlGroupPersistent=, ControlGroupAttribute= have been
8095 removed. Please use high-level attribute settings instead as
8096 well as slice units.
8098 * A new bus call SetUnitProperties() has been added to alter
8099 various runtime parameters of a unit. This is primarily
8100 useful to alter cgroup parameters dynamically in a nice way,
8101 but will be extended later on to make more properties
8102 modifiable at runtime. systemctl gained a new set-properties
8103 command that wraps this call.
8105 * A new tool "systemd-run" has been added which can be used to
8106 run arbitrary command lines as transient services or scopes,
8107 while configuring a number of settings via the command
8108 line. This tool is currently very basic, however already
8109 very useful. We plan to extend this tool to even allow
8110 queuing of execution jobs with time triggers from the
8111 command line, similar in fashion to "at".
8113 * nspawn will now inform the user explicitly that kernels with
8114 audit enabled break containers, and suggest the user to turn
8117 * Support for detecting the IMA and AppArmor security
8118 frameworks with ConditionSecurity= has been added.
8120 * journalctl gained a new "-k" switch for showing only kernel
8121 messages, mimicking dmesg output; in addition to "--user"
8122 and "--system" switches for showing only user's own logs
8125 * systemd-delta can now show information about drop-in
8126 snippets extending unit files.
8128 * libsystemd-bus has been substantially updated but is still
8129 not available as public API.
8131 * systemd will now look for the "debug" argument on the kernel
8132 command line and enable debug logging, similar to what
8133 "systemd.log_level=debug" already did before.
8135 * "systemctl set-default", "systemctl get-default" has been
8136 added to configure the default.target symlink, which
8137 controls what to boot into by default.
8139 * "systemctl set-log-level" has been added as a convenient
8140 way to raise and lower systemd logging threshold.
8142 * "systemd-analyze plot" will now show the time the various
8143 generators needed for execution, as well as information
8144 about the unit file loading.
8146 * libsystemd-journal gained a new sd_journal_open_files() call
8147 for opening specific journal files. journactl also gained a
8148 new switch to expose this new functionality. Previously we
8149 only supported opening all files from a directory, or all
8150 files from the system, as opening individual files only is
8151 racy due to journal file rotation.
8153 * systemd gained the new DefaultEnvironment= setting in
8154 /etc/systemd/system.conf to set environment variables for
8157 * If a privileged process logs a journal message with the
8158 OBJECT_PID= field set, then journald will automatically
8159 augment this with additional OBJECT_UID=, OBJECT_GID=,
8160 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
8161 system services want to log events about specific client
8162 processes. journactl/systemctl has been updated to make use
8163 of this information if all log messages regarding a specific
8166 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
8167 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
8168 Reisner, David Coppa, David King, David Strauss, Eelco
8169 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
8170 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
8171 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
8172 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
8173 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
8174 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
8175 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
8176 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
8177 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
8178 Łukasz Stelmach, 장동준
8182 * The Python bindings gained some minimal support for the APIs
8183 exposed by libsystemd-logind.
8185 * ConditionSecurity= gained support for detecting SMACK. Since
8186 this condition already supports SELinux and AppArmor we only
8187 miss IMA for this. Patches welcome!
8189 Contributions from: Karol Lewandowski, Lennart Poettering,
8190 Zbigniew Jędrzejewski-Szmek
8194 * systemd-nspawn will now create /etc/resolv.conf if
8195 necessary, before bind-mounting the host's file onto it.
8197 * systemd-nspawn will now store meta information about a
8198 container on the container's cgroup as extended attribute
8199 fields, including the root directory.
8201 * The cgroup hierarchy has been reworked in many ways. All
8202 objects any of the components systemd creates in the cgroup
8203 tree are now suffixed. More specifically, user sessions are
8204 now placed in cgroups suffixed with ".session", users in
8205 cgroups suffixed with ".user", and nspawn containers in
8206 cgroups suffixed with ".nspawn". Furthermore, all cgroup
8207 names are now escaped in a simple scheme to avoid collision
8208 of userspace object names with kernel filenames. This work
8209 is preparation for making these objects relocatable in the
8210 cgroup tree, in order to allow easy resource partitioning of
8211 these objects without causing naming conflicts.
8213 * systemctl list-dependencies gained the new switches
8214 --plain, --reverse, --after and --before.
8216 * systemd-inhibit now shows the process name of processes that
8217 have taken an inhibitor lock.
8219 * nss-myhostname will now also resolve "localhost"
8220 implicitly. This makes /etc/hosts an optional file and
8221 nicely handles that on IPv6 ::1 maps to both "localhost" and
8224 * libsystemd-logind.so gained a new call
8225 sd_get_machine_names() to enumerate running containers and
8226 VMs (currently only supported by very new libvirt and
8227 nspawn). sd_login_monitor can now be used to watch
8228 VMs/containers coming and going.
8230 * .include is not allowed recursively anymore, and only in
8231 unit files. Usually it is better to use drop-in snippets in
8232 .d/*.conf anyway, as introduced with systemd 198.
8234 * systemd-analyze gained a new "critical-chain" command that
8235 determines the slowest chain of units run during system
8236 boot-up. It is very useful for tracking down where
8237 optimizing boot time is the most beneficial.
8239 * systemd will no longer allow manipulating service paths in
8240 the name=systemd:/system cgroup tree using ControlGroup= in
8241 units. (But is still fine with it in all other dirs.)
8243 * There's a new systemd-nspawn@.service service file that may
8244 be used to easily run nspawn containers as system
8245 services. With the container's root directory in
8246 /var/lib/container/foobar it is now sufficient to run
8247 "systemctl start systemd-nspawn@foobar.service" to boot it.
8249 * systemd-cgls gained a new parameter "--machine" to list only
8250 the processes within a certain container.
8252 * ConditionSecurity= now can check for "apparmor". We still
8253 are lacking checks for SMACK and IMA for this condition
8254 check though. Patches welcome!
8256 * A new configuration file /etc/systemd/sleep.conf has been
8257 added that may be used to configure which kernel operation
8258 systemd is supposed to execute when "suspend", "hibernate"
8259 or "hybrid-sleep" is requested. This makes the new kernel
8260 "freeze" state accessible to the user.
8262 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
8263 the passed argument if applicable.
8265 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
8266 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
8267 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
8268 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
8269 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
8270 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
8271 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
8276 * The output of 'systemctl list-jobs' got some polishing. The
8277 '--type=' argument may now be passed more than once. A new
8278 command 'systemctl list-sockets' has been added which shows
8279 a list of kernel sockets systemd is listening on with the
8280 socket units they belong to, plus the units these socket
8283 * The experimental libsystemd-bus library got substantial
8284 updates to work in conjunction with the (also experimental)
8285 kdbus kernel project. It works well enough to exchange
8286 messages with some sophistication. Note that kdbus is not
8287 ready yet, and the library is mostly an elaborate test case
8288 for now, and not installable.
8290 * systemd gained a new unit 'systemd-static-nodes.service'
8291 that generates static device nodes earlier during boot, and
8292 can run in conjunction with udev.
8294 * libsystemd-login gained a new call sd_pid_get_user_unit()
8295 to retrieve the user systemd unit a process is running
8296 in. This is useful for systems where systemd is used as
8299 * systemd-nspawn now places all containers in the new /machine
8300 top-level cgroup directory in the name=systemd
8301 hierarchy. libvirt will soon do the same, so that we get a
8302 uniform separation of /system, /user and /machine for system
8303 services, user processes and containers/virtual
8304 machines. This new cgroup hierarchy is also useful to stick
8305 stable names to specific container instances, which can be
8306 recognized later this way (this name may be controlled
8307 via systemd-nspawn's new -M switch). libsystemd-login also
8308 gained a new call sd_pid_get_machine_name() to retrieve the
8309 name of the container/VM a specific process belongs to.
8311 * bootchart can now store its data in the journal.
8313 * libsystemd-journal gained a new call
8314 sd_journal_add_conjunction() for AND expressions to the
8315 matching logic. This can be used to express more complex
8316 logical expressions.
