1 systemd System and Service Manager
3 CHANGES WITH 243 in spe:
5 * This release enables unprivileged programs (i.e. requiring neither
6 setuid nor file capabilities) to send ICMP Echo (i.e. ping) requests
7 by turning on the "net.ipv4.ping_group_range" sysctl of the Linux
8 kernel for the whole UNIX group range, i.e. all processes. This
9 change should be reasonably safe, as the kernel support for it was
10 specifically implemented to allow safe access to ICMP Echo for
11 processes lacking any privileges. If this is not desirable, it can be
12 disabled again by setting the parameter to "1 0".
14 * Previously, filters defined with SystemCallFilter= would have the
15 effect that any calling of an offending system call would terminate
16 the calling thread. This behaviour never made much sense, since
17 killing individual threads of unsuspecting processes is likely to
18 create more problems than it solves. With this release the default
19 action changed from killing the thread to killing the whole
20 process. For this to work correctly both a kernel version (>= 4.14)
21 and a libseccomp version (>= 2.4.0) supporting this new seccomp
22 action is required. If an older kernel or libseccomp is used the old
23 behaviour continues to be used. This change does not affect any
24 services that have no system call filters defined, or that use
25 SystemCallErrorNumber= (and thus see EPERM or another error instead
26 of being killed when calling an offending system call). Note that
27 systemd documentation always claimed that the whole process is
28 killed. With this change behaviour is thus adjusted to match the
31 * On 64 bit systems, the "kernel.pid_max" sysctl is now bumped to
32 4194304 by default, i.e. the full 22bit range the kernel allows, up
33 from the old 16bit range. This should improve security and
34 robustness, as PID collisions are made less likely (though certainly
35 still possible). There are rumours this might create compatibility
36 problems, though at this moment no practical ones are known to
37 us. Downstream distributions are hence advised to undo this change in
38 their builds if they are concerned about maximum compatibility, but
39 for everybody else we recommend leaving the value bumped. Besides
40 improving security and robustness this should also simplify things as
41 the maximum number of allowed concurrent tasks was previously bounded
42 by both "kernel.pid_max" and "kernel.threads-max" and now effectively
43 only a single knob is left ("kernel.threads-max"). There have been
44 concerns that usability is affected by this change because larger PID
45 numbers are harder to type, but we believe the change from 5 digits
46 to 7 digits doesn't hamper usability.
48 * MemoryLow= and MemoryMin= gained hierarchy-aware counterparts,
49 DefaultMemoryLow= and DefaultMemoryMin=, which can be used to
50 hierarchically set default memory protection values for a particular
51 subtree of the unit hierarchy.
53 * Memory protection directives can now take a value of zero, allowing
54 explicit opting out of a default value propagated by an ancestor.
56 * systemd now defaults to the "unified" cgroup hierarchy setup during
57 build-time, i.e. -Ddefault-hierarchy=unified is now the build-time
58 default. Previously, -Ddefault-hierarchy=hybrid was the default. This
59 change reflects the fact that cgroupsv2 support has matured
60 substantially in both systemd and in the kernel, and is clearly the
61 way forward. Downstream production distributions might want to
62 continue to use -Ddefault-hierarchy=hybrid (or even =legacy) for
63 their builds as unfortunately the popular container managers have not
64 caught up with the kernel API changes.
66 * Man pages are not built by default anymore (html pages were already
67 disabled by default), to make development builds quicker. When
68 building systemd for a full installation with documentation, meson
69 should be called with -Dman=true and/or -Dhtml=true as appropriate.
70 The default was changed based on the assumption that quick one-off or
71 repeated development builds are much more common than full optimized
72 builds for installation, and people need to pass various other
73 options to when doing "proper" builds anyway, so the gain from making
74 development builds quicker is bigger than the one time disruption for
77 Two scripts are created in the *build* directory to generate and
78 preview man and html pages on demand, e.g.:
80 build/man/man systemctl
81 build/man/html systemd.index
83 * libidn2 is used by default if both libidn2 and libidn are installed.
84 Please use -Dlibidn=true if libidn is preferred.
86 * The D-Bus "wire format" of the CPUAffinity= attribute is changed on
87 big-endian machines. Before, bytes were written and read in native
88 machine order as exposed by the native libc __cpu_mask interface.
89 Now, little-endian order is always used (CPUs 0–7 are described by
90 bits 0–7 in byte 0, CPUs 8–15 are described by byte 1, and so on).
91 This change fixes D-Bus calls that cross endianness boundary.
93 The presentation format used for CPUAffinity= by "systemctl show" and
94 "systemd-analyze dump" is changed to present CPU indices instead of
95 the raw __cpu_mask bitmask. For example, CPUAffinity=0-1 would be
96 shown as CPUAffinity=03000000000000000000000000000… (on
97 little-endian) or CPUAffinity=00000000000000300000000000000… (on
98 64-bit big-endian), and is now shown as CPUAffinity=0-1, matching the
99 input format. The maximum integer that will be printed in the new
100 format is 8191 (four digits), while the old format always used a very
101 long number (with the length varying by architecture), so they can be
102 unambiguously distinguished.
104 * /usr/sbin/halt.local is no longer supported. Implementation in
105 distributions was inconsistent and it seems this functionality was
108 To replace this functionality, users should:
109 - either define a new unit and make it a dependency of final.target
110 (systemctl add-wants final.target my-halt-local.service)
111 - or move the shutdown script to /usr/lib/systemd/system-shutdown/
112 and ensure that it accepts "halt", "poweroff", "reboot", and
113 "kexec" as an argument, see the description in systemd-shutdown(8).
115 * When a [Match] section in .link or .network file is empty (contains
116 no match patterns), a warning will be emitted. Please add any "match
117 all" pattern instead, e.g. OriginalName=* or Name=* in case all
118 interfaces should really be matched.
120 * A new setting NUMAPolicy= may be used to set process memory
121 allocation policy. This setting can be specified in
122 /etc/systemd/system.conf and hence will set the default policy for
123 PID1. The default policy can be overridden on a per-service
124 basis. The related setting NUMAMask= is used to specify NUMA node
125 mask that should be associated with the selected policy.
127 * PID 1 will now listen to Out-Of-Memory (OOM) events the kernel
128 generates when processes it manages are reaching their memory limits,
129 and will place their units in a special state, and optionally kill or
132 * The service manager will now expose bus properties for the IO
133 resources used by units. This information is also shown in "systemctl
134 status" now (for services that have IOAccounting=yes set). Moreover,
135 the IO accounting data is included in the resource log message
136 generated whenever a unit stops.
138 * Units may now configure an explicit time-out to wait for when killed
139 with SIGABRT, for example when a service watchdog is hit. Previously,
140 the regular TimeoutStopSec= time-out was applied in this case too —
141 now a separate time-out may be set using TimeoutAbortSec=.
143 * Services may now send a special WATCHDOG=trigger message with
144 sd_notify() to trigger an immediate "watchdog missed" event, and thus
145 trigger service termination. This is useful both for testing watchdog
146 handling, but also for defining error paths in services, that shall
147 be handled the same way as watchdog events.
149 * There are two new per-unit settings IPIngressFilterPath= and
150 IPEgressFilterPath= which allow configuration of a BPF program
151 (usually by specifying a path to a program uploaded to /sys/fs/bpf/)
152 to apply to the IP packet ingress/egress path of all processes of a
153 unit. This is useful to allow running systemd services with BPF
154 programs set up externally.
156 * systemctl gained a new "clean" verb for removing the state, cache,
157 runtime or logs directories of a service while it is terminated. The
158 new verb may also be used to remove the state maintained on disk for
159 timer units that have Persistent= configured.
161 * During the last phase of shutdown systemd will now automatically
162 increase the log level configured in the "kernel.printk" sysctl so
163 that any relevant loggable events happening during late shutdown are
164 made visible. Previously, loggable events happening so late during
165 shutdown were generally lost if the "kernel.printk" sysctl was set to
166 high thresholds, as regular logging daemons are terminated at that
167 time and thus nothing is written to disk.
169 * If processes terminated during the last phase of shutdown do not exit
170 quickly systemd will now show their names after a short time, to make
171 debugging easier. After a longer time-out they are forcibly killed,
174 * journalctl (and the other tools that display logs) will now highlight
175 warnings in yellow (previously, both LOG_NOTICE and LOG_WARNING where
176 shown in bright bold, now only LOG_NOTICE is). Moreover, audit logs
177 are now shown in blue color, to separate them visually from regular
178 logs. References to configuration files are now turned into clickable
179 links on terminals that support that.
181 * systemd-journald will now stop logging to /var/log/journal during
182 shutdown when /var/ is on a separate mount, so that it can be
183 unmounted safely during shutdown.
185 * systemd-resolved gained support for a new 'strict' DNS-over-TLS mode.
187 * systemd-resolved "Cache=" configuration option in resolved.conf has
188 been extended to also accept the 'no-negative' value. Previously,
189 only a boolean option was allowed (yes/no), having yes as the
190 default. If this option is set to 'no-negative', negative answers are
191 not cached while the old cache heuristics are used positive answers.
192 The default remains unchanged.
194 * The predictable naming scheme for network devices now supports
195 generating predictable names for "netdevsim" devices.
197 Moreover, the "en" prefix was dropped from the ID_NET_NAME_ONBOARD
200 Those two changes form a new net.naming-policy-scheme= entry.
201 Distributions which want to preserve naming stability may want to set
202 the -Ddefault-net-naming-scheme= configuration option.
204 * systemd-networkd now supports MACsec, nlmon, IPVTAP and Xfrm
207 * systemd-networkd's bridge FDB support now allows configuration of a
208 destination address for each entry (Destination=), as well as the
209 VXLAN VNI (VNI=), as well as an option to declare what an entry is
210 associated with (AssociatedWith=).
212 * systemd-networkd's DHCPv4 support now understands a new MaxAttempts=
213 option for configuring the maximum number of DHCP lease requests. It
214 also learnt a new BlackList= option for blacklisting DHCP servers (a
215 similar setting has also been added to the IPv6 RA client), as well
216 as a SendRelease= option for configuring whether to send a DHCP
217 RELEASE message when terminating.
219 * systemd-networkd's DHCPv4 and DHCPv6 stacks can now be configured
220 separately in the [DHCPv4] and [DHCPv6] sections.
222 * systemd-networkd's DHCP support will now optionally create an
223 implicit host route to the DNS server specified in the DHCP lease, in
224 addition to the routes listed explicitly in the lease. This should
225 ensure that in multi-homed systems DNS traffic leaves the systems on
226 the interface that acquired the DNS server information even if other
227 routes such as default routes exist. This behaviour may be turned on
228 with the new RoutesToDNS= option.
230 * systemd-networkd's VXLAN support gained a new option
231 GenericProtocolExtension= for enabling VXLAN Generic Protocol
232 Extension support, as well as IPDoNotFragment= for setting the IP
233 "Don't fragment" bit on outgoing packets. A similar option has been
234 added to the GENEVE support.
236 * In systemd-networkd's [Route] section you may now configure
237 FastOpenNoCookie= for configuring per-route TCP fast-open support, as
238 well as TTLPropagate= for configuring Label Switched Path (LSP) TTL
239 propagation. The Type= setting now supports local, broadcast,
240 anycast, multicast, any, xresolve routes, too.
242 * systemd-networkd's [Network] section learnt a new option
243 DefaultRouteOnDevice= for automatically configuring a default route
244 onto the network device.
246 * systemd-networkd's bridging support gained two new options ProxyARP=
247 and ProxyARPWifi= for configuring proxy ARP behaviour as well as
248 MulticastRouter= for configuring multicast routing behaviour. A new
249 option MulticastIGMPVersion= may be used to change bridge's multicast
250 Internet Group Management Protocol (IGMP) version.
252 * systemd-networkd's FooOverUDP support gained the ability to configure
253 local and peer IP addresses via Local= and Peer=. A new option
254 PeerPort= may be used to configure the peer's IP port.
256 * systemd-networkd's TUN support gained a new setting VnetHeader= for
257 tweaking Generic Segment Offload support.
259 * networkctl gained a new "delete" command for removing virtual network
260 devices, as well as a new "--stats" switch for showing device
263 * networkd.conf gained a new setting SpeedMeter= and
264 SpeedMeterIntervalSec=, to measure bitrate of network interfaces. The
265 measured speed may be shown by 'networkctl status'.
267 * "networkctl status" now displays MTU and queue lengths, and more
268 detailed information about VXLAN and bridge devices.
270 * systemd-networkd's .network and .link files gained a new Property=
271 setting in the [Match] section, to match against devices with
272 specific udev properties.
274 * systemd-networkd's tunnel support gained a new option
275 AssignToLoopback= for selecting whether to use the loopback device
276 "lo" as underlying device.
278 * systemd-networkd's MACAddress= setting in the [Neighbor] section has
279 been renamed to LinkLayerAddress=, and it now allows configuration of
282 * A new tool systemd-network-generator has been added that may generate
283 .network, .netdev and .link files from IP configuration specified on
284 the kernel command line in the format used by Dracut.
286 * The CriticalConnection= setting in .network files is now deprecated,
287 and replaced by a new KeepConfiguration= setting which allows more
288 detailed configuration of the IP configuration to keep in place.
290 * systemd-analyze gained a few new verbs:
292 - "systemd-analyze timestamp" parses and converts timestamps. This is
293 similar to the existing "systemd-analyze calendar" command which
294 does the same for recurring calendar events.
296 - "systemd-analyze timespan" parses and converts timespans (i.e.
297 durations as opposed to points in time).
299 - "systemd-analyze condition" will parse and test ConditionXYZ=
302 - "systemd-analyze exit-status" will parse and convert exit status
303 codes to their names and back.
305 - "systemd-analyze unit-files" will print a list of all unit
306 file paths and unit aliases.
308 * SuccessExitStatus=, RestartPreventExitStatus=, and
309 RestartForceExitStatus= now accept exit status names (e.g. "DATAERR"
310 is equivalent to "65"). Those exit status name mappings may be
311 displayed with the sytemd-analyze exit-status verb describe above.
313 * systemd-logind now exposes a per-session SetBrightness() bus call,
314 which may be used to securely change the brightness of a kernel
315 brightness device, if it belongs to the session's seat. By using this
316 call unprivileged clients can make changes to "backlight" and "leds"
317 devices securely with strict requirements on session membership.
318 Desktop environments may use this to generically make brightness
319 changes to such devices without shipping private SUID binaries or
320 udev rules for that purpose.
322 * "udevadm info" gained a --wait-for-initialization switch to wait for
323 a device to be initialized.
325 * systemd-hibernate-resume-generator will now look for resumeflags= on
326 the kernel command line, which is similar to rootflags= and may be
327 used to configure device timeout for the hibernation device.
329 * sd-event learnt a new API call sd_event_source_disable_unref() for
330 disabling and unref'ing an event source in a single function. A
331 related call sd_event_source_disable_unrefp() has been added for use
332 with gcc's cleanup extension.
334 * The sd-id128.h public API gained a new definition
335 SD_ID128_UUID_FORMAT_STR for formatting a 128bit ID in UUID format
338 * "busctl introspect" gained a new switch --xml-interface for dumping
339 XML introspection data unmodified.
341 * PID 1 may now show the unit name instead of the unit description
342 string in its status output during boot. This may be configured in
343 the StatusUnitFormat= setting in /etc/systemd/system.conf or the
344 kernel command line option systemd.status_unit_format=.