8318 * journactl can now take multiple --unit= and --user-unit=
8321 * The cryptsetup logic now understands the "luks.key=" kernel
8322 command line switch for specifying a file to read the
8323 decryption key from. Also, if a configured key file is not
8324 found the tool will now automatically fall back to prompting
8327 * Python systemd.journal module was updated to wrap recently
8328 added functions from libsystemd-journal. The interface was
8329 changed to bring the low level interface in s.j._Reader
8330 closer to the C API, and the high level interface in
8331 s.j.Reader was updated to wrap and convert all data about
8334 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
8335 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
8336 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
8337 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
8338 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
8339 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
8343 * journalctl --update-catalog now understands a new --root=
8344 option to operate on catalogs found in a different root
8347 * During shutdown after systemd has terminated all running
8348 services a final killing loop kills all remaining left-over
8349 processes. We will now print the name of these processes
8350 when we send SIGKILL to them, since this usually indicates a
8353 * If /etc/crypttab refers to password files stored on
8354 configured mount points automatic dependencies will now be
8355 generated to ensure the specific mount is established first
8356 before the key file is attempted to be read.
8358 * 'systemctl status' will now show information about the
8359 network sockets a socket unit is listening on.
8361 * 'systemctl status' will also shown information about any
8362 drop-in configuration file for units. (Drop-In configuration
8363 files in this context are files such as
8364 /etc/systemd/systemd/foobar.service.d/*.conf)
8366 * systemd-cgtop now optionally shows summed up CPU times of
8367 cgroups. Press '%' while running cgtop to switch between
8368 percentage and absolute mode. This is useful to determine
8369 which cgroups use up the most CPU time over the entire
8370 runtime of the system. systemd-cgtop has also been updated
8371 to be 'pipeable' for processing with further shell tools.
8373 * 'hostnamectl set-hostname' will now allow setting of FQDN
8376 * The formatting and parsing of time span values has been
8377 changed. The parser now understands fractional expressions
8378 such as "5.5h". The formatter will now output fractional
8379 expressions for all time spans under 1min, i.e. "5.123456s"
8380 rather than "5s 123ms 456us". For time spans under 1s
8381 millisecond values are shown, for those under 1ms
8382 microsecond values are shown. This should greatly improve
8383 all time-related output of systemd.
8385 * libsystemd-login and libsystemd-journal gained new
8386 functions for querying the poll() events mask and poll()
8387 timeout value for integration into arbitrary event
8390 * localectl gained the ability to list available X11 keymaps
8391 (models, layouts, variants, options).
8393 * 'systemd-analyze dot' gained the ability to filter for
8394 specific units via shell-style globs, to create smaller,
8395 more useful graphs. I.e. it is now possible to create simple
8396 graphs of all the dependencies between only target units, or
8397 of all units that Avahi has dependencies with.
8399 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
8400 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
8401 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
8402 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
8403 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
8404 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
8405 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
8409 * The boot-time readahead implementation for rotating media
8410 will now read the read-ahead data in multiple passes which
8411 consist of all read requests made in equidistant time
8412 intervals. This means instead of strictly reading read-ahead
8413 data in its physical order on disk we now try to find a
8414 middle ground between physical and access time order.
8416 * /etc/os-release files gained a new BUILD_ID= field for usage
8417 on operating systems that provide continuous builds of OS
8420 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
8421 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
8422 William Douglas, Zbigniew Jędrzejewski-Szmek
8426 * systemd-python gained an API exposing libsystemd-daemon.
8428 * The SMACK setup logic gained support for uploading CIPSO
8431 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
8432 ReadOnlyDirectories= and InaccessibleDirectories= has
8433 changed. The private /tmp and /var/tmp directories are now
8434 shared by all processes of a service (which means
8435 ExecStartPre= may now leave data in /tmp that ExecStart= of
8436 the same service can still access). When a service is
8437 stopped its temporary directories are immediately deleted
8438 (normal clean-up with tmpfiles is still done in addition to
8441 * By default, systemd will now set a couple of sysctl
8442 variables in the kernel: the safe sysrq options are turned
8443 on, IP route verification is turned on, and source routing
8444 disabled. The recently added hardlink and softlink
8445 protection of the kernel is turned on. These settings should
8446 be reasonably safe, and good defaults for all new systems.
8448 * The predictable network naming logic may now be turned off
8449 with a new kernel command line switch: net.ifnames=0.
8451 * A new libsystemd-bus module has been added that implements a
8452 pretty complete D-Bus client library. For details see:
8454 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
8456 * journald will now explicitly flush the journal files to disk
8457 at the latest 5min after each write. The file will then also
8458 be marked offline until the next write. This should increase
8459 reliability in case of a crash. The synchronization delay
8460 can be configured via SyncIntervalSec= in journald.conf.
8462 * There's a new remote-fs-setup.target unit that can be used
8463 to pull in specific services when at least one remote file
8464 system is to be mounted.
8466 * There are new targets timers.target and paths.target as
8467 canonical targets to pull user timer and path units in
8468 from. This complements sockets.target with a similar
8469 purpose for socket units.
8471 * libudev gained a new call udev_device_set_attribute_value()
8472 to set sysfs attributes of a device.
8474 * The udev daemon now sets the default number of worker
8475 processes executed in parallel based on the number of available
8476 CPUs instead of the amount of available RAM. This is supposed
8477 to provide a more reliable default and limit a too aggressive
8478 parallelism for setups with 1000s of devices connected.
8480 Contributions from: Auke Kok, Colin Walters, Cristian
8481 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
8482 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
8483 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
8484 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
8485 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
8486 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
8487 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
8488 Zbigniew Jędrzejewski-Szmek
8492 * Configuration of unit files may now be extended via drop-in
8493 files without having to edit/override the unit files
8494 themselves. More specifically, if the administrator wants to
8495 change one value for a service file foobar.service he can
8496 now do so by dropping in a configuration snippet into
8497 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
8498 will load all these snippets and apply them on top of the
8499 main unit configuration file, possibly extending or
8500 overriding its settings. Using these drop-in snippets is
8501 generally nicer than the two earlier options for changing
8502 unit files locally: copying the files from
8503 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
8504 them there; or creating a new file in /etc/systemd/system/
8505 that incorporates the original one via ".include". Drop-in
8506 snippets into these .d/ directories can be placed in any
8507 directory systemd looks for units in, and the usual
8508 overriding semantics between /usr/lib, /etc and /run apply
8511 * Most unit file settings which take lists of items can now be
8512 reset by assigning the empty string to them. For example,
8513 normally, settings such as Environment=FOO=BAR append a new
8514 environment variable assignment to the environment block,
8515 each time they are used. By assigning Environment= the empty
8516 string the environment block can be reset to empty. This is
8517 particularly useful with the .d/*.conf drop-in snippets
8518 mentioned above, since this adds the ability to reset list
8519 settings from vendor unit files via these drop-ins.
8521 * systemctl gained a new "list-dependencies" command for
8522 listing the dependencies of a unit recursively.
8524 * Inhibitors are now honored and listed by "systemctl
8525 suspend", "systemctl poweroff" (and similar) too, not only
8526 GNOME. These commands will also list active sessions by
8529 * Resource limits (as exposed by the various control group
8530 controllers) can now be controlled dynamically at runtime
8531 for all units. More specifically, you can now use a command
8532 like "systemctl set-cgroup-attr foobar.service cpu.shares
8533 2000" to alter the CPU shares a specific service gets. These
8534 settings are stored persistently on disk, and thus allow the
8535 administrator to easily adjust the resource usage of
8536 services with a few simple commands. This dynamic resource
8537 management logic is also available to other programs via the
8538 bus. Almost any kernel cgroup attribute and controller is
8541 * systemd-vconsole-setup will now copy all font settings to
8542 all allocated VTs, where it previously applied them only to
8545 * libsystemd-login gained the new sd_session_get_tty() API
8548 * This release drops support for a few legacy or
8549 distribution-specific LSB facility names when parsing init
8550 scripts: $x-display-manager, $mail-transfer-agent,
8551 $mail-transport-agent, $mail-transfer-agent, $smtp,
8552 $null. Also, the mail-transfer-agent.target unit backing
8553 this has been removed. Distributions which want to retain
8554 compatibility with this should carry the burden for
8555 supporting this themselves and patch support for these back
8556 in, if they really need to. Also, the facilities $syslog and
8557 $local_fs are now ignored, since systemd does not support
8558 early-boot LSB init scripts anymore, and these facilities
8559 are implied anyway for normal services. syslog.target has
8562 * There are new bus calls on PID1's Manager object for
8563 cancelling jobs, and removing snapshot units. Previously,
8564 both calls were only available on the Job and Snapshot
8567 * systemd-journal-gatewayd gained SSL support.
8569 * The various "environment" files, such as /etc/locale.conf
8570 now support continuation lines with a backslash ("\") as
8571 last character in the line, similarly in style (but different)
8572 to how this is supported in shells.
8574 * For normal user processes the _SYSTEMD_USER_UNIT= field is
8575 now implicitly appended to every log entry logged. systemctl
8576 has been updated to filter by this field when operating on a
8577 user systemd instance.
8579 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
8580 CAP_AUDIT_CONTROL capabilities to the capabilities set for
8581 the container. This makes it easier to boot unmodified
8582 Fedora systems in a container, which however still requires
8583 audit=0 to be passed on the kernel command line. Auditing in
8584 kernel and userspace is unfortunately still too broken in
8585 context of containers, hence we recommend compiling it out
8586 of the kernel or using audit=0. Hopefully this will be fixed
8587 one day for good in the kernel.
8589 * nspawn gained the new --bind= and --bind-ro= parameters to
8590 bind mount specific directories from the host into the
8593 * nspawn will now mount its own devpts file system instance
8594 into the container, in order not to leak pty devices from
8595 the host into the container.
8597 * systemd will now read the firmware boot time performance
8598 information from the EFI variables, if the used boot loader
8599 supports this, and takes it into account for boot performance
8600 analysis via "systemd-analyze". This is currently supported
8601 only in conjunction with Gummiboot, but could be supported
8602 by other boot loaders too. For details see:
8604 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
8606 * A new generator has been added that automatically mounts the
8607 EFI System Partition (ESP) to /boot, if that directory
8608 exists, is empty, and no other file system has been
8609 configured to be mounted there.
8611 * logind will now send out PrepareForSleep(false) out
8612 unconditionally, after coming back from suspend. This may be
8613 used by applications as asynchronous notification for
8614 system resume events.
8616 * "systemctl unlock-sessions" has been added, that allows
8617 unlocking the screens of all user sessions at once, similar
8618 to how "systemctl lock-sessions" already locked all users
8619 sessions. This is backed by a new D-Bus call UnlockSessions().
8621 * "loginctl seat-status" will now show the master device of a
8622 seat. (i.e. the device of a seat that needs to be around for
8623 the seat to be considered available, usually the graphics
8626 * tmpfiles gained a new "X" line type, that allows
8627 configuration of files and directories (with wildcards) that
8628 shall be excluded from automatic cleanup ("aging").
8630 * udev default rules set the device node permissions now only
8631 at "add" events, and do not change them any longer with a
8632 later "change" event.
8634 * The log messages for lid events and power/sleep keypresses
8635 now carry a message ID.
8637 * We now have a substantially larger unit test suite, but this
8638 continues to be work in progress.
8640 * udevadm hwdb gained a new --root= parameter to change the
8641 root directory to operate relative to.
8643 * logind will now issue a background sync() request to the kernel
8644 early at shutdown, so that dirty buffers are flushed to disk early
8645 instead of at the last moment, in order to optimize shutdown
8648 * A new bootctl tool has been added that is an interface for
8649 certain boot loader operations. This is currently a preview
8650 and is likely to be extended into a small mechanism daemon
8651 like timedated, localed, hostnamed, and can be used by
8652 graphical UIs to enumerate available boot options, and
8653 request boot into firmware operations.
8655 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
8656 the rest of the package. It also has been updated to work
8657 correctly in initrds.
8659 * polkit previously has been runtime optional, and is now also
8660 compile time optional via a configure switch.
8662 * systemd-analyze has been reimplemented in C. Also "systemctl
8663 dot" has moved into systemd-analyze.
8665 * "systemctl status" with no further parameters will now print
8666 the status of all active or failed units.
8668 * Operations such as "systemctl start" can now be executed
8669 with a new mode "--irreversible" which may be used to queue
8670 operations that cannot accidentally be reversed by a later
8671 job queuing. This is by default used to make shutdown
8672 requests more robust.
8674 * The Python API of systemd now gained a new module for
8675 reading journal files.
8677 * A new tool kernel-install has been added that can install
8678 kernel images according to the Boot Loader Specification:
8680 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
8682 * Boot time console output has been improved to provide
8683 animated boot time output for hanging jobs.
8685 * A new tool systemd-activate has been added which can be used
8686 to test socket activation with, directly from the command
8687 line. This should make it much easier to test and debug
8688 socket activation in daemons.
8690 * journalctl gained a new "--reverse" (or -r) option to show
8691 journal output in reverse order (i.e. newest line first).
8693 * journalctl gained a new "--pager-end" (or -e) option to jump
8694 to immediately jump to the end of the journal in the
8695 pager. This is only supported in conjunction with "less".
8697 * journalctl gained a new "--user-unit=" option, that works
8698 similarly to "--unit=" but filters for user units rather than
8701 * A number of unit files to ease adoption of systemd in
8702 initrds has been added. This moves some minimal logic from
8703 the various initrd implementations into systemd proper.
8705 * The journal files are now owned by a new group
8706 "systemd-journal", which exists specifically to allow access
8707 to the journal, and nothing else. Previously, we used the
8708 "adm" group for that, which however possibly covers more
8709 than just journal/log file access. This new group is now
8710 already used by systemd-journal-gatewayd to ensure this
8711 daemon gets access to the journal files and as little else
8712 as possible. Note that "make install" will also set FS ACLs
8713 up for /var/log/journal to give "adm" and "wheel" read
8714 access to it, in addition to "systemd-journal" which owns
8715 the journal files. We recommend that packaging scripts also
8716 add read access to "adm" + "wheel" to /var/log/journal, and
8717 all existing/future journal files. To normal users and
8718 administrators little changes, however packagers need to
8719 ensure to create the "systemd-journal" system group at
8720 package installation time.
8722 * The systemd-journal-gatewayd now runs as unprivileged user
8723 systemd-journal-gateway:systemd-journal-gateway. Packaging
8724 scripts need to create these system user/group at
8727 * timedated now exposes a new boolean property CanNTP that
8728 indicates whether a local NTP service is available or not.
8730 * systemd-detect-virt will now also detect xen PVs
8732 * The pstore file system is now mounted by default, if it is
8735 * In addition to the SELinux and IMA policies we will now also
8736 load SMACK policies at early boot.
8738 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
8739 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
8740 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
8741 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
8742 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
8743 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
8744 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
8745 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
8746 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
8747 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
8748 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
8749 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
8750 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
8751 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
8755 * Timer units now support calendar time events in addition to
8756 monotonic time events. That means you can now trigger a unit
8757 based on a calendar time specification such as "Thu,Fri
8758 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
8759 or fifth day of any month of the year 2013, given that it is
8760 a thursday or friday. This brings timer event support
8761 considerably closer to cron's capabilities. For details on
8762 the supported calendar time specification language see
8765 * udev now supports a number of different naming policies for
8766 network interfaces for predictable names, and a combination
8767 of these policies is now the default. Please see this wiki
8768 document for details:
8770 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
8772 * Auke Kok's bootchart implementation has been added to the
8773 systemd tree. It is an optional component that can graph the
8774 boot in quite some detail. It is one of the best bootchart
8775 implementations around and minimal in its code and
8778 * nss-myhostname has been integrated into the systemd source
8779 tree. nss-myhostname guarantees that the local hostname
8780 always stays resolvable via NSS. It has been a weak
8781 requirement of systemd-hostnamed since a long time, and
8782 since its code is actually trivial we decided to just
8783 include it in systemd's source tree. It can be turned off
8784 with a configure switch.
8786 * The read-ahead logic is now capable of properly detecting
8787 whether a btrfs file system is on SSD or rotating media, in
8788 order to optimize the read-ahead scheme. Previously, it was
8789 only capable of detecting this on traditional file systems
8792 * In udev, additional device properties are now read from the
8793 IAB in addition to the OUI database. Also, Bluetooth company
8794 identities are attached to the devices as well.
8796 * In service files %U may be used as specifier that is
8797 replaced by the configured user name of the service.
8799 * nspawn may now be invoked without a controlling TTY. This
8800 makes it suitable for invocation as its own service. This
8801 may be used to set up a simple containerized server system
8802 using only core OS tools.
8804 * systemd and nspawn can now accept socket file descriptors
8805 when they are started for socket activation. This enables
8806 implementation of socket activated nspawn
8807 containers. i.e. think about autospawning an entire OS image
8808 when the first SSH or HTTP connection is received. We expect
8809 that similar functionality will also be added to libvirt-lxc
8812 * journalctl will now suppress ANSI color codes when
8813 presenting log data.
8815 * systemctl will no longer show control group information for
8816 a unit if the control group is empty anyway.