346 * PID 1 now understands a new option KExecWatchdogSec= in
347 /etc/systemd/system.conf to set a watchdog timeout for kexec reboots.
348 Previously watchdog functionality was only available for regular
349 reboots. The new setting defaults to off, because we don't know in
350 the general case if the watchdog will be reset after kexec (some
351 drivers do reset it, but not all), and the new userspace might not be
352 configured to handle the watchdog.
354 Moreover, the old ShutdownWatchdogSec= setting has been renamed to
355 RebootWatchdogSec= to more clearly communicate what it is about. The
356 old name is still accepted for compatibility.
358 * The systemd.debug_shell kernel command line option now optionally
359 takes a tty name to spawn the debug shell on, which allows a
360 different tty to be selected than the built-in default.
362 * Service units gained a new ExecCondition= setting which will run
363 before ExecStartPre= and either continue execution of the unit (for
364 clean exit codes), stop execution without marking the unit failed
365 (for exit codes 1 through 254), or stop execution and fail the unit
366 (for exit code 255 or abnormal termination).
368 * A new service systemd-pstore.service has been added that pulls data
369 from /sys/fs/pstore/ and saves it to /var/lib/pstore for later
372 * timedatectl gained new verbs for configuring per-interface NTP
373 service configuration for systemd-timesyncd.
375 * "localectl list-locales" won't list non-UTF-8 locales anymore. It's
376 2019. (You can set non-UTF-8 locales though, if you know their name.)
378 * If variable assignments in sysctl.d/ files are prefixed with "-" any
379 failures to apply them are now ignored.
381 * systemd-random-seed.service now optionally credits entropy when
382 applying the seed to the system. Set $SYSTEMD_RANDOM_SEED_CREDIT to
383 true for the service to enable this behaviour, but please consult the
384 documentation first, since this comes with a couple of caveats.
386 * systemd-random-seed.service is now a synchronization point for full
387 initialization of the kernel's entropy pool. Services that require
388 /dev/urandom to be correctly initialized should be ordered after this
391 * The systemd-boot boot loader has been updated to optionally maintain
392 a random seed file in the EFI System Partition (ESP). During the boot
393 phase, this random seed is read and updated with a new seed
394 cryptographically derived from it. Another derived seed is passed to
395 the OS. The latter seed is then credited to the kernel's entropy pool
396 very early during userspace initialization (from PID 1). This allows
397 systems to boot up with a fully initialized kernel entropy pool from
398 earliest boot on, and thus entirely removes all entropy pool
399 initialization delays from systems using systemd-boot. Special care
400 is taken to ensure different seeds are derived on system images
401 replicated to multiple systems. "bootctl status" will show whether
402 a seed was received from the boot loader.
404 * bootctl gained two new verbs:
406 - "bootctl random-seed" will generate the file in ESP and an EFI
407 variable to allow a random seed to be passed to the OS as described
410 - "bootctl is-installed" checks whether systemd-boot is currently
413 * bootctl will warn if it detects that boot entries are misconfigured
414 (for example if the kernel image was removed without purging the
417 * A new document has been added describing systemd's use and support
418 for the kernel's entropy pool subsystem:
420 https://systemd.io/RANDOM_SEEDS
422 * When the system is hibernated the swap device to write the
423 hibernation image to is now automatically picked from all available
424 swap devices, preferring the swap device with the highest configured
425 priority over all others, and picking the device with the most free
426 space if there are multiple devices with the highest priority.
428 * /etc/crypttab support has learnt a new keyfile-timeout= per-device
429 option that permits selecting the timout how long to wait for a
430 device with an encryption key before asking for the password.
432 * IOWeight= has learnt to properly set the IO weight when using the
433 BFQ scheduler officially found in kernels 5.0+.
435 Contributions from: Aaron Barany, Adrian Bunk, Alan Jenkins, Andrej
436 Valek, Anita Zhang, Arian van Putten, Balint Reczey, Bastien Nocera,
437 Ben Boeckel, Benjamin Robin, camoz, Chen Qi, Chris Chiu, Chris Down,
438 Christian Kellner, Clinton Roy, Connor Reeder, Daniele Medri, Dan
439 Streetman, Dave Reisner, Dave Ross, David Art, David Tardon, Debarshi
440 Ray, Dominick Grift, Donald Buczek, Douglas Christman, Eric DeVolder,
441 Evgeny Vereshchagin, Feldwor, Felix Riemann, Florian Dollinger, Franck
442 Bui, Frantisek Sumsal, Franz Pletz, Hans de Goede, Iago López Galeiras,
443 Insun Pyo, Ivan Shapovalov, Iwan Timmer, Jack, Jakob Unterwurzacher,
444 Jan Klötzke, Jan Pokorný, Jan Synacek, Jeka Pats, Jérémy Rosen, Jiri
445 Pirko, Joe Lin, Joerg Behrmann, Joe Richey, Jóhann B. Guðmundsson,
446 Johannes Schmitz, Jonathan Rouleau, Jorge Niedbalski, Kai Lüke, Karel
447 Zak, Kashyap Chamarthy, Krayushkin Konstantin, Lennart Poettering,
448 Lubomir Rintel, Luca Boccassi, Luís Ferreira, Marc-André Lureau, Markus
449 Felten, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Prokop,
450 Michael Stapelberg, Michael Zhivich, Michal Koutný, Michal Sekletar,
451 Mike Gilbert, Milan Broz, mpe85, Network Silence, Oliver Harley,
452 pan93412, Paul Menzel, pEJipE, Peter A. Bigot, Philip Withnall, Piotr
453 Drąg, Rafael Fontenelle, Roberto Santalla, root, RussianNeuroMancer,
454 Sebastian Jennen, shinygold, Shreyas Behera, Simon Schricker, Susant
455 Sahani, Thadeu Lima de Souza Cascardo, Theo Ouzhinski, Thiebaud
456 Weksteen, Thomas Haller, Thomas Weißschuh, Tomas Mraz, Topi Miettinen,
457 ven, Wieland Hoffmann, Xi Ruoyao, Yuri Chornoivan, Yu Watanabe, Zach
458 Smith, Zbigniew Jędrzejewski-Szmek, Zhang Xianwei
460 – Somewhere, SOME-TI-ME
464 * In .link files, MACAddressPolicy=persistent (the default) is changed
465 to cover more devices. For devices like bridges, tun, tap, bond, and
466 similar interfaces that do not have other identifying information,
467 the interface name is used as the basis for persistent seed for MAC
468 and IPv4LL addresses. The way that devices that were handled
469 previously is not changed, and this change is about covering more
470 devices then previously by the "persistent" policy.
472 MACAddressPolicy=random may be used to force randomized MACs and
473 IPv4LL addresses for a device if desired.
475 Hint: the log output from udev (at debug level) was enhanced to
476 clarify what policy is followed and which attributes are used.
477 `SYSTEMD_LOG_LEVEL=debug udevadm test-builtin net_setup_link /sys/class/net/<name>`
478 may be used to view this.
480 Hint: if a bridge interface is created without any slaves, and gains
481 a slave later, then now the bridge does not inherit slave's MAC.
482 To inherit slave's MAC, for example, create the following file:
484 # /etc/systemd/network/98-bridge-inherit-mac.link
489 MACAddressPolicy=none
492 * The .device units generated by systemd-fstab-generator and other
493 generators do not automatically pull in the corresponding .mount unit
494 as a Wants= dependency. This means that simply plugging in the device
495 will not cause the mount unit to be started automatically. But please
496 note that the mount unit may be started for other reasons, in
497 particular if it is part of local-fs.target, and any unit which
498 (transitively) depends on local-fs.target is started.
500 * networkctl list/status/lldp now accept globbing wildcards for network
501 interface names to match against all existing interfaces.
503 * The $PIDFILE environment variable is set to point the absolute path
504 configured with PIDFile= for processes of that service.
506 * The fallback DNS server list was augmented with Cloudflare public DNS
507 servers. Use `-Ddns-servers=` to set a different fallback.
509 * A new special target usb-gadget.target will be started automatically
510 when a USB Device Controller is detected (which means that the system
511 is a USB peripheral).
513 * A new unit setting CPUQuotaPeriodSec= assigns the time period
514 relatively to which the CPU time quota specified by CPUQuota= is
517 * A new unit setting ProtectHostname= may be used to prevent services
518 from modifying hostname information (even if they otherwise would
519 have privileges to do so).
521 * A new unit setting NetworkNamespacePath= may be used to specify a
522 namespace for service or socket units through a path referring to a
523 Linux network namespace pseudo-file.
525 * The PrivateNetwork= setting and JoinsNamespaceOf= dependencies now
526 have an effect on .socket units: when used the listening socket is
527 created within the configured network namespace instead of the host
530 * ExecStart= command lines in unit files may now be prefixed with ':'
531 in which case environment variable substitution is
532 disabled. (Supported for the other ExecXYZ= settings, too.)
534 * .timer units gained two new boolean settings OnClockChange= and
535 OnTimezoneChange= which may be used to also trigger a unit when the
536 system clock is changed or the local timezone is
537 modified. systemd-run has been updated to make these options easily
538 accessible from the command line for transient timers.
540 * Two new conditions for units have been added: ConditionMemory= may be
541 used to conditionalize a unit based on installed system
542 RAM. ConditionCPUs= may be used to conditionalize a unit based on
545 * The @default system call filter group understood by SystemCallFilter=
546 has been updated to include the new rseq() system call introduced in
549 * A new time-set.target has been added that indicates that the system
550 time has been set from a local source (possibly imprecise). The
551 existing time-sync.target is stronger and indicates that the time has
552 been synchronized with a precise external source. Services where
553 approximate time is sufficient should use the new target.
555 * "systemctl start" (and related commands) learnt a new
556 --show-transaction option. If specified brief information about all
557 jobs queued because of the requested operation is shown.
559 * systemd-networkd recognizes a new operation state 'enslaved', used
560 (instead of 'degraded' or 'carrier') for interfaces which form a
561 bridge, bond, or similar, and an new 'degraded-carrier' operational
562 state used for the bond or bridge master interface when one of the
563 enslaved devices is not operational.
565 * .network files learnt the new IgnoreCarrierLoss= option for leaving
566 networks configured even if the carrier is lost.
568 * The RequiredForOnline= setting in .network files may now specify a
569 minimum operational state required for the interface to be considered
570 "online" by systemd-networkd-wait-online. Related to this
571 systemd-networkd-wait-online gained a new option --operational-state=
572 to configure the same, and its --interface= option was updated to
573 optionally also take an operational state specific for an interface.
575 * systemd-networkd-wait-online gained a new setting --any for waiting
576 for only one of the requested interfaces instead of all of them.
578 * systemd-networkd now implements L2TP tunnels.
580 * Two new .network settings UseAutonomousPrefix= and UseOnLinkPrefix=
581 may be used to cause autonomous and onlink prefixes received in IPv6
582 Router Advertisements to be ignored.
584 * New MulticastFlood=, NeighborSuppression=, and Learning= .network
585 file settings may be used to tweak bridge behaviour.
587 * The new TripleSampling= option in .network files may be used to
588 configure CAN triple sampling.
590 * A new .netdev settings PrivateKeyFile= and PresharedKeyFile= may be
591 used to point to private or preshared key for a WireGuard interface.
593 * /etc/crypttab now supports the same-cpu-crypt and
594 submit-from-crypt-cpus options to tweak encryption work scheduling
597 * systemd-tmpfiles will now take a BSD file lock before operating on a
598 contents of directory. This may be used to temporarily exclude
599 directories from aging by taking the same lock (useful for example
600 when extracting a tarball into /tmp or /var/tmp as a privileged user,
601 which might create files with really old timestamps, which
602 nevertheless should not be deleted). For further details, see:
604 https://systemd.io/TEMPORARY_DIRECTORIES
606 * systemd-tmpfiles' h line type gained support for the
607 FS_PROJINHERIT_FL ('P') file attribute (introduced in kernel 4.5),
608 controlling project quota inheritance.
610 * sd-boot and bootctl now implement support for an Extended Boot Loader
611 (XBOOTLDR) partition, that is intended to be mounted to /boot, in
612 addition to the ESP partition mounted to /efi or /boot/efi.
613 Configuration file fragments, kernels, initrds and other EFI images
614 to boot will be loaded from both the ESP and XBOOTLDR partitions.
615 The XBOOTLDR partition was previously described by the Boot Loader
616 Specification, but implementation was missing in sd-boot. Support for
617 this concept allows using the sd-boot boot loader in more
618 conservative scenarios where the boot loader itself is placed in the
619 ESP but the kernels to boot (and their metadata) in a separate
622 * A system may now be booted with systemd.volatile=overlay on the
623 kernel command line, which causes the root file system to be set up
624 an overlayfs mount combining the root-only root directory with a
625 writable tmpfs. In this setup, the underlying root device is not
626 modified, and any changes are lost at reboot.
628 * Similar, systemd-nspawn can now boot containers with a volatile
629 overlayfs root with the new --volatile=overlay switch.
631 * systemd-nspawn can now consume OCI runtime bundles using a new
632 --oci-bundle= option. This implementation is fully usable, with most
633 features in the specification implemented, but since this a lot of
634 new code and functionality, this feature should most likely not
635 be used in production yet.
637 * systemd-nspawn now supports various options described by the OCI
638 runtime specification on the command-line and in .nspawn files:
639 --inaccessible=/Inaccessible= may be used to mask parts of the file
640 system tree, --console=/--pipe may be used to configure how standard
641 input, output, and error are set up.
643 * busctl learned the `emit` verb to generate D-Bus signals.
645 * systemd-analyze cat-config may be used to gather and display
646 configuration spread over multiple files, for example system and user
647 presets, tmpfiles.d, sysusers.d, udev rules, etc.
649 * systemd-analyze calendar now takes an optional new parameter
650 --iterations= which may be used to show a maximum number of iterations
651 the specified expression will elapse next.
653 * The sd-bus C API gained support for naming method parameters in the
656 * systemd-logind gained D-Bus APIs to specify the "reboot parameter"
657 the reboot() system call expects.
659 * journalctl learnt a new --cursor-file= option that points to a file
660 from which a cursor should be loaded in the beginning and to which
661 the updated cursor should be stored at the end.
663 * ACRN hypervisor and Windows Subsystem for Linux (WSL) are now
664 detected by systemd-detect-virt (and may also be used in
665 ConditionVirtualization=).
667 * The behaviour of systemd-logind may now be modified with environment
668 variables $SYSTEMD_REBOOT_TO_FIRMWARE_SETUP,
669 $SYSTEMD_REBOOT_TO_BOOT_LOADER_MENU, and
670 $SYSTEMD_REBOOT_TO_BOOT_LOADER_ENTRY. They cause logind to either
671 skip the relevant operation completely (when set to false), or to
672 create a flag file in /run/systemd (when set to true), instead of
673 actually commencing the real operation when requested. The presence
674 of /run/systemd/reboot-to-firmware-setup,
675 /run/systemd/reboot-to-boot-loader-menu, and
676 /run/systemd/reboot-to-boot-loader-entry, may be used by alternative
677 boot loader implementations to replace some steps logind performs
678 during reboot with their own operations.
680 * systemctl can be used to request a reboot into the boot loader menu
681 or a specific boot loader entry with the new --boot-load-menu= and
682 --boot-loader-entry= options to a reboot command. (This requires a
683 boot loader that supports this, for example sd-boot.)