8818 * logind can now automatically suspend/hibernate/shutdown the
8821 * /etc/machine-info and hostnamed now also expose the chassis
8822 type of the system. This can be used to determine whether
8823 the local system is a laptop, desktop, handset or
8824 tablet. This information may either be configured by the
8825 user/vendor or is automatically determined from ACPI and DMI
8826 information if possible.
8828 * A number of polkit actions are now bound together with "imply"
8829 rules. This should simplify creating UIs because many actions
8830 will now authenticate similar ones as well.
8832 * Unit files learnt a new condition ConditionACPower= which
8833 may be used to conditionalize a unit depending on whether an
8834 AC power source is connected or not, of whether the system
8835 is running on battery power.
8837 * systemctl gained a new "is-failed" verb that may be used in
8838 shell scripts and suchlike to check whether a specific unit
8839 is in the "failed" state.
8841 * The EnvironmentFile= setting in unit files now supports file
8842 globbing, and can hence be used to easily read a number of
8843 environment files at once.
8845 * systemd will no longer detect and recognize specific
8846 distributions. All distribution-specific #ifdeffery has been
8847 removed, systemd is now fully generic and
8848 distribution-agnostic. Effectively, not too much is lost as
8849 a lot of the code is still accessible via explicit configure
8850 switches. However, support for some distribution specific
8851 legacy configuration file formats has been dropped. We
8852 recommend distributions to simply adopt the configuration
8853 files everybody else uses now and convert the old
8854 configuration from packaging scripts. Most distributions
8855 already did that. If that's not possible or desirable,
8856 distributions are welcome to forward port the specific
8857 pieces of code locally from the git history.
8859 * When logging a message about a unit systemd will now always
8860 log the unit name in the message meta data.
8862 * localectl will now also discover system locale data that is
8863 not stored in locale archives, but directly unpacked.
8865 * logind will no longer unconditionally use framebuffer
8866 devices as seat masters, i.e. as devices that are required
8867 to be existing before a seat is considered preset. Instead,
8868 it will now look for all devices that are tagged as
8869 "seat-master" in udev. By default, framebuffer devices will
8870 be marked as such, but depending on local systems, other
8871 devices might be marked as well. This may be used to
8872 integrate graphics cards using closed source drivers (such
8873 as NVidia ones) more nicely into logind. Note however, that
8874 we recommend using the open source NVidia drivers instead,
8875 and no udev rules for the closed-source drivers will be
8876 shipped from us upstream.
8878 Contributions from: Adam Williamson, Alessandro Crismani, Auke
8879 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
8880 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
8881 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
8882 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
8883 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
8884 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
8885 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
8886 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
8887 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
8888 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
8889 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
8894 * udev gained support for loading additional device properties
8895 from an indexed database that is keyed by vendor/product IDs
8896 and similar device identifiers. For the beginning this
8897 "hwdb" is populated with data from the well-known PCI and
8898 USB database, but also includes PNP, ACPI and OID data. In
8899 the longer run this indexed database shall grow into
8900 becoming the one central database for non-essential
8901 userspace device metadata. Previously, data from the PCI/USB
8902 database was only attached to select devices, since the
8903 lookup was a relatively expensive operation due to O(n) time
8904 complexity (with n being the number of entries in the
8905 database). Since this is now O(1), we decided to add in this
8906 data for all devices where this is available, by
8907 default. Note that the indexed database needs to be rebuilt
8908 when new data files are installed. To achieve this you need
8909 to update your packaging scripts to invoke "udevadm hwdb
8910 --update" after installation of hwdb data files. For
8911 RPM-based distributions we introduced the new
8912 %udev_hwdb_update macro for this purpose.
8914 * The Journal gained support for the "Message Catalog", an
8915 indexed database to link up additional information with
8916 journal entries. For further details please check:
8918 https://www.freedesktop.org/wiki/Software/systemd/catalog
8920 The indexed message catalog database also needs to be
8921 rebuilt after installation of message catalog files. Use
8922 "journalctl --update-catalog" for this. For RPM-based
8923 distributions we introduced the %journal_catalog_update
8924 macro for this purpose.
8926 * The Python Journal bindings gained support for the standard
8927 Python logging framework.
8929 * The Journal API gained new functions for checking whether
8930 the underlying file system of a journal file is capable of
8931 properly reporting file change notifications, or whether
8932 applications that want to reflect journal changes "live"
8933 need to recheck journal files continuously in appropriate
8936 * It is now possible to set the "age" field for tmpfiles
8937 entries to 0, indicating that files matching this entry
8938 shall always be removed when the directories are cleaned up.
8940 * coredumpctl gained a new "gdb" verb which invokes gdb
8941 right-away on the selected coredump.
8943 * There's now support for "hybrid sleep" on kernels that
8944 support this, in addition to "suspend" and "hibernate". Use
8945 "systemctl hybrid-sleep" to make use of this.
8947 * logind's HandleSuspendKey= setting (and related settings)
8948 now gained support for a new "lock" setting to simply
8949 request the screen lock on all local sessions, instead of
8950 actually executing a suspend or hibernation.
8952 * systemd will now mount the EFI variables file system by
8955 * Socket units now gained support for configuration of the
8956 SMACK security label.
8958 * timedatectl will now output the time of the last and next
8959 daylight saving change.
8961 * We dropped support for various legacy and distro-specific
8962 concepts, such as insserv, early-boot SysV services
8963 (i.e. those for non-standard runlevels such as 'b' or 'S')
8964 or ArchLinux /etc/rc.conf support. We recommend the
8965 distributions who still need support this to either continue
8966 to maintain the necessary patches downstream, or find a
8967 different solution. (Talk to us if you have questions!)
8969 * Various systemd components will now bypass polkit checks for
8970 root and otherwise handle properly if polkit is not found to
8971 be around. This should fix most issues for polkit-less
8972 systems. Quite frankly this should have been this way since
8973 day one. It is absolutely our intention to make systemd work
8974 fine on polkit-less systems, and we consider it a bug if
8975 something does not work as it should if polkit is not around.
8977 * For embedded systems it is now possible to build udev and
8978 systemd without blkid and/or kmod support.
8980 * "systemctl switch-root" is now capable of switching root
8981 more than once. I.e. in addition to transitions from the
8982 initrd to the host OS it is now possible to transition to
8983 further OS images from the host. This is useful to implement
8984 offline updating tools.
8986 * Various other additions have been made to the RPM macros
8987 shipped with systemd. Use %udev_rules_update() after
8988 installing new udev rules files. %_udevhwdbdir,
8989 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
8990 %_sysctldir are now available which resolve to the right
8991 directories for packages to place various data files in.
8993 * journalctl gained the new --full switch (in addition to
8994 --all, to disable ellipsation for long messages.
8996 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
8997 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
8998 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
8999 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
9000 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
9001 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
9002 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
9003 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
9004 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
9008 * journalctl gained new --since= and --until= switches to
9009 filter by time. It also now supports nice filtering for
9010 units via --unit=/-u.
9012 * Type=oneshot services may use ExecReload= and do the
9015 * The journal daemon now supports time-based rotation and
9016 vacuuming, in addition to the usual disk-space based
9019 * The journal will now index the available field values for
9020 each field name. This enables clients to show pretty drop
9021 downs of available match values when filtering. The bash
9022 completion of journalctl has been updated
9023 accordingly. journalctl gained a new switch -F to list all
9024 values a certain field takes in the journal database.
9026 * More service events are now written as structured messages
9027 to the journal, and made recognizable via message IDs.
9029 * The timedated, localed and hostnamed mini-services which
9030 previously only provided support for changing time, locale
9031 and hostname settings from graphical DEs such as GNOME now
9032 also have a minimal (but very useful) text-based client
9033 utility each. This is probably the nicest way to changing
9034 these settings from the command line now, especially since
9035 it lists available options and is fully integrated with bash
9038 * There's now a new tool "systemd-coredumpctl" to list and
9039 extract coredumps from the journal.
9041 * We now install a README each in /var/log/ and
9042 /etc/rc.d/init.d explaining where the system logs and init
9043 scripts went. This hopefully should help folks who go to
9044 that dirs and look into the otherwise now empty void and
9045 scratch their heads.
9047 * When user-services are invoked (by systemd --user) the
9048 $MANAGERPID env var is set to the PID of systemd.
9050 * SIGRTMIN+24 when sent to a --user instance will now result
9051 in immediate termination of systemd.
9053 * gatewayd received numerous feature additions such as a
9054 "follow" mode, for live syncing and filtering.
9056 * browse.html now allows filtering and showing detailed
9057 information on specific entries. Keyboard navigation and
9058 mouse screen support has been added.
9060 * gatewayd/journalctl now supports HTML5/JSON
9061 Server-Sent-Events as output.
9063 * The SysV init script compatibility logic will now
9064 heuristically determine whether a script supports the
9065 "reload" verb, and only then make this available as
9068 * "systemctl status --follow" has been removed, use "journalctl
9071 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
9072 have been removed since they are hardly useful to be
9075 * And I'd like to take the opportunity to specifically mention
9076 Zbigniew for his great contributions. Zbigniew, you rock!
9078 Contributions from: Andrew Eikum, Christian Hesse, Colin
9079 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
9080 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
9081 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
9082 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
9083 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
9084 Jędrzejewski-Szmek, Сковорода Никита Андреевич
9088 * If /etc/vconsole.conf is non-existent or empty we will no
9089 longer load any console font or key map at boot by
9090 default. Instead the kernel defaults will be left
9091 intact. This is definitely the right thing to do, as no
9092 configuration should mean no configuration, and hard-coding
9093 font names that are different on all archs is probably a bad
9094 idea. Also, the kernel default key map and font should be
9095 good enough for most cases anyway, and mostly identical to
9096 the userspace fonts/key maps we previously overloaded them
9097 with. If distributions want to continue to default to a
9098 non-kernel font or key map they should ship a default
9099 /etc/vconsole.conf with the appropriate contents.
9101 Contributions from: Colin Walters, Daniel J Walsh, Dave
9102 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
9103 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
9107 * journalctl gained a new --cursor= switch to show entries
9108 starting from the specified location in the journal.
9110 * We now enforce a size limit on journal entry fields exported
9111 with "-o json" in journalctl. Fields larger than 4K will be
9112 assigned null. This can be turned off with --all.
9114 * An (optional) journal gateway daemon is now available as
9115 "systemd-journal-gatewayd.service". This service provides
9116 access to the journal via HTTP and JSON. This functionality
9117 will be used to implement live log synchronization in both
9118 pull and push modes, but has various other users too, such
9119 as easy log access for debugging of embedded devices. Right
9120 now it is already useful to retrieve the journal via HTTP:
9122 # systemctl start systemd-journal-gatewayd.service
9123 # wget http://localhost:19531/entries
9125 This will download the journal contents in a
9126 /var/log/messages compatible format. The same as JSON:
9128 # curl -H"Accept: application/json" http://localhost:19531/entries
9130 This service is also accessible via a web browser where a
9131 single static HTML5 app is served that uses the JSON logic
9132 to enable the user to do some basic browsing of the
9133 journal. This will be extended later on. Here's an example
9134 screenshot of this app in its current state:
9136 http://0pointer.de/public/journal-gatewayd
9138 Contributions from: Kay Sievers, Lennart Poettering, Robert
9139 Milasan, Tom Gundersen
9143 * The bash completion logic is now available for journalctl
9146 * We do not mount the "cpuset" controller anymore together with
9147 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
9148 started if no parameters are assigned to it. "cpuset" hence
9149 broke code that assumed it could create "cpu" groups and
9152 * journalctl -f will now subscribe to terminal size changes,
9153 and line break accordingly.
9155 Contributions from: Dave Reisner, Kay Sievers, Lennart
9156 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
9160 * nspawn will now create a symlink /etc/localtime in the
9161 container environment, copying the host's timezone
9162 setting. Previously this has been done via a bind mount, but
9163 since symlinks cannot be bind mounted this has now been
9164 changed to create/update the appropriate symlink.
9166 * journalctl -n's line number argument is now optional, and
9167 will default to 10 if omitted.
9169 * journald will now log the maximum size the journal files may
9170 take up on disk. This is particularly useful if the default
9171 built-in logic of determining this parameter from the file
9172 system size is used. Use "systemctl status
9173 systemd-journald.service" to see this information.
9175 * The multi-seat X wrapper tool has been stripped down. As X
9176 is now capable of enumerating graphics devices via udev in a
9177 seat-aware way the wrapper is not strictly necessary
9178 anymore. A stripped down temporary stop-gap is still shipped
9179 until the upstream display managers have been updated to
9180 fully support the new X logic. Expect this wrapper to be
9181 removed entirely in one of the next releases.
9183 * HandleSleepKey= in logind.conf has been split up into
9184 HandleSuspendKey= and HandleHibernateKey=. The old setting
9185 is not available anymore. X11 and the kernel are
9186 distinguishing between these keys and we should too. This
9187 also means the inhibition lock for these keys has been split
9190 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
9191 Poettering, Lukas Nykryn, Václav Pavlín
9195 * Whenever a unit changes state we will now log this to the
9196 journal and show along the unit's own log output in
9199 * ConditionPathIsMountPoint= can now properly detect bind
9200 mount points too. (Previously, a bind mount of one file
9201 system to another place in the same file system could not be
9202 detected as mount, since they shared struct stat's st_dev
9205 * We will now mount the cgroup controllers cpu, cpuacct,
9206 cpuset and the controllers net_cls, net_prio together by
9209 * nspawn containers will now have a virtualized boot
9210 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
9211 over with a randomized ID at container initialization). This
9212 has the effect of making "journalctl -b" do the right thing
9215 * The JSON output journal serialization has been updated not
9216 to generate "endless" list objects anymore, but rather one
9217 JSON object per line. This is more in line how most JSON
9218 parsers expect JSON objects. The new output mode
9219 "json-pretty" has been added to provide similar output, but
9220 neatly aligned for readability by humans.
9222 * We dropped all explicit sync() invocations in the shutdown
9223 code. The kernel does this implicitly anyway in the kernel
9224 reboot() syscall. halt(8)'s -n option is now a compatibility
9227 * We now support virtualized reboot() in containers, as
9228 supported by newer kernels. We will fall back to exit() if
9229 CAP_SYS_REBOOT is not available to the container. Also,
9230 nspawn makes use of this now and will actually reboot the
9231 container if the containerized OS asks for that.
9233 * journalctl will only show local log output by default
9234 now. Use --merge (-m) to show remote log output, too.
9236 * libsystemd-journal gained the new sd_journal_get_usage()
9237 call to determine the current disk usage of all journal
9238 files. This is exposed in the new "journalctl --disk-usage"
9241 * journald gained a new configuration setting SplitMode= in
9242 journald.conf which may be used to control how user journals
9243 are split off. See journald.conf(5) for details.
9245 * A new condition type ConditionFileNotEmpty= has been added.
9247 * tmpfiles' "w" lines now support file globbing, to write
9248 multiple files at once.
9250 * We added Python bindings for the journal submission
9251 APIs. More Python APIs for a number of selected APIs will
9252 likely follow. Note that we intend to add native bindings
9253 only for the Python language, as we consider it common
9254 enough to deserve bindings shipped within systemd. There are
9255 various projects outside of systemd that provide bindings
9256 for languages such as PHP or Lua.
9258 * Many conditions will now resolve specifiers such as %i. In
9259 addition, PathChanged= and related directives of .path units
9260 now support specifiers as well.
9262 * There's now a new RPM macro definition for the system preset
9265 * journald will now warn if it ca not forward a message to the
9266 syslog daemon because its socket is full.
9268 * timedated will no longer write or process /etc/timezone,
9269 except on Debian. As we do not support late mounted /usr
9270 anymore /etc/localtime always being a symlink is now safe,
9271 and hence the information in /etc/timezone is not necessary
9274 * logind will now always reserve one VT for a text getty (VT6
9275 by default). Previously if more than 6 X sessions where
9276 started they took up all the VTs with auto-spawned gettys,
9277 so that no text gettys were available anymore.
9279 * udev will now automatically inform the btrfs kernel logic
9280 about btrfs RAID components showing up. This should make
9281 simple hotplug based btrfs RAID assembly work.
9283 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
9284 (but not for its children which will stay at the kernel
9285 default). This should allow setups with a lot more listening
9288 * systemd will now always pass the configured timezone to the
9289 kernel at boot. timedated will do the same when the timezone
9292 * logind's inhibition logic has been updated. By default,
9293 logind will now handle the lid switch, the power and sleep
9294 keys all the time, even in graphical sessions. If DEs want
9295 to handle these events on their own they should take the new
9296 handle-power-key, handle-sleep-key and handle-lid-switch
9297 inhibitors during their runtime. A simple way to achieve
9298 that is to invoke the DE wrapped in an invocation of:
9300 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
9302 * Access to unit operations is now checked via SELinux taking
9303 the unit file label and client process label into account.