685 * kernel-install will no longer unconditionally create the output
686 directory (e.g. /efi/<machine-id>/<kernel-version>) for boot loader
687 snippets, but will do only if the machine-specific parent directory
688 (i.e. /efi/<machine-id>/) already exists. bootctl has been modified
689 to create this parent directory during sd-boot installation.
691 This makes it easier to use kernel-install with plugins which support
692 a different layout of the bootloader partitions (for example grub2).
694 * During package installation (with `ninja install`), we would create
695 symlinks for getty@tty1.service, systemd-networkd.service,
696 systemd-networkd.socket, systemd-resolved.service,
697 remote-cryptsetup.target, remote-fs.target,
698 systemd-networkd-wait-online.service, and systemd-timesyncd.service
699 in /etc, as if `systemctl enable` was called for those units, to make
700 the system usable immediately after installation. Now this is not
701 done anymore, and instead calling `systemctl preset-all` is
702 recommended after the first installation of systemd.
704 * A new boolean sandboxing option RestrictSUIDSGID= has been added that
705 is built on seccomp. When turned on creation of SUID/SGID files is
708 * The NoNewPrivileges= and the new RestrictSUIDSGID= options are now
709 implied if DynamicUser= is turned on for a service. This hardens
710 these services, so that they neither can benefit from nor create
711 SUID/SGID executables. This is a minor compatibility breakage, given
712 that when DynamicUser= was first introduced SUID/SGID behaviour was
713 unaffected. However, the security benefit of these two options is
714 substantial, and the setting is still relatively new, hence we opted
715 to make it mandatory for services with dynamic users.
717 Contributions from: Adam Jackson, Alexander Tsoy, Andrey Yashkin,
718 Andrzej Pietrasiewicz, Anita Zhang, Balint Reczey, Beniamino Galvani,
719 Ben Iofel, Benjamin Berg, Benjamin Dahlhoff, Chris, Chris Morin,
720 Christopher Wong, Claudius Ellsel, Clemens Gruber, dana, Daniel Black,
721 Davide Cavalca, David Michael, David Rheinsberg, emersion, Evgeny
722 Vereshchagin, Filipe Brandenburger, Franck Bui, Frantisek Sumsal,
723 Giacinto Cifelli, Hans de Goede, Hugo Kindel, Ignat Korchagin, Insun
724 Pyo, Jan Engelhardt, Jonas Dorel, Jonathan Lebon, Jonathon Kowalski,
725 Jörg Sommer, Jörg Thalheim, Jussi Pakkanen, Kai-Heng Feng, Lennart
726 Poettering, Lubomir Rintel, Luís Ferreira, Martin Pitt, Matthias
727 Klumpp, Michael Biebl, Michael Niewöhner, Michael Olbrich, Michal
728 Sekletar, Mike Lothian, Paul Menzel, Piotr Drąg, Riccardo Schirone,
729 Robin Elvedi, Roman Kulikov, Ronald Tschalär, Ross Burton, Ryan
730 Gonzalez, Sebastian Krzyszkowiak, Stephane Chazelas, StKob, Susant
731 Sahani, Sylvain Plantefève, Szabolcs Fruhwald, Taro Yamada, Theo
732 Ouzhinski, Thomas Haller, Tobias Jungel, Tom Yan, Tony Asleson, Topi
733 Miettinen, unixsysadmin, Van Laser, Vesa Jääskeläinen, Yu, Li-Yu,
734 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
740 * The default locale can now be configured at compile time. Otherwise,
741 a suitable default will be selected automatically (one of C.UTF-8,
744 * The version string shown by systemd and other tools now includes the
745 git commit hash when built from git. An override may be specified
746 during compilation, which is intended to be used by distributions to
747 include the package release information.
749 * systemd-cat can now filter standard input and standard error streams
750 for different syslog priorities using the new --stderr-priority=
753 * systemd-journald and systemd-journal-remote reject entries which
754 contain too many fields (CVE-2018-16865) and set limits on the
755 process' command line length (CVE-2018-16864).
757 * $DBUS_SESSION_BUS_ADDRESS environment variable is set by pam_systemd
760 * A new network device NamePolicy "keep" is implemented for link files,
761 and used by default in 99-default.link (the fallback configuration
762 provided by systemd). With this policy, if the network device name
763 was already set by userspace, the device will not be renamed again.
764 This matches the naming scheme that was implemented before
765 systemd-240. If naming-scheme < 240 is specified, the "keep" policy
766 is also enabled by default, even if not specified. Effectively, this
767 means that if naming-scheme >= 240 is specified, network devices will
768 be renamed according to the configuration, even if they have been
769 renamed already, if "keep" is not specified as the naming policy in
770 the .link file. The 99-default.link file provided by systemd includes
771 "keep" for backwards compatibility, but it is recommended for user
772 installed .link files to *not* include it.
774 The "kernel" policy, which keeps kernel names declared to be
775 "persistent", now works again as documented.
777 * kernel-install script now optionally takes the paths to one or more
778 initrd files, and passes them to all plugins.
780 * The mincore() system call has been dropped from the @system-service
781 system call filter group, as it is pretty exotic and may potentially
782 used for side-channel attacks.
784 * -fPIE is dropped from compiler and linker options. Please specify
785 -Db_pie=true option to meson to build position-independent
786 executables. Note that the meson option is supported since meson-0.49.
788 * The fs.protected_regular and fs.protected_fifos sysctls, which were
789 added in Linux 4.19 to make some data spoofing attacks harder, are
790 now enabled by default. While this will hopefully improve the
791 security of most installations, it is technically a backwards
792 incompatible change; to disable these sysctls again, place the
793 following lines in /etc/sysctl.d/60-protected.conf or a similar file:
795 fs.protected_regular = 0
796 fs.protected_fifos = 0
798 Note that the similar hardlink and symlink protection has been
799 enabled since v199, and may be disabled likewise.
801 * The files read from the EnvironmentFile= setting in unit files now
802 parse backslashes inside quotes literally, matching the behaviour of
805 * udevadm trigger, udevadm control, udevadm settle and udevadm monitor
806 now automatically become NOPs when run in a chroot() environment.
808 * The tmpfiles.d/ "C" line type will now copy directory trees not only
809 when the destination is so far missing, but also if it already exists
810 as a directory and is empty. This is useful to cater for systems
811 where directory trees are put together from multiple separate mount
812 points but otherwise empty.
814 * A new function sd_bus_close_unref() (and the associated
815 sd_bus_close_unrefp()) has been added to libsystemd, that combines
816 sd_bus_close() and sd_bus_unref() in one.
818 * udevadm control learnt a new option for --ping for testing whether a
819 systemd-udevd instance is running and reacting.
821 * udevadm trigger learnt a new option for --wait-daemon for waiting
822 systemd-udevd daemon to be initialized.
824 Contributions from: Aaron Plattner, Alberts Muktupāvels, Alex Mayer,
825 Ayman Bagabas, Beniamino Galvani, Burt P, Chris Down, Chris Lamb, Chris
826 Morin, Christian Hesse, Claudius Ellsel, dana, Daniel Axtens, Daniele
827 Medri, Dave Reisner, David Santamaría Rogado, Diego Canuhe, Dimitri
828 John Ledkov, Evgeny Vereshchagin, Fabrice Fontaine, Filipe
829 Brandenburger, Franck Bui, Frantisek Sumsal, govwin, Hans de Goede,
830 James Hilliard, Jan Engelhardt, Jani Uusitalo, Jan Janssen, Jan
831 Synacek, Jonathan McDowell, Jonathan Roemer, Jonathon Kowalski, Joost
832 Heitbrink, Jörg Thalheim, Lance, Lennart Poettering, Louis Taylor,
833 Lucas Werkmeister, Mantas Mikulėnas, Marc-Antoine Perennou,
834 marvelousblack, Michael Biebl, Michael Sloan, Michal Sekletar, Mike
835 Auty, Mike Gilbert, Mikhail Kasimov, Neil Brown, Niklas Hambüchen,
836 Patrick Williams, Paul Seyfert, Peter Hutterer, Philip Withnall, Roger
837 James, Ronnie P. Thomas, Ryan Gonzalez, Sam Morris, Stephan Edel,
838 Stephan Gerhold, Susant Sahani, Taro Yamada, Thomas Haller, Topi
839 Miettinen, YiFei Zhu, YmrDtnJu, YunQiang Su, Yu Watanabe, Zbigniew
840 Jędrzejewski-Szmek, zsergeant77, Дамјан Георгиевски
846 * NoNewPrivileges=yes has been set for all long-running services
847 implemented by systemd. Previously, this was problematic due to
848 SELinux (as this would also prohibit the transition from PID1's label
849 to the service's label). This restriction has since been lifted, but
850 an SELinux policy update is required.
851 (See e.g. https://github.com/fedora-selinux/selinux-policy/pull/234.)
853 * DynamicUser=yes is dropped from systemd-networkd.service,
854 systemd-resolved.service and systemd-timesyncd.service, which was
855 enabled in v239 for systemd-networkd.service and systemd-resolved.service,
856 and since v236 for systemd-timesyncd.service. The users and groups
857 systemd-network, systemd-resolve and systemd-timesync are created
858 by systemd-sysusers again. Distributors or system administrators
859 may need to create these users and groups if they not exist (or need
860 to re-enable DynamicUser= for those units) while upgrading systemd.
861 Also, the clock file for systemd-timesyncd may need to move from
862 /var/lib/private/systemd/timesync/clock to /var/lib/systemd/timesync/clock.
864 * When unit files are loaded from disk, previously systemd would
865 sometimes (depending on the unit loading order) load units from the
866 target path of symlinks in .wants/ or .requires/ directories of other
867 units. This meant that unit could be loaded from different paths
868 depending on whether the unit was requested explicitly or as a
869 dependency of another unit, not honouring the priority of directories
870 in search path. It also meant that it was possible to successfully
871 load and start units which are not found in the unit search path, as
872 long as they were requested as a dependency and linked to from
873 .wants/ or .requires/. The target paths of those symlinks are not
874 used for loading units anymore and the unit file must be found in
877 * A new service type has been added: Type=exec. It's very similar to
878 Type=simple but ensures the service manager will wait for both fork()
879 and execve() of the main service binary to complete before proceeding
880 with follow-up units. This is primarily useful so that the manager
881 propagates any errors in the preparation phase of service execution
882 back to the job that requested the unit to be started. For example,
883 consider a service that has ExecStart= set to a file system binary
884 that doesn't exist. With Type=simple starting the unit would be
885 considered instantly successful, as only fork() has to complete
886 successfully and the manager does not wait for execve(), and hence
887 its failure is seen "too late". With the new Type=exec service type
888 starting the unit will fail, as the manager will wait for the
889 execve() and notice its failure, which is then propagated back to the
892 NOTE: with the next release 241 of systemd we intend to change the
893 systemd-run tool to default to Type=exec for transient services
894 started by it. This should be mostly safe, but in specific corner
895 cases might result in problems, as the systemd-run tool will then
896 block on NSS calls (such as user name look-ups due to User=) done
897 between the fork() and execve(), which under specific circumstances
898 might cause problems. It is recommended to specify "-p Type=simple"
899 explicitly in the few cases where this applies. For regular,
900 non-transient services (i.e. those defined with unit files on disk)
901 we will continue to default to Type=simple.
903 * The Linux kernel's current default RLIMIT_NOFILE resource limit for
904 userspace processes is set to 1024 (soft) and 4096
905 (hard). Previously, systemd passed this on unmodified to all
906 processes it forked off. With this systemd release the hard limit
907 systemd passes on is increased to 512K, overriding the kernel's
908 defaults and substantially increasing the number of simultaneous file
909 descriptors unprivileged userspace processes can allocate. Note that
910 the soft limit remains at 1024 for compatibility reasons: the
911 traditional UNIX select() call cannot deal with file descriptors >=
912 1024 and increasing the soft limit globally might thus result in
913 programs unexpectedly allocating a high file descriptor and thus
914 failing abnormally when attempting to use it with select() (of
915 course, programs shouldn't use select() anymore, and prefer
916 poll()/epoll, but the call unfortunately remains undeservedly popular
917 at this time). This change reflects the fact that file descriptor
918 handling in the Linux kernel has been optimized in more recent
919 kernels and allocating large numbers of them should be much cheaper
920 both in memory and in performance than it used to be. Programs that
921 want to take benefit of the increased limit have to "opt-in" into
922 high file descriptors explicitly by raising their soft limit. Of
923 course, when they do that they must acknowledge that they cannot use
924 select() anymore (and neither can any shared library they use — or
925 any shared library used by any shared library they use and so on).
926 Which default hard limit is most appropriate is of course hard to
927 decide. However, given reports that ~300K file descriptors are used
928 in real-life applications we believe 512K is sufficiently high as new
929 default for now. Note that there are also reports that using very
930 high hard limits (e.g. 1G) is problematic: some software allocates
931 large arrays with one element for each potential file descriptor
932 (Java, …) — a high hard limit thus triggers excessively large memory
933 allocations in these applications. Hopefully, the new default of 512K
934 is a good middle ground: higher than what real-life applications
935 currently need, and low enough for avoid triggering excessively large
936 allocations in problematic software. (And yes, somebody should fix
939 * The fs.nr_open and fs.file-max sysctls are now automatically bumped
940 to the highest possible values, as separate accounting of file
941 descriptors is no longer necessary, as memcg tracks them correctly as
942 part of the memory accounting anyway. Thus, from the four limits on
943 file descriptors currently enforced (fs.file-max, fs.nr_open,
944 RLIMIT_NOFILE hard, RLIMIT_NOFILE soft) we turn off the first two,
945 and keep only the latter two. A set of build-time options
946 (-Dbump-proc-sys-fs-file-max=false and -Dbump-proc-sys-fs-nr-open=false)
947 has been added to revert this change in behaviour, which might be
948 an option for systems that turn off memcg in the kernel.
950 * When no /etc/locale.conf file exists (and hence no locale settings
951 are in place), systemd will now use the "C.UTF-8" locale by default,
952 and set LANG= to it. This locale is supported by various
953 distributions including Fedora, with clear indications that upstream
954 glibc is going to make it available too. This locale enables UTF-8
955 mode by default, which appears appropriate for 2018.
957 * The "net.ipv4.conf.all.rp_filter" sysctl will now be set to 2 by
958 default. This effectively switches the RFC3704 Reverse Path filtering
959 from Strict mode to Loose mode. This is more appropriate for hosts
960 that have multiple links with routes to the same networks (e.g.
961 a client with a Wi-Fi and Ethernet both connected to the internet).
963 Consult the kernel documentation for details on this sysctl:
964 https://www.kernel.org/doc/Documentation/networking/ip-sysctl.txt
966 * CPUAccounting=yes no longer enables the CPU controller when using
967 kernel 4.15+ and the unified cgroup hierarchy, as required accounting
968 statistics are now provided independently from the CPU controller.
970 * Support for disabling a particular cgroup controller within a sub-tree
971 has been added through the DisableControllers= directive.
973 * cgroup_no_v1=all on the kernel command line now also implies
974 using the unified cgroup hierarchy, unless one explicitly passes
975 systemd.unified_cgroup_hierarchy=0 on the kernel command line.
977 * The new "MemoryMin=" unit file property may now be used to set the
978 memory usage protection limit of processes invoked by the unit. This
979 controls the cgroup v2 memory.min attribute. Similarly, the new
980 "IODeviceLatencyTargetSec=" property has been added, wrapping the new
981 cgroup v2 io.latency cgroup property for configuring per-service I/O
984 * systemd now supports the cgroup v2 devices BPF logic, as counterpart
985 to the cgroup v1 "devices" cgroup controller.