9305 * systemd will now notify the administrator in the journal
9306 when he over-mounts a non-empty directory.
9308 * There are new specifiers that are resolved in unit files,
9309 for the host name (%H), the machine ID (%m) and the boot ID
9312 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
9313 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
9314 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
9315 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
9316 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
9317 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
9318 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
9322 * Support for reading structured kernel messages from
9323 /dev/kmsg has now been added and is enabled by default.
9325 * Support for reading kernel messages from /proc/kmsg has now
9326 been removed. If you want kernel messages in the journal
9327 make sure to run a recent kernel (>= 3.5) that supports
9328 reading structured messages from /dev/kmsg (see
9329 above). /proc/kmsg is now exclusive property of classic
9330 syslog daemons again.
9332 * The libudev API gained the new
9333 udev_device_new_from_device_id() call.
9335 * The logic for file system namespace (ReadOnlyDirectory=,
9336 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
9337 require pivot_root() anymore. This means fewer temporary
9338 directories are created below /tmp for this feature.
9340 * nspawn containers will now see and receive all submounts
9341 made on the host OS below the root file system of the
9344 * Forward Secure Sealing is now supported for Journal files,
9345 which provide cryptographical sealing of journal files so
9346 that attackers cannot alter log history anymore without this
9347 being detectable. Lennart will soon post a blog story about
9348 this explaining it in more detail.
9350 * There are two new service settings RestartPreventExitStatus=
9351 and SuccessExitStatus= which allow configuration of exit
9352 status (exit code or signal) which will be excepted from the
9353 restart logic, resp. consider successful.
9355 * journalctl gained the new --verify switch that can be used
9356 to check the integrity of the structure of journal files and
9357 (if Forward Secure Sealing is enabled) the contents of
9360 * nspawn containers will now be run with /dev/stdin, /dev/fd/
9361 and similar symlinks pre-created. This makes running shells
9362 as container init process a lot more fun.
9364 * The fstab support can now handle PARTUUID= and PARTLABEL=
9367 * A new ConditionHost= condition has been added to match
9368 against the hostname (with globs) and machine ID. This is
9369 useful for clusters where a single OS image is used to
9370 provision a large number of hosts which shall run slightly
9371 different sets of services.
9373 * Services which hit the restart limit will now be placed in a
9376 Contributions from: Bertram Poettering, Dave Reisner, Huang
9377 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
9378 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
9382 * When running in --user mode systemd will now become a
9383 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
9384 tree a lot more organized.
9386 * A new PartOf= unit dependency type has been introduced that
9387 may be used to group services in a natural way.
9389 * "systemctl enable" may now be used to enable instances of
9392 * journalctl now prints error log levels in red, and
9393 warning/notice log levels in bright white. It also supports
9394 filtering by log level now.
9396 * cgtop gained a new -n switch (similar to top), to configure
9397 the maximum number of iterations to run for. It also gained
9398 -b, to run in batch mode (accepting no input).
9400 * The suffix ".service" may now be omitted on most systemctl
9401 command lines involving service unit names.
9403 * There's a new bus call in logind to lock all sessions, as
9404 well as a loginctl verb for it "lock-sessions".
9406 * libsystemd-logind.so gained a new call sd_journal_perror()
9407 that works similar to libc perror() but logs to the journal
9408 and encodes structured information about the error number.
9410 * /etc/crypttab entries now understand the new keyfile-size=
9413 * shutdown(8) now can send a (configurable) wall message when
9414 a shutdown is cancelled.
9416 * The mount propagation mode for the root file system will now
9417 default to "shared", which is useful to make containers work
9418 nicely out-of-the-box so that they receive new mounts from
9419 the host. This can be undone locally by running "mount
9420 --make-rprivate /" if needed.
9422 * The prefdm.service file has been removed. Distributions
9423 should maintain this unit downstream if they intend to keep
9424 it around. However, we recommend writing normal unit files
9425 for display managers instead.
9427 * Since systemd is a crucial part of the OS we will now
9428 default to a number of compiler switches that improve
9429 security (hardening) such as read-only relocations, stack
9430 protection, and suchlike.
9432 * The TimeoutSec= setting for services is now split into
9433 TimeoutStartSec= and TimeoutStopSec= to allow configuration
9434 of individual time outs for the start and the stop phase of
9437 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
9438 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
9439 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
9440 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
9441 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
9442 Gundersen, Zbigniew Jędrzejewski-Szmek
9446 * The journal and id128 C APIs are now fully documented as man
9449 * Extra safety checks have been added when transitioning from
9450 the initial RAM disk to the main system to avoid accidental
9453 * /etc/crypttab entries now understand the new keyfile-offset=
9456 * systemctl -t can now be used to filter by unit load state.
9458 * The journal C API gained the new sd_journal_wait() call to
9459 make writing synchronous journal clients easier.
9461 * journalctl gained the new -D switch to show journals from a
9464 * journalctl now displays a special marker between log
9465 messages of two different boots.
9467 * The journal is now explicitly flushed to /var via a service
9468 systemd-journal-flush.service, rather than implicitly simply
9469 by seeing /var/log/journal to be writable.
9471 * journalctl (and the journal C APIs) can now match for much
9472 more complex expressions, with alternatives and
9475 * When transitioning from the initial RAM disk to the main
9476 system we will now kill all processes in a killing spree to
9477 ensure no processes stay around by accident.
9479 * Three new specifiers may be used in unit files: %u, %h, %s
9480 resolve to the user name, user home directory resp. user
9481 shell. This is useful for running systemd user instances.
9483 * We now automatically rotate journal files if their data
9484 object hash table gets a fill level > 75%. We also size the
9485 hash table based on the configured maximum file size. This
9486 together should lower hash collisions drastically and thus
9487 speed things up a bit.
9489 * journalctl gained the new "--header" switch to introspect
9490 header data of journal files.
9492 * A new setting SystemCallFilters= has been added to services
9493 which may be used to apply blacklists or whitelists to
9494 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
9496 * nspawn gained a new --link-journal= switch (and quicker: -j)
9497 to link the container journal with the host. This makes it
9498 very easy to centralize log viewing on the host for all
9499 guests while still keeping the journal files separated.
9501 * Many bugfixes and optimizations
9503 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
9504 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
9505 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
9510 * Several tools now understand kernel command line arguments,
9511 which are only read when run in an initial RAM disk. They
9512 usually follow closely their normal counterparts, but are
9515 * There's a new tool to analyze the readahead files that are
9516 automatically generated at boot. Use:
9518 /usr/lib/systemd/systemd-readahead analyze /.readahead
9520 * We now provide an early debug shell on tty9 if this enabled. Use:
9522 systemctl enable debug-shell.service
9524 * All plymouth related units have been moved into the Plymouth
9525 package. Please make sure to upgrade your Plymouth version
9528 * systemd-tmpfiles now supports getting passed the basename of
9529 a configuration file only, in which case it will look for it
9530 in all appropriate directories automatically.
9532 * udevadm info now takes a /dev or /sys path as argument, and
9533 does the right thing. Example:
9535 udevadm info /dev/sda
9536 udevadm info /sys/class/block/sda
9538 * systemctl now prints a warning if a unit is stopped but a
9539 unit that might trigger it continues to run. Example: a
9540 service is stopped but the socket that activates it is left
9543 * "systemctl status" will now mention if the log output was
9544 shortened due to rotation since a service has been started.
9546 * The journal API now exposes functions to determine the
9547 "cutoff" times due to rotation.
9549 * journald now understands SIGUSR1 and SIGUSR2 for triggering
9550 immediately flushing of runtime logs to /var if possible,
9551 resp. for triggering immediate rotation of the journal
9554 * It is now considered an error if a service is attempted to
9555 be stopped that is not loaded.
9557 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
9559 * systemd-analyze now supports Python 3
9561 * tmpfiles now supports cleaning up directories via aging
9562 where the first level dirs are always kept around but
9563 directories beneath it automatically aged. This is enabled
9564 by prefixing the age field with '~'.
9566 * Seat objects now expose CanGraphical, CanTTY properties
9567 which is required to deal with very fast bootups where the
9568 display manager might be running before the graphics drivers
9569 completed initialization.
9571 * Seat objects now expose a State property.
9573 * We now include RPM macros for service enabling/disabling
9574 based on the preset logic. We recommend RPM based
9575 distributions to make use of these macros if possible. This
9576 makes it simpler to reuse RPM spec files across
9579 * We now make sure that the collected systemd unit name is
9580 always valid when services log to the journal via
9583 * There's a new man page kernel-command-line(7) detailing all
9584 command line options we understand.
9586 * The fstab generator may now be disabled at boot by passing
9587 fstab=0 on the kernel command line.
9589 * A new kernel command line option modules-load= is now understood
9590 to load a specific kernel module statically, early at boot.
9592 * Unit names specified on the systemctl command line are now
9593 automatically escaped as needed. Also, if file system or
9594 device paths are specified they are automatically turned
9595 into the appropriate mount or device unit names. Example:
9597 systemctl status /home
9598 systemctl status /dev/sda
9600 * The SysVConsole= configuration option has been removed from
9601 system.conf parsing.
9603 * The SysV search path is no longer exported on the D-Bus
9606 * The Names= option has been removed from unit file parsing.
9608 * There's a new man page bootup(7) detailing the boot process.
9610 * Every unit and every generator we ship with systemd now
9611 comes with full documentation. The self-explanatory boot is
9614 * A couple of services gained "systemd-" prefixes in their
9615 name if they wrap systemd code, rather than only external
9616 code. Among them fsck@.service which is now
9617 systemd-fsck@.service.
9619 * The HaveWatchdog property has been removed from the D-Bus
9622 * systemd.confirm_spawn= on the kernel command line should now
9625 * There's a new man page crypttab(5) which details all options
9626 we actually understand.
9628 * systemd-nspawn gained a new --capability= switch to pass
9629 additional capabilities to the container.
9631 * timedated will now read known NTP implementation unit names
9632 from /usr/lib/systemd/ntp-units.d/*.list,
9633 systemd-timedated-ntp.target has been removed.
9635 * journalctl gained a new switch "-b" that lists log data of
9636 the current boot only.
9638 * The notify socket is in the abstract namespace again, in
9639 order to support daemons which chroot() at start-up.
9641 * There is a new Storage= configuration option for journald
9642 which allows configuration of where log data should go. This
9643 also provides a way to disable journal logging entirely, so
9644 that data collected is only forwarded to the console, the
9645 kernel log buffer or another syslog implementation.
9647 * Many bugfixes and optimizations
9649 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
9650 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
9651 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
9652 Shawn Landden, Tom Gundersen
9656 * "systemctl help <unit>" now shows the man page if one is
9659 * Several new man pages have been added.
9661 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
9662 MaxLevelConsole= can now be specified in
9663 journald.conf. These options allow reducing the amount of
9664 data stored on disk or forwarded by the log level.
9666 * TimerSlackNSec= can now be specified in system.conf for
9667 PID1. This allows system-wide power savings.
9669 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
9670 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
9675 * logind is now capable of (optionally) handling power and
9676 sleep keys as well as the lid switch.
9678 * journalctl now understands the syntax "journalctl
9679 /usr/bin/avahi-daemon" to get all log output of a specific
9682 * CapabilityBoundingSet= in system.conf now also influences
9683 the capability bound set of usermode helpers of the kernel.
9685 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
9686 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
9687 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
9688 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
9692 * Note that we skipped 139 releases here in order to set the
9693 new version to something that is greater than both udev's
9694 and systemd's most recent version number.
9696 * udev: all udev sources are merged into the systemd source tree now.
9697 All future udev development will happen in the systemd tree. It
9698 is still fully supported to use the udev daemon and tools without
9699 systemd running, like in initramfs or other init systems. Building
9700 udev though, will require the *build* of the systemd tree, but
9701 udev can be properly *run* without systemd.
9703 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
9704 should be used to create dead device nodes as workarounds for broken
9707 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
9708 no longer supported. udev_monitor_new_from_netlink() needs to be
9709 used to subscribe to events.
9711 * udev: when udevd is started by systemd, processes which are left
9712 behind by forking them off of udev rules, are unconditionally cleaned
9713 up and killed now after the event handling has finished. Services or
9714 daemons must be started as systemd services. Services can be
9715 pulled-in by udev to get started, but they can no longer be directly
9716 forked by udev rules.
9718 * udev: the daemon binary is called systemd-udevd now and installed
9719 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
9720 to adapt to that, create symlink, or rename the binary after building
9723 * libudev no longer provides these symbols:
9724 udev_monitor_from_socket()
9725 udev_queue_get_failed_list_entry()
9726 udev_get_{dev,sys,run}_path()
9727 The versions number was bumped and symbol versioning introduced.
9729 * systemd-loginctl and systemd-journalctl have been renamed
9730 to loginctl and journalctl to match systemctl.
9732 * The config files: /etc/systemd/systemd-logind.conf and
9733 /etc/systemd/systemd-journald.conf have been renamed to
9734 logind.conf and journald.conf. Package updates should rename
9735 the files to the new names on upgrade.
9737 * For almost all files the license is now LGPL2.1+, changed
9738 from the previous GPL2.0+. Exceptions are some minor stuff
9739 of udev (which will be changed to LGPL2.1 eventually, too),
9740 and the MIT licensed sd-daemon.[ch] library that is suitable
9741 to be used as drop-in files.
9743 * systemd and logind now handle system sleep states, in
9744 particular suspending and hibernating.
9746 * logind now implements a sleep/shutdown/idle inhibiting logic
9747 suitable for a variety of uses. Soonishly Lennart will blog
9748 about this in more detail.
9750 * var-run.mount and var-lock.mount are no longer provided
9751 (which previously bind mounted these directories to their new
9752 places). Distributions which have not converted these
9753 directories to symlinks should consider stealing these files
9754 from git history and add them downstream.
9756 * We introduced the Documentation= field for units and added
9757 this to all our shipped units. This is useful to make it
9758 easier to explore the boot and the purpose of the various
9761 * All smaller setup units (such as
9762 systemd-vconsole-setup.service) now detect properly if they
9763 are run in a container and are skipped when
9764 appropriate. This guarantees an entirely noise-free boot in
9765 Linux container environments such as systemd-nspawn.
9767 * A framework for implementing offline system updates is now
9768 integrated, for details see:
9769 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
9771 * A new service type Type=idle is available now which helps us
9772 avoiding ugly interleaving of getty output and boot status
9775 * There's now a system-wide CapabilityBoundingSet= option to
9776 globally reduce the set of capabilities for the
9777 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
9778 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
9779 even CAP_NET_ADMIN system-wide for secure systems.
9781 * There are now system-wide DefaultLimitXXX= options to
9782 globally change the defaults of the various resource limits
9783 for all units started by PID 1.
9785 * Harald Hoyer's systemd test suite has been integrated into
9786 systemd which allows easy testing of systemd builds in qemu
9787 and nspawn. (This is really awesome! Ask us for details!)
9789 * The fstab parser is now implemented as generator, not inside
9792 * systemctl will now warn you if .mount units generated from
9793 /etc/fstab are out of date due to changes in fstab that
9794 have not been read by systemd yet.
9796 * systemd is now suitable for usage in initrds. Dracut has
9797 already been updated to make use of this. With this in place
9798 initrds get a slight bit faster but primarily are much
9799 easier to introspect and debug since "systemctl status" in
9800 the host system can be used to introspect initrd services,
9801 and the journal from the initrd is kept around too.
9803 * systemd-delta has been added, a tool to explore differences
9804 between user/admin configuration and vendor defaults.
9806 * PrivateTmp= now affects both /tmp and /var/tmp.
9808 * Boot time status messages are now much prettier and feature
9809 proper english language. Booting up systemd has never been
9812 * Read-ahead pack files now include the inode number of all
9813 files to pre-cache. When the inode changes the pre-caching
9814 is not attempted. This should be nicer to deal with updated
9815 packages which might result in changes of read-ahead
9818 * We now temporaritly lower the kernel's read_ahead_kb variable
9819 when collecting read-ahead data to ensure the kernel's
9820 built-in read-ahead does not add noise to our measurements
9821 of necessary blocks to pre-cache.
9823 * There's now RequiresMountsFor= to add automatic dependencies
9824 for all mounts necessary for a specific file system path.
9826 * MountAuto= and SwapAuto= have been removed from
9827 system.conf. Mounting file systems at boot has to take place
9830 * nspawn now learned a new switch --uuid= to set the machine
9831 ID on the command line.
9833 * nspawn now learned the -b switch to automatically search
9836 * vt102 is now the default TERM for serial TTYs, upgraded from
9839 * systemd-logind now works on VT-less systems.
9841 * The build tree has been reorganized. The individual
9842 components now have directories of their own.
9844 * A new condition type ConditionPathIsReadWrite= is now available.