987 * systemd-escape now is able to combine --unescape with --template. It
988 also learnt a new option --instance for extracting and unescaping the
989 instance part of a unit name.
991 * sd-bus now provides the sd_bus_message_readv() which is similar to
992 sd_bus_message_read() but takes a va_list object. The pair
993 sd_bus_set_method_call_timeout() and sd_bus_get_method_call_timeout()
994 has been added for configuring the default method call timeout to
995 use. sd_bus_error_move() may be used to efficiently move the contents
996 from one sd_bus_error structure to another, invalidating the
997 source. sd_bus_set_close_on_exit() and sd_bus_get_close_on_exit() may
998 be used to control whether a bus connection object is automatically
999 flushed when an sd-event loop is exited.
1001 * When processing classic BSD syslog log messages, journald will now
1002 save the original time-stamp string supplied in the new
1003 SYSLOG_TIMESTAMP= journal field. This permits consumers to
1004 reconstruct the original BSD syslog message more correctly.
1006 * StandardOutput=/StandardError= in service files gained support for
1007 new "append:…" parameters, for connecting STDOUT/STDERR of a service
1008 to a file, and appending to it.
1010 * The signal to use as last step of killing of unit processes is now
1011 configurable. Previously it was hard-coded to SIGKILL, which may now
1012 be overridden with the new KillSignal= setting. Note that this is the
1013 signal used when regular termination (i.e. SIGTERM) does not suffice.
1014 Similarly, the signal used when aborting a program in case of a
1015 watchdog timeout may now be configured too (WatchdogSignal=).
1017 * The XDG_SESSION_DESKTOP environment variable may now be configured in
1018 the pam_systemd argument line, using the new desktop= switch. This is
1019 useful to initialize it properly from a display manager without
1020 having to touch C code.
1022 * Most configuration options that previously accepted percentage values
1023 now also accept permille values with the '‰' suffix (instead of '%').
1025 * systemd-resolved may now optionally use OpenSSL instead of GnuTLS for
1028 * systemd-resolved's configuration file resolved.conf gained a new
1029 option ReadEtcHosts= which may be used to turn off processing and
1030 honoring /etc/hosts entries.
1032 * The "--wait" switch may now be passed to "systemctl
1033 is-system-running", in which case the tool will synchronously wait
1034 until the system finished start-up.
1036 * hostnamed gained a new bus call to determine the DMI product UUID.
1038 * On x86-64 systemd will now prefer using the RDRAND processor
1039 instruction over /dev/urandom whenever it requires randomness that
1040 neither has to be crypto-grade nor should be reproducible. This
1041 should substantially reduce the amount of entropy systemd requests
1042 from the kernel during initialization on such systems, though not
1043 reduce it to zero. (Why not zero? systemd still needs to allocate
1044 UUIDs and such uniquely, which require high-quality randomness.)
1046 * networkd gained support for Foo-Over-UDP, ERSPAN and ISATAP
1047 tunnels. It also gained a new option ForceDHCPv6PDOtherInformation=
1048 for forcing the "Other Information" bit in IPv6 RA messages. The
1049 bonding logic gained four new options AdActorSystemPriority=,
1050 AdUserPortKey=, AdActorSystem= for configuring various 802.3ad
1051 aspects, and DynamicTransmitLoadBalancing= for enabling dynamic
1052 shuffling of flows. The tunnel logic gained a new
1053 IPv6RapidDeploymentPrefix= option for configuring IPv6 Rapid
1054 Deployment. The policy rule logic gained four new options IPProtocol=,
1055 SourcePort= and DestinationPort=, InvertRule=. The bridge logic gained
1056 support for the MulticastToUnicast= option. networkd also gained
1057 support for configuring static IPv4 ARP or IPv6 neighbor entries.
1059 * .preset files (as read by 'systemctl preset') may now be used to
1060 instantiate services.
1062 * /etc/crypttab now understands the sector-size= option to configure
1063 the sector size for an encrypted partition.
1065 * Key material for encrypted disks may now be placed on a formatted
1066 medium, and referenced from /etc/crypttab by the UUID of the file
1067 system, followed by "=" suffixed by the path to the key file.
1069 * The "collect" udev component has been removed without replacement, as
1070 it is neither used nor maintained.
1072 * When the RuntimeDirectory=, StateDirectory=, CacheDirectory=,
1073 LogsDirectory=, ConfigurationDirectory= settings are used in a
1074 service the executed processes will now receive a set of environment
1075 variables containing the full paths of these directories.
1076 Specifically, RUNTIME_DIRECTORY=, STATE_DIRECTORY, CACHE_DIRECTORY,
1077 LOGS_DIRECTORY, CONFIGURATION_DIRECTORY are now set if these options
1078 are used. Note that these options may be used multiple times per
1079 service in which case the resulting paths will be concatenated and
1080 separated by colons.
1082 * Predictable interface naming has been extended to cover InfiniBand
1083 NICs. They will be exposed with an "ib" prefix.
1085 * tmpfiles.d/ line types may now be suffixed with a '-' character, in
1086 which case the respective line failing is ignored.
1088 * .link files may now be used to configure the equivalent to the
1089 "ethtool advertise" commands.
1091 * The sd-device.h and sd-hwdb.h APIs are now exported, as an
1092 alternative to libudev.h. Previously, the latter was just an internal
1093 wrapper around the former, but now these two APIs are exposed
1096 * sd-id128.h gained a new function sd_id128_get_boot_app_specific()
1097 which calculates an app-specific boot ID similar to how
1098 sd_id128_get_machine_app_specific() generates an app-specific machine
1101 * A new tool systemd-id128 has been added that can be used to determine
1102 and generate various 128bit IDs.
1104 * /etc/os-release gained two new standardized fields DOCUMENTATION_URL=
1107 * systemd-hibernate-resume-generator will now honor the "noresume"
1108 kernel command line option, in which case it will bypass resuming
1109 from any hibernated image.
1111 * The systemd-sleep.conf configuration file gained new options
1112 AllowSuspend=, AllowHibernation=, AllowSuspendThenHibernate=,
1113 AllowHybridSleep= for prohibiting specific sleep modes even if the
1114 kernel exports them.
1116 * portablectl is now officially supported and has thus moved to
1119 * bootctl learnt the two new commands "set-default" and "set-oneshot"
1120 for setting the default boot loader item to boot to (either
1121 persistently or only for the next boot). This is currently only
1122 compatible with sd-boot, but may be implemented on other boot loaders
1123 too, that follow the boot loader interface. The updated interface is
1124 now documented here:
1126 https://systemd.io/BOOT_LOADER_INTERFACE
1128 * A new kernel command line option systemd.early_core_pattern= is now
1129 understood which may be used to influence the core_pattern PID 1
1130 installs during early boot.
1132 * busctl learnt two new options -j and --json= for outputting method
1133 call replies, properties and monitoring output in JSON.
1135 * journalctl's JSON output now supports simple ANSI coloring as well as
1136 a new "json-seq" mode for generating RFC7464 output.
1138 * Unit files now support the %g/%G specifiers that resolve to the UNIX
1139 group/GID of the service manager runs as, similar to the existing
1140 %u/%U specifiers that resolve to the UNIX user/UID.
1142 * systemd-logind learnt a new global configuration option
1143 UserStopDelaySec= that may be set in logind.conf. It specifies how
1144 long the systemd --user instance shall remain started after a user
1145 logs out. This is useful to speed up repetitive re-connections of the
1146 same user, as it means the user's service manager doesn't have to be
1147 stopped/restarted on each iteration, but can be reused between
1148 subsequent options. This setting defaults to 10s. systemd-logind also
1149 exports two new properties on its Manager D-Bus objects indicating
1150 whether the system's lid is currently closed, and whether the system
1153 * systemd gained support for a generic boot counting logic, which
1154 generically permits automatic reverting to older boot loader entries
1155 if newer updated ones don't work. The boot loader side is implemented
1156 in sd-boot, but is kept open for other boot loaders too. For details
1159 https://systemd.io/AUTOMATIC_BOOT_ASSESSMENT
1161 * The SuccessAction=/FailureAction= unit file settings now learnt two
1162 new parameters: "exit" and "exit-force", which result in immediate
1163 exiting of the service manager, and are only useful in systemd --user
1164 and container environments.
1166 * Unit files gained support for a pair of options
1167 FailureActionExitStatus=/SuccessActionExitStatus= for configuring the
1168 exit status to use as service manager exit status when
1169 SuccessAction=/FailureAction= is set to exit or exit-force.
1171 * A pair of LogRateLimitIntervalSec=/LogRateLimitBurst= per-service
1172 options may now be used to configure the log rate limiting applied by
1173 journald per-service.
1175 * systemd-analyze gained a new verb "timespan" for parsing and
1176 normalizing time span values (i.e. strings like "5min 7s 8us").
1178 * systemd-analyze also gained a new verb "security" for analyzing the
1179 security and sand-boxing settings of services in order to determine an
1180 "exposure level" for them, indicating whether a service would benefit
1181 from more sand-boxing options turned on for them.
1183 * "systemd-analyze syscall-filter" will now also show system calls
1184 supported by the local kernel but not included in any of the defined
1187 * .nspawn files now understand the Ephemeral= setting, matching the
1188 --ephemeral command line switch.
1190 * sd-event gained the new APIs sd_event_source_get_floating() and
1191 sd_event_source_set_floating() for controlling whether a specific
1192 event source is "floating", i.e. destroyed along with the even loop
1195 * Unit objects on D-Bus gained a new "Refs" property that lists all
1196 clients that currently have a reference on the unit (to ensure it is
1199 * The JoinControllers= option in system.conf is no longer supported, as
1200 it didn't work correctly, is hard to support properly, is legacy (as
1201 the concept only exists on cgroup v1) and apparently wasn't used.
1203 * Journal messages that are generated whenever a unit enters the failed
1204 state are now tagged with a unique MESSAGE_ID. Similarly, messages
1205 generated whenever a service process exits are now made recognizable,
1206 too. A tagged message is also emitted whenever a unit enters the
1207 "dead" state on success.
1209 * systemd-run gained a new switch --working-directory= for configuring
1210 the working directory of the service to start. A shortcut -d is
1211 equivalent, setting the working directory of the service to the
1212 current working directory of the invoking program. The new --shell
1213 (or just -S) option has been added for invoking the $SHELL of the
1214 caller as a service, and implies --pty --same-dir --wait --collect
1215 --service-type=exec. Or in other words, "systemd-run -S" is now the
1216 quickest way to quickly get an interactive in a fully clean and
1217 well-defined system service context.
1219 * machinectl gained a new verb "import-fs" for importing an OS tree
1220 from a directory. Moreover, when a directory or tarball is imported
1221 and single top-level directory found with the OS itself below the OS
1222 tree is automatically mangled and moved one level up.
1224 * systemd-importd will no longer set up an implicit btrfs loop-back
1225 file system on /var/lib/machines. If one is already set up, it will
1226 continue to be used.
1228 * A new generator "systemd-run-generator" has been added. It will
1229 synthesize a unit from one or more program command lines included in
1230 the kernel command line. This is very useful in container managers
1233 # systemd-nspawn -i someimage.raw -b systemd.run='"some command line"'
1235 This will run "systemd-nspawn" on an image, invoke the specified
1236 command line and immediately shut down the container again, returning
1237 the command line's exit code.
1239 * The block device locking logic is now documented:
1241 https://systemd.io/BLOCK_DEVICE_LOCKING
1243 * loginctl and machinectl now optionally output the various tables in
1244 JSON using the --output= switch. It is our intention to add similar
1245 support to systemctl and all other commands.
1247 * udevadm's query and trigger verb now optionally take a .device unit
1250 * systemd-udevd's network naming logic now understands a new
1251 net.naming-scheme= kernel command line switch, which may be used to
1252 pick a specific version of the naming scheme. This helps stabilizing
1253 interface names even as systemd/udev are updated and the naming logic
1256 * sd-id128.h learnt two new auxiliary helpers: sd_id128_is_allf() and
1257 SD_ID128_ALLF to test if a 128bit ID is set to all 0xFF bytes, and to
1258 initialize one to all 0xFF.
1260 * After loading the SELinux policy systemd will now recursively relabel
1261 all files and directories listed in
1262 /run/systemd/relabel-extra.d/*.relabel (which should be simple
1263 newline separated lists of paths) in addition to the ones it already
1264 implicitly relabels in /run, /dev and /sys. After the relabelling is
1265 completed the *.relabel files (and /run/systemd/relabel-extra.d/) are
1266 removed. This is useful to permit initrds (i.e. code running before
1267 the SELinux policy is in effect) to generate files in the host
1268 filesystem safely and ensure that the correct label is applied during
1269 the transition to the host OS.
1271 * KERNEL API BREAKAGE: Linux kernel 4.18 changed behaviour regarding
1272 mknod() handling in user namespaces. Previously mknod() would always
1273 fail with EPERM in user namespaces. Since 4.18 mknod() will succeed
1274 but device nodes generated that way cannot be opened, and attempts to
1275 open them result in EPERM. This breaks the "graceful fallback" logic
1276 in systemd's PrivateDevices= sand-boxing option. This option is
1277 implemented defensively, so that when systemd detects it runs in a
1278 restricted environment (such as a user namespace, or an environment
1279 where mknod() is blocked through seccomp or absence of CAP_SYS_MKNOD)
1280 where device nodes cannot be created the effect of PrivateDevices= is
1281 bypassed (following the logic that 2nd-level sand-boxing is not
1282 essential if the system systemd runs in is itself already sand-boxed
1283 as a whole). This logic breaks with 4.18 in container managers where
1284 user namespacing is used: suddenly PrivateDevices= succeeds setting
1285 up a private /dev/ file system containing devices nodes — but when
1286 these are opened they don't work.
1288 At this point it is recommended that container managers utilizing
1289 user namespaces that intend to run systemd in the payload explicitly
1290 block mknod() with seccomp or similar, so that the graceful fallback
1293 We are very sorry for the breakage and the requirement to change
1294 container configurations for newer kernels. It's purely caused by an
1295 incompatible kernel change. The relevant kernel developers have been
1296 notified about this userspace breakage quickly, but they chose to
1299 * PermissionsStartOnly= setting is deprecated (but is still supported
1300 for backwards compatibility). The same functionality is provided by
1301 the more flexible "+", "!", and "!!" prefixes to ExecStart= and other
1304 * $DBUS_SESSION_BUS_ADDRESS environment variable is not set by
1305 pam_systemd anymore.