9846 * nspawn learned the new -C switch to create cgroups for the
9847 container in other hierarchies.
9849 * We now have support for hardware watchdogs, configurable in
9852 * The scheduled shutdown logic now has a public API.
9854 * We now mount /tmp as tmpfs by default, but this can be
9855 masked and /etc/fstab can override it.
9857 * Since udisks does not make use of /media anymore we are not
9858 mounting a tmpfs on it anymore.
9860 * journalctl gained a new --local switch to only interleave
9861 locally generated journal files.
9863 * We can now load the IMA policy at boot automatically.
9865 * The GTK tools have been split off into a systemd-ui.
9867 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
9868 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
9869 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
9870 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
9871 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
9872 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
9873 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
9874 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
9875 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
9880 * This is mostly a bugfix release
9882 * Support optional initialization of the machine ID from the
9883 KVM or container configured UUID.
9885 * Support immediate reboots with "systemctl reboot -ff"
9887 * Show /etc/os-release data in systemd-analyze output
9889 * Many bugfixes for the journal, including endianness fixes and
9890 ensuring that disk space enforcement works
9892 * sd-login.h is C++ compatible again
9894 * Extend the /etc/os-release format on request of the Debian
9897 * We now refuse non-UTF8 strings used in various configuration
9898 and unit files. This is done to ensure we do not pass invalid
9899 data over D-Bus or expose it elsewhere.
9901 * Register Mimo USB Screens as suitable for automatic seat
9904 * Read SELinux client context from journal clients in a race
9907 * Reorder configuration file lookup order. /etc now always
9908 overrides /run in order to allow the administrator to always
9909 and unconditionally override vendor-supplied or
9910 automatically generated data.
9912 * The various user visible bits of the journal now have man
9913 pages. We still lack man pages for the journal API calls
9916 * We now ship all man pages in HTML format again in the
9919 Contributions from: Dave Reisner, Dirk Eibach, Frederic
9920 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
9921 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
9926 * This is mostly a bugfix release
9928 * systems lacking /etc/os-release are no longer supported.
9930 * Various functionality updates to libsystemd-login.so
9932 * Track class of PAM logins to distinguish greeters from
9935 Contributions from: Kay Sievers, Lennart Poettering, Michael
9940 * This is an important bugfix release for v41.
9942 * Building man pages is now optional which should be useful
9943 for those building systemd from git but unwilling to install
9946 * Watchdog support for supervising services is now usable. In
9947 a future release support for hardware watchdogs
9948 (i.e. /dev/watchdog) will be added building on this.
9950 * Service start rate limiting is now configurable and can be
9951 turned off per service. When a start rate limit is hit a
9952 reboot can automatically be triggered.
9954 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
9956 Contributions from: Benjamin Franzke, Bill Nottingham,
9957 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
9958 Schmidt, Michał Górny, Piotr Drąg
9962 * The systemd binary is installed /usr/lib/systemd/systemd now;
9963 An existing /sbin/init symlink needs to be adapted with the
9966 * The code that loads kernel modules has been ported to invoke
9967 libkmod directly, instead of modprobe. This means we do not
9968 support systems with module-init-tools anymore.
9970 * Watchdog support is now already useful, but still not
9973 * A new kernel command line option systemd.setenv= is
9974 understood to set system wide environment variables
9975 dynamically at boot.
9977 * We now limit the set of capabilities of systemd-journald.
9979 * We now set SIGPIPE to ignore by default, since it only is
9980 useful in shell pipelines, and has little use in general
9981 code. This can be disabled with IgnoreSIPIPE=no in unit
9984 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
9985 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
9990 * This is mostly a bugfix release
9992 * We now expose the reason why a service failed in the
9993 "Result" D-Bus property.
9995 * Rudimentary service watchdog support (will be completed over
9996 the next few releases.)
9998 * When systemd forks off in order execute some service we will
9999 now immediately changes its argv[0] to reflect which process
10000 it will execute. This is useful to minimize the time window
10001 with a generic argv[0], which makes bootcharts more useful
10003 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
10004 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
10005 Mike Kazantsev, Ray Strode
10009 * This is mostly a test release, but incorporates many
10012 * New systemd-cgtop tool to show control groups by their
10015 * Linking against libacl for ACLs is optional again. If
10016 disabled, support tracking device access for active logins
10017 goes becomes unavailable, and so does access to the user
10018 journals by the respective users.
10020 * If a group "adm" exists, journal files are automatically
10021 owned by them, thus allow members of this group full access
10022 to the system journal as well as all user journals.
10024 * The journal now stores the SELinux context of the logging
10025 client for all entries.
10027 * Add C++ inclusion guards to all public headers
10029 * New output mode "cat" in the journal to print only text
10030 messages, without any meta data like date or time.
10032 * Include tiny X server wrapper as a temporary stop-gap to
10033 teach XOrg udev display enumeration. This is used by display
10034 managers such as gdm, and will go away as soon as XOrg
10035 learned native udev hotplugging for display devices.
10037 * Add new systemd-cat tool for executing arbitrary programs
10038 with STDERR/STDOUT connected to the journal. Can also act as
10039 BSD logger replacement, and does so by default.
10041 * Optionally store all locally generated coredumps in the
10042 journal along with meta data.
10044 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
10045 writing short strings to files (for usage for /sys), and for
10046 creating symlinks, character and block device nodes.
10048 * New unit file option ControlGroupPersistent= to make cgroups
10049 persistent, following the mechanisms outlined in
10050 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
10052 * Support multiple local RTCs in a sane way
10054 * No longer monopolize IO when replaying readahead data on
10055 rotating disks, since we might starve non-file-system IO to
10056 death, since fanotify() will not see accesses done by blkid,
10059 * Do not show kernel threads in systemd-cgls anymore, unless
10060 requested with new -k switch.
10062 Contributions from: Dan Horák, Kay Sievers, Lennart
10063 Poettering, Michal Schmidt
10067 * This is mostly a test release, but incorporates many
10070 * The git repository moved to:
10071 git://anongit.freedesktop.org/systemd/systemd
10072 ssh://git.freedesktop.org/git/systemd/systemd
10074 * First release with the journal
10075 http://0pointer.de/blog/projects/the-journal.html
10077 * The journal replaces both systemd-kmsg-syslogd and
10078 systemd-stdout-bridge.
10080 * New sd_pid_get_unit() API call in libsystemd-logind
10082 * Many systemadm clean-ups
10084 * Introduce remote-fs-pre.target which is ordered before all
10085 remote mounts and may be used to start services before all
10088 * Added Mageia support
10090 * Add bash completion for systemd-loginctl
10092 * Actively monitor PID file creation for daemons which exit in
10093 the parent process before having finished writing the PID
10094 file in the daemon process. Daemons which do this need to be
10095 fixed (i.e. PID file creation must have finished before the
10096 parent exits), but we now react a bit more gracefully to them.
10098 * Add colourful boot output, mimicking the well-known output
10099 of existing distributions.
10101 * New option PassCredentials= for socket units, for
10102 compatibility with a recent kernel ABI breakage.
10104 * /etc/rc.local is now hooked in via a generator binary, and
10105 thus will no longer act as synchronization point during
10108 * systemctl list-unit-files now supports --root=.
10110 * systemd-tmpfiles now understands two new commands: z, Z for
10111 relabelling files according to the SELinux database. This is
10112 useful to apply SELinux labels to specific files in /sys,
10113 among other things.
10115 * Output of SysV services is now forwarded to both the console
10116 and the journal by default, not only just the console.
10118 * New man pages for all APIs from libsystemd-login.
10120 * The build tree got reorganized and the build system is a
10121 lot more modular allowing embedded setups to specifically
10122 select the components of systemd they are interested in.
10124 * Support for Linux systems lacking the kernel VT subsystem is
10127 * configure's --with-rootdir= got renamed to
10128 --with-rootprefix= to follow the naming used by udev and
10131 * Unless specified otherwise we will now install to /usr instead
10132 of /usr/local by default.
10134 * Processes with '@' in argv[0][0] are now excluded from the
10135 final shut-down killing spree, following the logic explained
10137 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
10139 * All processes remaining in a service cgroup when we enter
10140 the START or START_PRE states are now killed with
10141 SIGKILL. That means it is no longer possible to spawn
10142 background processes from ExecStart= lines (which was never
10143 supported anyway, and bad style).
10145 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
10146 reloading of units together.
10148 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
10149 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
10150 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
10151 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
10152 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
projects / platform / upstream / systemd.git / blob