1307 * The naming scheme for network devices was changed to always rename
1308 devices, even if they were already renamed by userspace. The "kernel"
1309 policy was changed to only apply as a fallback, if no other naming
1312 * The requirements to build systemd is bumped to meson-0.46 and
1315 Contributions from: afg, Alan Jenkins, Aleksei Timofeyev, Alexander
1316 Filippov, Alexander Kurtz, Alexey Bogdanenko, Andreas Henriksson,
1317 Andrew Jorgensen, Anita Zhang, apnix-uk, Arkan49, Arseny Maslennikov,
1318 asavah, Asbjørn Apeland, aszlig, Bastien Nocera, Ben Boeckel, Benedikt
1319 Morbach, Benjamin Berg, Bruce Zhang, Carlo Caione, Cedric Viou, Chen
1320 Qi, Chris Chiu, Chris Down, Chris Morin, Christian Rebischke, Claudius
1321 Ellsel, Colin Guthrie, dana, Daniel, Daniele Medri, Daniel Kahn
1322 Gillmor, Daniel Rusek, Daniel van Vugt, Dariusz Gadomski, Dave Reisner,
1323 David Anderson, Davide Cavalca, David Leeds, David Malcolm, David
1324 Strauss, David Tardon, Dimitri John Ledkov, Dmitry Torokhov, dj-kaktus,
1325 Dongsu Park, Elias Probst, Emil Soleyman, Erik Kooistra, Ervin Peters,
1326 Evgeni Golov, Evgeny Vereshchagin, Fabrice Fontaine, Faheel Ahmad,
1327 Faizal Luthfi, Felix Yan, Filipe Brandenburger, Franck Bui, Frank
1328 Schaefer, Frantisek Sumsal, Gautier Husson, Gianluca Boiano, Giuseppe
1329 Scrivano, glitsj16, Hans de Goede, Harald Hoyer, Harry Mallon, Harshit
1330 Jain, Helmut Grohne, Henry Tung, Hui Yiqun, imayoda, Insun Pyo, Iwan
1331 Timmer, Jan Janssen, Jan Pokorný, Jan Synacek, Jason A. Donenfeld,
1332 javitoom, Jérémy Nouhaud, Jeremy Su, Jiuyang Liu, João Paulo Rechi
1333 Vita, Joe Hershberger, Joe Rayhawk, Joerg Behrmann, Joerg Steffens,
1334 Jonas Dorel, Jon Ringle, Josh Soref, Julian Andres Klode, Jun Bo Bi,
1335 Jürg Billeter, Keith Busch, Khem Raj, Kirill Marinushkin, Larry
1336 Bernstone, Lennart Poettering, Lion Yang, Li Song, Lorenz
1337 Hübschle-Schneider, Lubomir Rintel, Lucas Werkmeister, Ludwin Janvier,
1338 Lukáš Nykrýn, Luke Shumaker, mal, Marc-Antoine Perennou, Marcin
1339 Skarbek, Marco Trevisan (Treviño), Marian Cepok, Mario Hros, Marko
1340 Myllynen, Markus Grimm, Martin Pitt, Martin Sobotka, Martin Wilck,
1341 Mathieu Trudel-Lapierre, Matthew Leeds, Michael Biebl, Michael Olbrich,
1342 Michael 'pbone' Pobega, Michael Scherer, Michal Koutný, Michal
1343 Sekletar, Michal Soltys, Mike Gilbert, Mike Palmer, Muhammet Kara, Neal
1344 Gompa, Neil Brown, Network Silence, Niklas Tibbling, Nikolas Nyby,
1345 Nogisaka Sadata, Oliver Smith, Patrik Flykt, Pavel Hrdina, Paweł
1346 Szewczyk, Peter Hutterer, Piotr Drąg, Ray Strode, Reinhold Mueller,
1347 Renaud Métrich, Roman Gushchin, Ronny Chevalier, Rubén Suárez Alvarez,
1348 Ruixin Bao, RussianNeuroMancer, Ryutaroh Matsumoto, Saleem Rashid, Sam
1349 Morris, Samuel Morris, Sandy Carter, scootergrisen, Sébastien Bacher,
1350 Sergey Ptashnick, Shawn Landden, Shengyao Xue, Shih-Yuan Lee
1351 (FourDollars), Silvio Knizek, Sjoerd Simons, Stasiek Michalski, Stephen
1352 Gallagher, Steven Allen, Steve Ramage, Susant Sahani, Sven Joachim,
1353 Sylvain Plantefève, Tanu Kaskinen, Tejun Heo, Thiago Macieira, Thomas
1354 Blume, Thomas Haller, Thomas H. P. Andersen, Tim Ruffing, TJ, Tobias
1355 Jungel, Todd Walton, Tommi Rantala, Tomsod M, Tony Novak, Tore
1356 Anderson, Trevonn, Victor Laskurain, Victor Tapia, Violet Halo, Vojtech
1357 Trefny, welaq, William A. Kennington III, William Douglas, Wyatt Ward,
1358 Xiang Fan, Xi Ruoyao, Xuanwo, Yann E. Morin, YmrDtnJu, Yu Watanabe,
1359 Zbigniew Jędrzejewski-Szmek, Zhang Xianwei, Zsolt Dollenstein
1361 — Warsaw, 2018-12-21
1365 * NETWORK INTERFACE DEVICE NAMING CHANGES: systemd-udevd's "net_id"
1366 builtin will name network interfaces differently than in previous
1367 versions for virtual network interfaces created with SR-IOV and NPAR
1368 and for devices where the PCI network controller device does not have
1369 a slot number associated.
1371 SR-IOV virtual devices are now named based on the name of the parent
1372 interface, with a suffix of "v<N>", where <N> is the virtual device
1373 number. Previously those virtual devices were named as if completely
1376 The ninth and later NPAR virtual devices will be named following the
1377 scheme used for the first eight NPAR partitions. Previously those
1378 devices were not renamed and the kernel default (eth<n>) was used.
1380 "net_id" will also generate names for PCI devices where the PCI
1381 network controller device does not have an associated slot number
1382 itself, but one of its parents does. Previously those devices were
1383 not renamed and the kernel default (eth<n>) was used.
1385 * AF_INET and AF_INET6 are dropped from RestrictAddressFamilies= in
1386 systemd-logind.service. Since v235, IPAddressDeny=any has been set to
1387 the unit. So, it is expected that the default behavior of
1388 systemd-logind is not changed. However, if distribution packagers or
1389 administrators disabled or modified IPAddressDeny= setting by a
1390 drop-in config file, then it may be necessary to update the file to
1391 re-enable AF_INET and AF_INET6 to support network user name services,
1394 * When the RestrictNamespaces= unit property is specified multiple
1395 times, then the specified types are merged now. Previously, only the
1396 last assignment was used. So, if distribution packagers or
1397 administrators modified the setting by a drop-in config file, then it
1398 may be necessary to update the file.
1400 * When OnFailure= is used in combination with Restart= on a service
1401 unit, then the specified units will no longer be triggered on
1402 failures that result in restarting. Previously, the specified units
1403 would be activated each time the unit failed, even when the unit was
1404 going to be restarted automatically. This behaviour contradicted the
1405 documentation. With this release the code is adjusted to match the
1408 * systemd-tmpfiles will now print a notice whenever it encounters
1409 tmpfiles.d/ lines referencing the /var/run/ directory. It will
1410 recommend reworking them to use the /run/ directory instead (for
1411 which /var/run/ is simply a symlinked compatibility alias). This way
1412 systemd-tmpfiles can properly detect line conflicts and merge lines
1413 referencing the same file by two paths, without having to access
1416 * systemctl disable/unmask/preset/preset-all cannot be used with
1417 --runtime. Previously this was allowed, but resulted in unintuitive
1418 behaviour that wasn't useful. systemctl disable/unmask will now undo
1419 both runtime and persistent enablement/masking, i.e. it will remove
1420 any relevant symlinks both in /run and /etc.
1422 * Note that all long-running system services shipped with systemd will
1423 now default to a system call whitelist (rather than a blacklist, as
1424 before). In particular, systemd-udevd will now enforce one too. For
1425 most cases this should be safe, however downstream distributions
1426 which disabled sandboxing of systemd-udevd (specifically the
1427 MountFlags= setting), might want to disable this security feature
1428 too, as the default whitelisting will prohibit all mount, swap,
1429 reboot and clock changing operations from udev rules.
1431 * sd-boot acquired new loader configuration settings to optionally turn
1432 off Windows and MacOS boot partition discovery as well as
1433 reboot-into-firmware menu items. It is also able to pick a better
1434 screen resolution for HiDPI systems, and now provides loader
1435 configuration settings to change the resolution explicitly.
1437 * systemd-resolved now supports DNS-over-TLS. It's still
1438 turned off by default, use DNSOverTLS=opportunistic to turn it on in
1439 resolved.conf. We intend to make this the default as soon as couple
1440 of additional techniques for optimizing the initial latency caused by
1441 establishing a TLS/TCP connection are implemented.
1443 * systemd-resolved.service and systemd-networkd.service now set
1444 DynamicUser=yes. The users systemd-resolve and systemd-network are
1445 not created by systemd-sysusers anymore.
1447 NOTE: This has a chance of breaking nss-ldap and similar NSS modules
1448 that embed a network facing module into any process using getpwuid()
1449 or related call: the dynamic allocation of the user ID for
1450 systemd-resolved.service means the service manager has to check NSS
1451 if the user name is already taken when forking off the service. Since
1452 the user in the common case won't be defined in /etc/passwd the
1453 lookup is likely to trigger nss-ldap which in turn might use NSS to
1454 ask systemd-resolved for hostname lookups. This will hence result in
1455 a deadlock: a user name lookup in order to start
1456 systemd-resolved.service will result in a host name lookup for which
1457 systemd-resolved.service needs to be started already. There are
1458 multiple ways to work around this problem: pre-allocate the
1459 "systemd-resolve" user on such systems, so that nss-ldap won't be
1460 triggered; or use a different NSS package that doesn't do networking
1461 in-process but provides a local asynchronous name cache; or configure
1462 the NSS package to avoid lookups for UIDs in the range `pkg-config
1463 systemd --variable=dynamicuidmin` … `pkg-config systemd
1464 --variable=dynamicuidmax`, so that it does not consider itself
1465 authoritative for the same UID range systemd allocates dynamic users
1468 * The systemd-resolve tool has been renamed to resolvectl (it also
1469 remains available under the old name, for compatibility), and its
1470 interface is now verb-based, similar in style to the other <xyz>ctl
1471 tools, such as systemctl or loginctl.
1473 * The resolvectl/systemd-resolve tool also provides 'resolvconf'
1474 compatibility. It may be symlinked under the 'resolvconf' name, in
1475 which case it will take arguments and input compatible with the
1476 Debian and FreeBSD resolvconf tool.
1478 * Support for suspend-then-hibernate has been added, i.e. a sleep mode
1479 where the system initially suspends, and after a timeout resumes and
1482 * networkd's ClientIdentifier= now accepts a new option "duid-only". If
1483 set the client will only send a DUID as client identifier.
1485 * The nss-systemd glibc NSS module will now enumerate dynamic users and
1486 groups in effect. Previously, it could resolve UIDs/GIDs to user
1487 names/groups and vice versa, but did not support enumeration.
1489 * journald's Compress= configuration setting now optionally accepts a
1490 byte threshold value. All journal objects larger than this threshold
1491 will be compressed, smaller ones will not. Previously this threshold
1492 was not configurable and set to 512.
1494 * A new system.conf setting NoNewPrivileges= is now available which may
1495 be used to turn off acquisition of new privileges system-wide
1496 (i.e. set Linux' PR_SET_NO_NEW_PRIVS for PID 1 itself, and thus also
1497 for all its children). Note that turning this option on means setuid
1498 binaries and file system capabilities lose their special powers.
1499 While turning on this option is a big step towards a more secure
1500 system, doing so is likely to break numerous pre-existing UNIX tools,
1501 in particular su and sudo.
1503 * A new service systemd-time-sync-wait.service has been added. If
1504 enabled it will delay the time-sync.target unit at boot until time
1505 synchronization has been received from the network. This
1506 functionality is useful on systems lacking a local RTC or where it is
1507 acceptable that the boot process shall be delayed by external network
1510 * When hibernating, systemd will now inform the kernel of the image
1511 write offset, on kernels new enough to support this. This means swap
1512 files should work for hibernation now.
1514 * When loading unit files, systemd will now look for drop-in unit files
1515 extensions in additional places. Previously, for a unit file name
1516 "foo-bar-baz.service" it would look for dropin files in
1517 "foo-bar-baz.service.d/*.conf". Now, it will also look in
1518 "foo-bar-.service.d/*.conf" and "foo-.service.d/", i.e. at the
1519 service name truncated after all inner dashes. This scheme allows
1520 writing drop-ins easily that apply to a whole set of unit files at
1521 once. It's particularly useful for mount and slice units (as their
1522 naming is prefix based), but is also useful for service and other
1523 units, for packages that install multiple unit files at once,
1524 following a strict naming regime of beginning the unit file name with
1525 the package's name. Two new specifiers are now supported in unit
1526 files to match this: %j and %J are replaced by the part of the unit
1527 name following the last dash.
1529 * Unit files and other configuration files that support specifier
1530 expansion now understand another three new specifiers: %T and %V will
1531 resolve to /tmp and /var/tmp respectively, or whatever temporary
1532 directory has been set for the calling user. %E will expand to either
1533 /etc (for system units) or $XDG_CONFIG_HOME (for user units).
1535 * The ExecStart= lines of unit files are no longer required to
1536 reference absolute paths. If non-absolute paths are specified the
1537 specified binary name is searched within the service manager's
1538 built-in $PATH, which may be queried with 'systemd-path
1539 search-binaries-default'. It's generally recommended to continue to
1540 use absolute paths for all binaries specified in unit files.
1542 * Units gained a new load state "bad-setting", which is used when a
1543 unit file was loaded, but contained fatal errors which prevent it
1544 from being started (for example, a service unit has been defined
1545 lacking both ExecStart= and ExecStop= lines).
1547 * coredumpctl's "gdb" verb has been renamed to "debug", in order to
1548 support alternative debuggers, for example lldb. The old name
1549 continues to be available however, for compatibility reasons. Use the
1550 new --debugger= switch or the $SYSTEMD_DEBUGGER environment variable
1551 to pick an alternative debugger instead of the default gdb.
1553 * systemctl and the other tools will now output escape sequences that
1554 generate proper clickable hyperlinks in various terminal emulators
1555 where useful (for example, in the "systemctl status" output you can
1556 now click on the unit file name to quickly open it in the
1557 editor/viewer of your choice). Note that not all terminal emulators
1558 support this functionality yet, but many do. Unfortunately, the
1559 "less" pager doesn't support this yet, hence this functionality is
1560 currently automatically turned off when a pager is started (which
1561 happens quite often due to auto-paging). We hope to remove this
1562 limitation as soon as "less" learns these escape sequences. This new
1563 behaviour may also be turned off explicitly with the $SYSTEMD_URLIFY
1564 environment variable. For details on these escape sequences see:
1565 https://gist.github.com/egmontkob/eb114294efbcd5adb1944c9f3cb5feda
1567 * networkd's .network files now support a new IPv6MTUBytes= option for
1568 setting the MTU used by IPv6 explicitly as well as a new MTUBytes=
1569 option in the [Route] section to configure the MTU to use for
1570 specific routes. It also gained support for configuration of the DHCP
1571 "UserClass" option through the new UserClass= setting. It gained
1572 three new options in the new [CAN] section for configuring CAN
1573 networks. The MULTICAST and ALLMULTI interface flags may now be
1574 controlled explicitly with the new Multicast= and AllMulticast=
1577 * networkd will now automatically make use of the kernel's route
1578 expiration feature, if it is available.
1580 * udevd's .link files now support setting the number of receive and
1581 transmit channels, using the RxChannels=, TxChannels=,
1582 OtherChannels=, CombinedChannels= settings.
1584 * Support for UDPSegmentationOffload= has been removed, given its
1585 limited support in hardware, and waning software support.
1587 * networkd's .netdev files now support creating "netdevsim" interfaces.
1589 * PID 1 learnt a new bus call GetUnitByControlGroup() which may be used
1590 to query the unit belonging to a specific kernel control group.
1592 * systemd-analyze gained a new verb "cat-config", which may be used to
1593 dump the contents of any configuration file, with all its matching
1594 drop-in files added in, and honouring the usual search and masking
1595 logic applied to systemd configuration files. For example use
1596 "systemd-analyze cat-config systemd/system.conf" to get the complete
1597 system configuration file of systemd how it would be loaded by PID 1
1598 itself. Similar to this, various tools such as systemd-tmpfiles or
1599 systemd-sysusers, gained a new option "--cat-config", which does the
1600 corresponding operation for their own configuration settings. For
1601 example, "systemd-tmpfiles --cat-config" will now output the full
1602 list of tmpfiles.d/ lines in place.
1604 * timedatectl gained three new verbs: "show" shows bus properties of
1605 systemd-timedated, "timesync-status" shows the current NTP
1606 synchronization state of systemd-timesyncd, and "show-timesync"
1607 shows bus properties of systemd-timesyncd.
1609 * systemd-timesyncd gained a bus interface on which it exposes details
1612 * A new environment variable $SYSTEMD_TIMEDATED_NTP_SERVICES is now
1613 understood by systemd-timedated. It takes a colon-separated list of
1614 unit names of NTP client services. The list is used by
1615 "timedatectl set-ntp".
1617 * systemd-nspawn gained a new --rlimit= switch for setting initial
1618 resource limits for the container payload. There's a new switch
1619 --hostname= to explicitly override the container's hostname. A new
1620 --no-new-privileges= switch may be used to control the
1621 PR_SET_NO_NEW_PRIVS flag for the container payload. A new
1622 --oom-score-adjust= switch controls the OOM scoring adjustment value
1623 for the payload. The new --cpu-affinity= switch controls the CPU
1624 affinity of the container payload. The new --resolv-conf= switch
1625 allows more detailed control of /etc/resolv.conf handling of the
1626 container. Similarly, the new --timezone= switch allows more detailed
1627 control of /etc/localtime handling of the container.
1629 * systemd-detect-virt gained a new --list switch, which will print a
1630 list of all currently known VM and container environments.
1632 * Support for "Portable Services" has been added, see
1633 doc/PORTABLE_SERVICES.md for details. Currently, the support is still
1634 experimental, but this is expected to change soon. Reflecting this
1635 experimental state, the "portablectl" binary is not installed into
1636 /usr/bin yet. The binary has to be called with the full path
1637 /usr/lib/systemd/portablectl instead.
1639 * journalctl's and systemctl's -o switch now knows a new log output
1640 mode "with-unit". The output it generates is very similar to the
1641 regular "short" mode, but displays the unit name instead of the
1642 syslog tag for each log line. Also, the date is shown with timezone
1643 information. This mode is probably more useful than the classic
1644 "short" output mode for most purposes, except where pixel-perfect
1645 compatibility with classic /var/log/messages formatting is required.
1647 * A new --dump-bus-properties switch has been added to the systemd
1648 binary, which may be used to dump all supported D-Bus properties.
1649 (Options which are still supported, but are deprecated, are *not*
1652 * sd-bus gained a set of new calls:
1653 sd_bus_slot_set_floating()/sd_bus_slot_get_floating() may be used to
1654 enable/disable the "floating" state of a bus slot object,
1655 i.e. whether the slot object pins the bus it is allocated for into
1656 memory or if the bus slot object gets disconnected when the bus goes
1657 away. sd_bus_open_with_description(),
1658 sd_bus_open_user_with_description(),
1659 sd_bus_open_system_with_description() may be used to allocate bus
1660 objects and set their description string already during allocation.
1662 * sd-event gained support for watching inotify events from the event
1663 loop, in an efficient way, sharing inotify handles between multiple
1664 users. For this a new function sd_event_add_inotify() has been added.
1666 * sd-event and sd-bus gained support for calling special user-supplied
1667 destructor functions for userdata pointers associated with
1668 sd_event_source, sd_bus_slot, and sd_bus_track objects. For this new
1669 functions sd_bus_slot_set_destroy_callback,
1670 sd_bus_slot_get_destroy_callback, sd_bus_track_set_destroy_callback,
1671 sd_bus_track_get_destroy_callback,
1672 sd_event_source_set_destroy_callback,
1673 sd_event_source_get_destroy_callback have been added.
1675 * The "net.ipv4.tcp_ecn" sysctl will now be turned on by default.
1677 * PID 1 will now automatically reschedule .timer units whenever the
1678 local timezone changes. (They previously got rescheduled
1679 automatically when the system clock changed.)
1681 * New documentation has been added to document cgroups delegation,
1682 portable services and the various code quality tools we have set up:
1684 https://github.com/systemd/systemd/blob/master/docs/CGROUP_DELEGATION.md
1685 https://github.com/systemd/systemd/blob/master/docs/PORTABLE_SERVICES.md
1686 https://github.com/systemd/systemd/blob/master/docs/CODE_QUALITY.md
1688 * The Boot Loader Specification has been added to the source tree.
1690 https://github.com/systemd/systemd/blob/master/docs/BOOT_LOADER_SPECIFICATION.md
1692 While moving it into our source tree we have updated it and further
1693 changes are now accepted through the usual github PR workflow.
1695 * pam_systemd will now look for PAM userdata fields systemd.memory_max,
1696 systemd.tasks_max, systemd.cpu_weight, systemd.io_weight set by
1697 earlier PAM modules. The data in these fields is used to initialize
1698 the session scope's resource properties. Thus external PAM modules
1699 may now configure per-session limits, for example sourced from
1700 external user databases.
1702 * socket units with Accept=yes will now maintain a "refused" counter in
1703 addition to the existing "accepted" counter, counting connections
1704 refused due to the enforced limits.
1706 * The "systemd-path search-binaries-default" command may now be use to
1707 query the default, built-in $PATH PID 1 will pass to the services it
1710 * A new unit file setting PrivateMounts= has been added. It's a boolean
1711 option. If enabled the unit's processes are invoked in their own file
1712 system namespace. Note that this behaviour is also implied if any
1713 other file system namespacing options (such as PrivateTmp=,
1714 PrivateDevices=, ProtectSystem=, …) are used. This option is hence
1715 primarily useful for services that do not use any of the other file
1716 system namespacing options. One such service is systemd-udevd.service
1717 where this is now used by default.
1719 * ConditionSecurity= gained a new value "uefi-secureboot" that is true
1720 when the system is booted in UEFI "secure mode".
1722 * A new unit "system-update-pre.target" is added, which defines an
1723 optional synchronization point for offline system updates, as
1724 implemented by the pre-existing "system-update.target" unit. It
1725 allows ordering services before the service that executes the actual
1726 update process in a generic way.
1728 * Systemd now emits warnings whenever .include syntax is used.
1730 Contributions from: Adam Duskett, Alan Jenkins, Alessandro Casale,
1731 Alexander Kurtz, Alex Gartrell, Anssi Hannula, Arnaud Rebillout, Brian
1732 J. Murrell, Bruno Vernay, Chris Lamb, Chris Lesiak, Christian Brauner,
1733 Christian Hesse, Christian Rebischke, Colin Guthrie, Daniel Dao, Daniel
1734 Lin, Danylo Korostil, Davide Cavalca, David Tardon, Dimitri John
1735 Ledkov, Dmitriy Geels, Douglas Christman, Elia Geretto, emelenas, Emil
1736 Velikov, Evgeny Vereshchagin, Felipe Sateler, Feng Sun, Filipe
1737 Brandenburger, Franck Bui, futpib, Giuseppe Scrivano, Guillem Jover,
1738 guixxx, Hannes Reinecke, Hans de Goede, Harald Hoyer, Henrique Dante de
1739 Almeida, Hiram van Paassen, Ian Miell, Igor Gnatenko, Ivan Shapovalov,
1740 Iwan Timmer, James Cowgill, Jan Janssen, Jan Synacek, Jared Kazimir,
1741 Jérémy Rosen, João Paulo Rechi Vita, Joost Heitbrink, Jui-Chi Ricky
1742 Liang, Jürg Billeter, Kai-Heng Feng, Karol Augustin, Kay Sievers,
1743 Krzysztof Nowicki, Lauri Tirkkonen, Lennart Poettering, Leonard König,
1744 Long Li, Luca Boccassi, Lucas Werkmeister, Marcel Hoppe, Marc
1745 Kleine-Budde, Mario Limonciello, Martin Jansa, Martin Wilck, Mathieu
1746 Malaterre, Matteo F. Vescovi, Matthew McGinn, Matthias-Christian Ott,
1747 Michael Biebl, Michael Olbrich, Michael Prokop, Michal Koutný, Michal
1748 Sekletar, Mike Gilbert, Mikhail Kasimov, Milan Broz, Milan Pässler,
1749 Mladen Pejaković, Muhammet Kara, Nicolas Boichat, Omer Katz, Paride
1750 Legovini, Paul Menzel, Paul Milliken, Pavel Hrdina, Peter A. Bigot,
1751 Peter D'Hoye, Peter Hutterer, Peter Jones, Philip Sequeira, Philip
1752 Withnall, Piotr Drąg, Radostin Stoyanov, Ricardo Salveti de Araujo,
1753 Ronny Chevalier, Rosen Penev, Rubén Suárez Alvarez, Ryan Gonzalez,
1754 Salvo Tomaselli, Sebastian Reichel, Sergey Ptashnick, Sergio Lindo
1755 Mansilla, Stefan Schweter, Stephen Hemminger, Stuart Hayes, Susant
1756 Sahani, Sylvain Plantefève, Thomas H. P. Andersen, Tobias Jungel,
1757 Tomasz Torcz, Vito Caputo, Will Dietz, Will Thompson, Wim van Mourik,
1758 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
1760 — Berlin, 2018-06-22
1764 * The MemoryAccounting= unit property now defaults to on. After
1765 discussions with the upstream control group maintainers we learnt
1766 that the negative impact of cgroup memory accounting on current
1767 kernels is finally relatively minimal, so that it should be safe to
1768 enable this by default without affecting system performance. Besides
1769 memory accounting only task accounting is turned on by default, all
1770 other forms of resource accounting (CPU, IO, IP) remain off for now,
1771 because it's not clear yet that their impact is small enough to move
1772 from opt-in to opt-out. We recommend downstreams to leave memory
1773 accounting on by default if kernel 4.14 or higher is primarily
1774 used. On very resource constrained systems or when support for old
1775 kernels is a necessity, -Dmemory-accounting-default=false can be used
1776 to revert this change.
1778 * rpm scriptlets to update the udev hwdb and rules (%udev_hwdb_update,
1779 %udev_rules_update) and the journal catalog (%journal_catalog_update)
1780 from the upgrade scriptlets of individual packages now do nothing.
1781 Transfiletriggers have been added which will perform those updates
1782 once at the end of the transaction.
1784 Similar transfiletriggers have been added to execute any sysctl.d
1785 and binfmt.d rules. Thus, it should be unnecessary to provide any
1786 scriptlets to execute this configuration from package installation
1789 * systemd-sysusers gained a mode where the configuration to execute is
1790 specified on the command line, but this configuration is not executed
1791 directly, but instead it is merged with the configuration on disk,
1792 and the result is executed. This is useful for package installation
1793 scripts which want to create the user before installing any files on
1794 disk (in case some of those files are owned by that user), while
1795 still allowing local admin overrides.
1797 This functionality is exposed to rpm scriptlets through a new
1798 %sysusers_create_package macro. Old %sysusers_create and
1799 %sysusers_create_inline macros are deprecated.
1801 A transfiletrigger for sysusers.d configuration is now installed,
1802 which means that it should be unnecessary to call systemd-sysusers from
1803 package installation scripts, unless the package installs any files
1804 owned by those newly-created users, in which case
1805 %sysusers_create_package should be used.
1807 * Analogous change has been done for systemd-tmpfiles: it gained a mode
1808 where the command-line configuration is merged with the configuration
1809 on disk. This is exposed as the new %tmpfiles_create_package macro,
1810 and %tmpfiles_create is deprecated. A transfiletrigger is installed
1811 for tmpfiles.d, hence it should be unnecessary to call systemd-tmpfiles
1812 from package installation scripts.
1814 * sysusers.d configuration for a user may now also specify the group
1815 number, in addition to the user number ("u username 123:456"), or
1816 without the user number ("u username -:456").
1818 * Configution items for systemd-sysusers can now be specified as
1819 positional arguments when the new --inline switch is used.
1821 * The login shell of users created through sysusers.d may now be
1822 specified (previously, it was always /bin/sh for root and
1823 /sbin/nologin for other users).
1825 * systemd-analyze gained a new --global switch to look at global user
1826 configuration. It also gained a unit-paths verb to list the unit load
1827 paths that are compiled into systemd (which can be used with
1828 --systemd, --user, or --global).
1830 * udevadm trigger gained a new --settle/-w option to wait for any
1831 triggered events to finish (but just those, and not any other events
1832 which are triggered meanwhile).
1834 * The action that systemd-logind takes when the lid is closed and the
1835 machine is connected to external power can now be configured using
1836 HandleLidSwitchExternalPower= in logind.conf. Previously, this action
1837 was determined by HandleLidSwitch=, and, for backwards compatibility,
1838 is still is, if HandleLidSwitchExternalPower= is not explicitly set.
1840 * journalctl will periodically call sd_journal_process() to make it
1841 resilient against inotify queue overruns when journal files are
1842 rotated very quickly.
1844 * Two new functions in libsystemd — sd_bus_get_n_queued_read and
1845 sd_bus_get_n_queued_write — may be used to check the number of
1846 pending bus messages.
1848 * systemd gained a new
1849 org.freedesktop.systemd1.Manager.AttachProcessesToUnit dbus call
1850 which can be used to migrate foreign processes to scope and service
1851 units. The primary user for this new API is systemd itself: the
1852 systemd --user instance uses this call of the systemd --system
1853 instance to migrate processes if it itself gets the request to
1854 migrate processes and the kernel refuses this due to access
1855 restrictions. Thanks to this "systemd-run --scope --user …" works
1856 again in pure cgroup v2 environments when invoked from the user
1859 * A new TemporaryFileSystem= setting can be used to mask out part of
1860 the real file system tree with tmpfs mounts. This may be combined
1861 with BindPaths= and BindReadOnlyPaths= to hide files or directories
1862 not relevant to the unit, while still allowing some paths lower in
1863 the tree to be accessed.
1865 ProtectHome=tmpfs may now be used to hide user home and runtime
1866 directories from units, in a way that is mostly equivalent to
1867 "TemporaryFileSystem=/home /run/user /root".
1869 * Non-service units are now started with KeyringMode=shared by default.
1870 This means that mount and swapon and other mount tools have access
1871 to keys in the main keyring.
1873 * /sys/fs/bpf is now mounted automatically.
1875 * QNX virtualization is now detected by systemd-detect-virt and may
1876 be used in ConditionVirtualization=.
1878 * IPAccounting= may now be enabled also for slice units.
1880 * A new -Dsplit-bin= build configuration switch may be used to specify
1881 whether bin and sbin directories are merged, or if they should be
1882 included separately in $PATH and various listings of executable
1883 directories. The build configuration scripts will try to autodetect
1884 the proper values of -Dsplit-usr= and -Dsplit-bin= based on build
1885 system, but distributions are encouraged to configure this
1888 * A new -Dok-color= build configuration switch may be used to change
1889 the colour of "OK" status messages.
1891 * UPGRADE ISSUE: serialization of units using JoinsNamespaceOf= with
1892 PrivateNetwork=yes was buggy in previous versions of systemd. This
1893 means that after the upgrade and daemon-reexec, any such units must
1896 * INCOMPATIBILITY: as announced in the NEWS for 237, systemd-tmpfiles
1897 will not exclude read-only files owned by root from cleanup.
1899 Contributions from: Alan Jenkins, Alexander F Rødseth, Alexis Jeandet,
1900 Andika Triwidada, Andrei Gherzan, Ansgar Burchardt, antizealot1337,
1901 Batuhan Osman Taşkaya, Beniamino Galvani, Bill Yodlowsky, Caio Marcelo
1902 de Oliveira Filho, CuBiC, Daniele Medri, Daniel Mouritzen, Daniel
1903 Rusek, Davide Cavalca, Dimitri John Ledkov, Douglas Christman, Evgeny
1904 Vereshchagin, Faalagorn, Filipe Brandenburger, Franck Bui, futpib,
1905 Giacomo Longo, Gunnar Hjalmarsson, Hans de Goede, Hermann Gausterer,
1906 Iago López Galeiras, Jakub Filak, Jan Synacek, Jason A. Donenfeld,
1907 Javier Martinez Canillas, Jérémy Rosen, Lennart Poettering, Lucas
1908 Werkmeister, Mao Huang, Marco Gulino, Michael Biebl, Michael Vogt,
1909 MilhouseVH, Neal Gompa (ニール・ゴンパ), Oleander Reis, Olof Mogren,
1910 Patrick Uiterwijk, Peter Hutterer, Peter Portante, Piotr Drąg, Robert
1911 Antoni Buj Gelonch, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
1912 Fowler, SjonHortensius, snorreflorre, Susant Sahani, Sylvain
1913 Plantefève, Thomas Blume, Thomas Haller, Vito Caputo, Yu Watanabe,
1914 Zbigniew Jędrzejewski-Szmek, Марко М. Костић (Marko M. Kostić)
1916 — Warsaw, 2018-03-05
1920 * Some keyboards come with a zoom see-saw or rocker which until now got
1921 mapped to the Linux "zoomin/out" keys in hwdb. However, these
1922 keycodes are not recognized by any major desktop. They now produce
1923 Up/Down key events so that they can be used for scrolling.
1925 * INCOMPATIBILITY: systemd-tmpfiles' "f" lines changed behaviour
1926 slightly: previously, if an argument was specified for lines of this
1927 type (i.e. the right-most column was set) this string was appended to
1928 existing files each time systemd-tmpfiles was run. This behaviour was
1929 different from what the documentation said, and not particularly
1930 useful, as repeated systemd-tmpfiles invocations would not be
1931 idempotent and grow such files without bounds. With this release
1932 behaviour has been altered to match what the documentation says:
1933 lines of this type only have an effect if the indicated files don't
1934 exist yet, and only then the argument string is written to the file.
1936 * FUTURE INCOMPATIBILITY: In systemd v238 we intend to slightly change
1937 systemd-tmpfiles behaviour: previously, read-only files owned by root
1938 were always excluded from the file "aging" algorithm (i.e. the
1939 automatic clean-up of directories like /tmp based on
1940 atime/mtime/ctime). We intend to drop this restriction, and age files
1941 by default even when owned by root and read-only. This behaviour was
1942 inherited from older tools, but there have been requests to remove
1943 it, and it's not obvious why this restriction was made in the first
1944 place. Please speak up now, if you are aware of software that reqires
1945 this behaviour, otherwise we'll remove the restriction in v238.
1947 * A new environment variable $SYSTEMD_OFFLINE is now understood by
1948 systemctl. It takes a boolean argument. If on, systemctl assumes it
1949 operates on an "offline" OS tree, and will not attempt to talk to the
1950 service manager. Previously, this mode was implicitly enabled if a
1951 chroot() environment was detected, and this new environment variable
1952 now provides explicit control.
1954 * .path and .socket units may now be created transiently, too.
1955 Previously only service, mount, automount and timer units were
1956 supported as transient units. The systemd-run tool has been updated
1957 to expose this new functionality, you may hence use it now to bind
1958 arbitrary commands to path or socket activation on-the-fly from the
1959 command line. Moreover, almost all properties are now exposed for the
1960 unit types that already supported transient operation.
1962 * The systemd-mount command gained support for a new --owner= parameter
1963 which takes a user name, which is then resolved and included in uid=
1964 and gid= mount options string of the file system to mount.
1966 * A new unit condition ConditionControlGroupController= has been added
1967 that checks whether a specific cgroup controller is available.
1969 * Unit files, udev's .link files, and systemd-networkd's .netdev and
1970 .network files all gained support for a new condition
1971 ConditionKernelVersion= for checking against specific kernel
1974 * In systemd-networkd, the [IPVLAN] section in .netdev files gained
1975 support for configuring device flags in the Flags= setting. In the
1976 same files, the [Tunnel] section gained support for configuring
1977 AllowLocalRemote=. The [Route] section in .network files gained
1978 support for configuring InitialCongestionWindow=,
1979 InitialAdvertisedReceiveWindow= and QuickAck=. The [DHCP] section now
1980 understands RapidCommit=.
1982 * systemd-networkd's DHCPv6 support gained support for Prefix
1985 * sd-bus gained support for a new "watch-bind" feature. When this
1986 feature is enabled, an sd_bus connection may be set up to connect to
1987 an AF_UNIX socket in the file system as soon as it is created. This
1988 functionality is useful for writing early-boot services that
1989 automatically connect to the system bus as soon as it is started,
1990 without ugly time-based polling. systemd-networkd and
1991 systemd-resolved have been updated to make use of this
1992 functionality. busctl exposes this functionality in a new
1993 --watch-bind= command line switch.
1995 * sd-bus will now optionally synthesize a local "Connected" signal as
1996 soon as a D-Bus connection is set up fully. This message mirrors the
1997 already existing "Disconnected" signal which is synthesized when the
1998 connection is terminated. This signal is generally useful but
1999 particularly handy in combination with the "watch-bind" feature
2000 described above. Synthesizing of this message has to be requested
2001 explicitly through the new API call sd_bus_set_connected_signal(). In
2002 addition a new call sd_bus_is_ready() has been added that checks
2003 whether a connection is fully set up (i.e. between the "Connected" and
2004 "Disconnected" signals).
2006 * sd-bus gained two new calls sd_bus_request_name_async() and
2007 sd_bus_release_name_async() for asynchronously registering bus
2008 names. Similar, there is now sd_bus_add_match_async() for installing
2009 a signal match asynchronously. All of systemd's own services have
2010 been updated to make use of these calls. Doing these operations
2011 asynchronously has two benefits: it reduces the risk of deadlocks in
2012 case of cyclic dependencies between bus services, and it speeds up
2013 service initialization since synchronization points for bus
2014 round-trips are removed.
2016 * sd-bus gained two new calls sd_bus_match_signal() and
2017 sd_bus_match_signal_async(), which are similar to sd_bus_add_match()
2018 and sd_bus_add_match_async() but instead of taking a D-Bus match
2019 string take match fields as normal function parameters.
2021 * sd-bus gained two new calls sd_bus_set_sender() and
2022 sd_bus_message_set_sender() for setting the sender name of outgoing
2023 messages (either for all outgoing messages or for just one specific
2024 one). These calls are only useful in direct connections as on
2025 brokered connections the broker fills in the sender anyway,
2026 overwriting whatever the client filled in.
2028 * sd-event gained a new pseudo-handle that may be specified on all API
2029 calls where an "sd_event*" object is expected: SD_EVENT_DEFAULT. When
2030 used this refers to the default event loop object of the calling
2031 thread. Note however that this does not implicitly allocate one —
2032 which has to be done prior by using sd_event_default(). Similarly
2033 sd-bus gained three new pseudo-handles SD_BUS_DEFAULT,
2034 SD_BUS_DEFAULT_USER, SD_BUS_DEFAULT_SYSTEM that may be used to refer
2035 to the default bus of the specified type of the calling thread. Here
2036 too this does not implicitly allocate bus connection objects, this
2037 has to be done prior with sd_bus_default() and friends.
2039 * sd-event gained a new call pair
2040 sd_event_source_{get|set}_io_fd_own(). This may be used to request
2041 automatic closure of the file descriptor an IO event source watches
2042 when the event source is destroyed.
2044 * systemd-networkd gained support for natively configuring WireGuard
2047 * In previous versions systemd synthesized user records both for the
2048 "nobody" (UID 65534) and "root" (UID 0) users in nss-systemd and
2049 internally. In order to simplify distribution-wide renames of the
2050 "nobody" user (like it is planned in Fedora: nfsnobody → nobody), a
2051 new transitional flag file has been added: if
2052 /etc/systemd/dont-synthesize-nobody exists synthesizing of the 65534
2053 user and group record within the systemd codebase is disabled.
2055 * systemd-notify gained a new --uid= option for selecting the source
2056 user/UID to use for notification messages sent to the service
2059 * journalctl gained a new --grep= option to list only entries in which
2060 the message matches a certain pattern. By default matching is case
2061 insensitive if the pattern is lowercase, and case sensitive
2062 otherwise. Option --case-sensitive=yes|no can be used to override
2063 this an specify case sensitivity or case insensitivity.
2065 * There's now a "systemd-analyze service-watchdogs" command for printing
2066 the current state of the service runtime watchdog, and optionally
2067 enabling or disabling the per-service watchdogs system-wide if given a
2068 boolean argument (i.e. the concept you configure in WatchdogSec=), for
2069 debugging purposes. There's also a kernel command line option
2070 systemd.service_watchdogs= for controlling the same.
2072 * Two new "log-level" and "log-target" options for systemd-analyze were
2073 added that merge the now deprecated get-log-level, set-log-level and
2074 get-log-target, set-log-target pairs. The deprecated options are still
2075 understood for backwards compatibility. The two new options print the
2076 current value when no arguments are given, and set them when a
2077 level/target is given as an argument.
2079 * sysusers.d's "u" lines now optionally accept both a UID and a GID
2080 specification, separated by a ":" character, in order to create users
2081 where UID and GID do not match.
2083 Contributions from: Adam Duskett, Alan Jenkins, Alexander Kuleshov,
2084 Alexis Deruelle, Andrew Jeddeloh, Armin Widegreen, Batuhan Osman
2085 Taşkaya, Björn Esser, bleep_blop, Bruce A. Johnson, Chris Down, Clinton
2086 Roy, Colin Walters, Daniel Rusek, Dimitri John Ledkov, Dmitry Rozhkov,
2087 Evgeny Vereshchagin, Ewout van Mansom, Felipe Sateler, Franck Bui,
2088 Frantisek Sumsal, George Gaydarov, Gianluca Boiano, Hans-Christian
2089 Noren Egtvedt, Hans de Goede, Henrik Grindal Bakken, Jan Alexander
2090 Steffens, Jan Klötzke, Jason A. Donenfeld, jdkbx, Jérémy Rosen,
2091 Jerónimo Borque, John Lin, John Paul Herold, Jonathan Rudenberg, Jörg
2092 Thalheim, Ken (Bitsko) MacLeod, Larry Bernstone, Lennart Poettering,
2093 Lucas Werkmeister, Maciej S. Szmigiero, Marek Čermák, Martin Pitt,
2094 Mathieu Malaterre, Matthew Thode, Matthias-Christian Ott, Max Harmathy,
2095 Michael Biebl, Michael Vogt, Michal Koutný, Michal Sekletar, Michał
2096 Szczepański, Mike Gilbert, Nathaniel McCallum, Nicolas Chauvet, Olaf
2097 Hering, Olivier Schwander, Patrik Flykt, Paul Cercueil, Peter Hutterer,
2098 Piotr Drąg, Raphael Vogelgsang, Reverend Homer, Robert Kolchmeyer,
2099 Samuel Dionne-Riel, Sergey Ptashnick, Shawn Landden, Susant Sahani,
2100 Sylvain Plantefève, Thomas H. P. Andersen, Thomas Huth, Tomasz
2101 Bachorski, Vladislav Vishnyakov, Wieland Hoffmann, Yu Watanabe, Zachary
2102 Winnerman, Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски, Дилян
2109 * The modprobe.d/ drop-in for the bonding.ko kernel module introduced
2110 in v235 has been extended to also set the dummy.ko module option
2111 numdummies=0, preventing the kernel from automatically creating
2112 dummy0. All dummy interfaces must now be explicitly created.
2114 * Unknown '%' specifiers in configuration files are now rejected. This
2115 applies to units and tmpfiles.d configuration. Any percent characters
2116 that are followed by a letter or digit that are not supposed to be
2117 interpreted as the beginning of a specifier should be escaped by
2118 doubling ("%%"). (So "size=5%" is still accepted, as well as
2119 "size=5%,foo=bar", but not "LABEL=x%y%z" since %y and %z are not
2120 valid specifiers today.)
2122 * systemd-resolved now maintains a new dynamic
2123 /run/systemd/resolve/stub-resolv.conf compatibility file. It is
2124 recommended to make /etc/resolv.conf a symlink to it. This file
2125 points at the systemd-resolved stub DNS 127.0.0.53 resolver and
2126 includes dynamically acquired search domains, achieving more correct
2127 DNS resolution by software that bypasses local DNS APIs such as NSS.
2129 * The "uaccess" udev tag has been dropped from /dev/kvm and
2130 /dev/dri/renderD*. These devices now have the 0666 permissions by
2131 default (but this may be changed at build-time). /dev/dri/renderD*
2132 will now be owned by the "render" group along with /dev/kfd.
2134 * "DynamicUser=yes" has been enabled for systemd-timesyncd.service,
2135 systemd-journal-gatewayd.service and
2136 systemd-journal-upload.service. This means "nss-systemd" must be
2137 enabled in /etc/nsswitch.conf to ensure the UIDs assigned to these
2138 services are resolved properly.
2140 * In /etc/fstab two new mount options are now understood:
2141 x-systemd.makefs and x-systemd.growfs. The former has the effect that
2142 the configured file system is formatted before it is mounted, the
2143 latter that the file system is resized to the full block device size
2144 after it is mounted (i.e. if the file system is smaller than the
2145 partition it resides on, it's grown). This is similar to the fsck
2146 logic in /etc/fstab, and pulls in systemd-makefs@.service and
2147 systemd-growfs@.service as necessary, similar to
2148 systemd-fsck@.service. Resizing is currently only supported on ext4
2151 * In systemd-networkd, the IPv6 RA logic now optionally may announce
2152 DNS server and domain information.
2154 * Support for the LUKS2 on-disk format for encrypted partitions has
2155 been added. This requires libcryptsetup2 during compilation and
2158 * The systemd --user instance will now signal "readiness" when its
2159 basic.target unit has been reached, instead of when the run queue ran
2160 empty for the first time.
2162 * Tmpfiles.d with user configuration are now also supported.
2163 systemd-tmpfiles gained a new --user switch, and snippets placed in
2164 ~/.config/user-tmpfiles.d/ and corresponding directories will be
2165 executed by systemd-tmpfiles --user running in the new
2166 systemd-tmpfiles-setup.service and systemd-tmpfiles-clean.service
2167 running in the user session.
2169 * Unit files and tmpfiles.d snippets learnt three new % specifiers:
2170 %S resolves to the top-level state directory (/var/lib for the system
2171 instance, $XDG_CONFIG_HOME for the user instance), %C resolves to the
2172 top-level cache directory (/var/cache for the system instance,
2173 $XDG_CACHE_HOME for the user instance), %L resolves to the top-level
2174 logs directory (/var/log for the system instance,
2175 $XDG_CONFIG_HOME/log/ for the user instance). This matches the
2176 existing %t specifier, that resolves to the top-level runtime
2177 directory (/run for the system instance, and $XDG_RUNTIME_DIR for the
2180 * journalctl learnt a new parameter --output-fields= for limiting the
2181 set of journal fields to output in verbose and JSON output modes.
2183 * systemd-timesyncd's configuration file gained a new option
2184 RootDistanceMaxSec= for setting the maximum root distance of servers
2185 it'll use, as well as the new options PollIntervalMinSec= and
2186 PollIntervalMaxSec= to tweak the minimum and maximum poll interval.
2188 * bootctl gained a new command "list" for listing all available boot
2189 menu items on systems that follow the boot loader specification.
2191 * systemctl gained a new --dry-run switch that shows what would be done
2192 instead of doing it, and is currently supported by the shutdown and
2195 * ConditionSecurity= can now detect the TOMOYO security module.
2197 * Unit file [Install] sections are now also respected in unit drop-in
2198 files. This is intended to be used by drop-ins under /usr/lib/.
2200 * systemd-firstboot may now also set the initial keyboard mapping.
2202 * Udev "changed" events for devices which are exposed as systemd
2203 .device units are now propagated to units specified in
2204 ReloadPropagatedFrom= as reload requests.
2206 * If a udev device has a SYSTEMD_WANTS= property containing a systemd
2207 unit template name (i.e. a name in the form of 'foobar@.service',
2208 without the instance component between the '@' and - the '.'), then
2209 the escaped sysfs path of the device is automatically used as the
2212 * SystemCallFilter= in unit files has been extended so that an "errno"
2213 can be specified individually for each system call. Example:
2214 SystemCallFilter=~uname:EILSEQ.
2216 * The cgroup delegation logic has been substantially updated. Delegate=
2217 now optionally takes a list of controllers (instead of a boolean, as
2218 before), which lists the controllers to delegate at least.
2220 * The networkd DHCPv6 client now implements the FQDN option (RFC 4704).
2222 * A new LogLevelMax= setting configures the maximum log level any
2223 process of the service may log at (i.e. anything with a lesser
2224 priority than what is specified is automatically dropped). A new
2225 LogExtraFields= setting allows configuration of additional journal
2226 fields to attach to all log records generated by any of the unit's
2229 * New StandardInputData= and StandardInputText= settings along with the
2230 new option StandardInput=data may be used to configure textual or
2231 binary data that shall be passed to the executed service process via
2232 standard input, encoded in-line in the unit file.
2234 * StandardInput=, StandardOutput= and StandardError= may now be used to
2235 connect stdin/stdout/stderr of executed processes directly with a
2236 file or AF_UNIX socket in the file system, using the new "file:" option.
2238 * A new unit file option CollectMode= has been added, that allows
2239 tweaking the garbage collection logic for units. It may be used to
2240 tell systemd to garbage collect units that have failed automatically
2241 (normally it only GCs units that exited successfully). systemd-run
2242 and systemd-mount expose this new functionality with a new -G option.
2244 * "machinectl bind" may now be used to bind mount non-directories
2245 (i.e. regularfiles, devices, fifos, sockets).
2247 * systemd-analyze gained a new verb "calendar" for validating and
2248 testing calendar time specifications to use for OnCalendar= in timer
2249 units. Besides validating the expression it will calculate the next
2250 time the specified expression would elapse.
2252 * In addition to the pre-existing FailureAction= unit file setting
2253 there's now SuccessAction=, for configuring a shutdown action to
2254 execute when a unit completes successfully. This is useful in
2255 particular inside containers that shall terminate after some workload
2256 has been completed. Also, both options are now supported for all unit
2257 types, not just services.
2259 * networkds's IP rule support gained two new options
2260 IncomingInterface= and OutgoingInterface= for configuring the incoming
2261 and outgoing interfaces of configured rules. systemd-networkd also
2262 gained support for "vxcan" network devices.
2264 * networkd gained a new setting RequiredForOnline=, taking a
2265 boolean. If set, systemd-wait-online will take it into consideration
2266 when determining that the system is up, otherwise it will ignore the
2267 interface for this purpose.
2269 * The sd_notify() protocol gained support for a new operation: with
2270 FDSTOREREMOVE=1 file descriptors may be removed from the per-service
2271 store again, ahead of POLLHUP or POLLERR when they are removed
2274 * A new document doc/UIDS-GIDS.md has been added to the source tree,
2275 that documents the UID/GID range and assignment assumptions and
2276 requirements of systemd.
2278 * The watchdog device PID 1 will ping may now be configured through the
2279 WatchdogDevice= configuration file setting, or by setting the
2280 systemd.watchdog_service= kernel commandline option.
2282 * systemd-resolved's gained support for registering DNS-SD services on
2283 the local network using MulticastDNS. Services may either be
2284 registered by dropping in a .dnssd file in /etc/systemd/dnssd/ (or
2285 the same dir below /run, /usr/lib), or through its D-Bus API.
2287 * The sd_notify() protocol can now with EXTEND_TIMEOUT_USEC=microsecond
2288 extend the effective start, runtime, and stop time. The service must
2289 continue to send EXTEND_TIMEOUT_USEC within the period specified to
2290 prevent the service manager from making the service as timedout.
2292 * systemd-resolved's DNSSEC support gained support for RFC 8080
2293 (Ed25519 keys and signatures).
2295 * The systemd-resolve command line tool gained a new set of options
2296 --set-dns=, --set-domain=, --set-llmnr=, --set-mdns=, --set-dnssec=,
2297 --set-nta= and --revert to configure per-interface DNS configuration
2298 dynamically during runtime. It's useful for pushing DNS information
2299 into systemd-resolved from DNS hook scripts that various interface
2300 managing software supports (such as pppd).
2302 * systemd-nspawn gained a new --network-namespace-path= command line
2303 option, which may be used to make a container join an existing
2304 network namespace, by specifying a path to a "netns" file.
2306 Contributions from: Alan Jenkins, Alan Robertson, Alessandro Ghedini,
2307 Andrew Jeddeloh, Antonio Rojas, Ari, asavah, bleep_blop, Carsten
2308 Strotmann, Christian Brauner, Christian Hesse, Clinton Roy, Collin
2309 Eggert, Cong Wang, Daniel Black, Daniel Lockyer, Daniel Rusek, Dimitri
2310 John Ledkov, Dmitry Rozhkov, Dongsu Park, Edward A. James, Evgeny
2311 Vereshchagin, Florian Klink, Franck Bui, Gwendal Grignou, Hans de
2312 Goede, Harald Hoyer, Hristo Venev, Iago López Galeiras, Ikey Doherty,
2313 Jakub Wilk, Jérémy Rosen, Jiahui Xie, John Lin, José Bollo, Josef
2314 Andersson, juga0, Krzysztof Nowicki, Kyle Walker, Lars Karlitski, Lars
2315 Kellogg-Stedman, Lauri Tirkkonen, Lennart Poettering, Lubomir Rintel,
2316 Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn, Lukáš Říha, Lukasz
2317 Rubaszewski, Maciej S. Szmigiero, Mantas Mikulėnas, Marcus Folkesson,
2318 Martin Steuer, Mathieu Trudel-Lapierre, Matija Skala,
2319 Matthias-Christian Ott, Max Resch, Michael Biebl, Michael Vogt, Michal
2320 Koutný, Michal Sekletar, Mike Gilbert, Muhammet Kara, Neil Brown, Olaf
2321 Hering, Ondrej Kozina, Patrik Flykt, Patryk Kocielnik, Peter Hutterer,
2322 Piotr Drąg, Razvan Cojocaru, Robin McCorkell, Roland Hieber, Saran
2323 Tunyasuvunakool, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
2324 Arlott, Simon Peeters, Stanislav Angelovič, Stefan Agner, Susant
2325 Sahani, Sylvain Plantefève, Thomas Blume, Thomas Haller, Tiago Salem
2326 Herrmann, Tinu Weber, Tom Stellard, Topi Miettinen, Torsten Hilbrich,
2327 Vito Caputo, Vladislav Vishnyakov, WaLyong Cho, Yu Watanabe, Zbigniew
2328 Jędrzejewski-Szmek, Zeal Jagannatha
2330 — Berlin, 2017-12-14
2334 * INCOMPATIBILITY: systemd-logind.service and other long-running
2335 services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP
2336 communication with the outside. This generally improves security of
2337 the system, and is in almost all cases a safe and good choice, as
2338 these services do not and should not provide any network-facing
2339 functionality. However, systemd-logind uses the glibc NSS API to
2340 query the user database. This creates problems on systems where NSS
2341 is set up to directly consult network services for user database
2342 lookups. In particular, this creates incompatibilities with the
2343 "nss-nis" module, which attempts to directly contact the NIS/YP
2344 network servers it is configured for, and will now consistently
2345 fail. In such cases, it is possible to turn off IP sandboxing for
2346 systemd-logind.service (set IPAddressDeny= in its [Service] section
2347 to the empty string, via a .d/ unit file drop-in). Downstream
2348 distributions might want to update their nss-nis packaging to include
2349 such a drop-in snippet, accordingly, to hide this incompatibility
2350 from the user. Another option is to make use of glibc's nscd service
2351 to proxy such network requests through a privilege-separated, minimal
2352 local caching daemon, or to switch to more modern technologies such
2353 sssd, whose NSS hook-ups generally do not involve direct network
2354 access. In general, we think it's definitely time to question the
2355 implementation choices of nss-nis, i.e. whether it's a good idea
2356 today to embed a network-facing loadable module into all local
2357 processes that need to query the user database, including the most
2358 trivial and benign ones, such as "ls". For more details about
2359 IPAddressDeny= see below.
2361 * A new modprobe.d drop-in is now shipped by default that sets the
2362 bonding module option max_bonds=0. This overrides the kernel default,
2363 to avoid conflicts and ambiguity as to whether or not bond0 should be
2364 managed by systemd-networkd or not. This resolves multiple issues
2365 with bond0 properties not being applied, when bond0 is configured
2366 with systemd-networkd. Distributors may choose to not package this,
2367 however in that case users will be prevented from correctly managing
2368 bond0 interface using systemd-networkd.
2370 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
2371 which print the logging level and target of the system manager. They
2372 complement the existing "set-log-level" and "set-log-target" verbs
2373 used to change those values.
2375 * journald.conf gained a new boolean setting ReadKMsg= which defaults
2376 to on. If turned off kernel log messages will not be read by
2377 systemd-journald or included in the logs. It also gained a new
2378 setting LineMax= for configuring the maximum line length in
2379 STDOUT/STDERR log streams. The new default for this value is 48K, up
2380 from the previous hardcoded 2048.
2382 * A new unit setting RuntimeDirectoryPreserve= has been added, which
2383 allows more detailed control of what to do with a runtime directory
2384 configured with RuntimeDirectory= (i.e. a directory below /run or
2385 $XDG_RUNTIME_DIR) after a unit is stopped.
2387 * The RuntimeDirectory= setting for units gained support for creating
2388 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
2389 one top-level directory.
2391 * Units gained new options StateDirectory=, CacheDirectory=,
2392 LogsDirectory= and ConfigurationDirectory= which are closely related
2393 to RuntimeDirectory= but manage per-service directories below
2394 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
2395 possible to write unit files which when activated automatically gain
2396 properly owned service specific directories in these locations, thus
2397 making unit files self-contained and increasing compatibility with
2398 stateless systems and factory reset where /etc or /var are
2399 unpopulated at boot. Matching these new settings there's also
2400 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
2401 ConfigurationDirectoryMode= for configuring the access mode of these
2402 directories. These settings are particularly useful in combination
2403 with DynamicUser=yes as they provide secure, properly-owned,
2404 writable, and stateful locations for storage, excluded from the
2405 sandbox that such services live in otherwise.
2407 * Automake support has been removed from this release. systemd is now
2410 * systemd-journald will now aggressively cache client metadata during
2411 runtime, speeding up log write performance under pressure. This comes
2412 at a small price though: as much of the metadata is read
2413 asynchronously from /proc/ (and isn't implicitly attached to log
2414 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
2415 metadata stored alongside a log entry might be slightly
2416 out-of-date. Previously it could only be slightly newer than the log
2417 message. The time window is small however, and given that the kernel
2418 is unlikely to be improved anytime soon in this regard, this appears
2421 * nss-myhostname/systemd-resolved will now by default synthesize an
2422 A/AAAA resource record for the "_gateway" hostname, pointing to the
2423 current default IP gateway. Previously it did that for the "gateway"
2424 name, hampering adoption, as some distributions wanted to leave that
2425 host name open for local use. The old behaviour may still be
2426 requested at build time.
2428 * systemd-networkd's [Address] section in .network files gained a new
2429 Scope= setting for configuring the IP address scope. The [Network]
2430 section gained a new boolean setting ConfigureWithoutCarrier= that
2431 tells systemd-networkd to ignore link sensing when configuring the
2432 device. The [DHCP] section gained a new Anonymize= boolean option for
2433 turning on a number of options suggested in RFC 7844. A new
2434 [RoutingPolicyRule] section has been added for configuring the IP
2435 routing policy. The [Route] section has gained support for a new
2436 Type= setting which permits configuring
2437 blackhole/unreachable/prohibit routes.
2439 * The [VRF] section in .netdev files gained a new Table= setting for
2440 configuring the routing table to use. The [Tunnel] section gained a
2441 new Independent= boolean field for configuring tunnels independent of
2442 an underlying network interface. The [Bridge] section gained a new
2443 GroupForwardMask= option for configuration of propagation of link
2444 local frames between bridge ports.
2446 * The WakeOnLan= setting in .link files gained support for a number of
2447 new modes. A new TCP6SegmentationOffload= setting has been added for
2448 configuring TCP/IPv6 hardware segmentation offload.
2450 * The IPv6 RA sender implementation may now optionally send out RDNSS
2451 and RDNSSL records to supply DNS configuration to peers.
2453 * systemd-nspawn gained support for a new --system-call-filter= command
2454 line option for adding and removing entries in the default system
2455 call filter it applies. Moreover systemd-nspawn has been changed to
2456 implement a system call whitelist instead of a blacklist.
2458 * systemd-run gained support for a new --pipe command line option. If
2459 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
2460 are directly passed on to the activated transient service
2461 executable. This allows invoking arbitrary processes as systemd
2462 services (for example to take benefit of dependency management,
2463 accounting management, resource management or log management that is
2464 done automatically for services) — while still allowing them to be
2465 integrated in a classic UNIX shell pipeline.
2467 * When a service sends RELOAD=1 via sd_notify() and reload propagation
2468 using ReloadPropagationTo= is configured, a reload is now propagated
2469 to configured units. (Previously this was only done on explicitly
2470 requested reloads, using "systemctl reload" or an equivalent
2473 * For each service unit a restart counter is now kept: it is increased
2474 each time the service is restarted due to Restart=, and may be
2475 queried using "systemctl show -p NRestarts …".
2477 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
2478 @signal and @timer have been added, for usage with SystemCallFilter=
2479 in unit files and the new --system-call-filter= command line option
2480 of systemd-nspawn (see above).
2482 * ExecStart= lines in unit files gained two new modifiers: when a
2483 command line is prefixed with "!" the command will be executed as
2484 configured, except for the credentials applied by
2485 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
2486 "+", but does still apply namespacing options unlike "+". There's
2487 also "!!" now, which is mostly identical, but becomes a NOP on
2488 systems that support ambient capabilities. This is useful to write
2489 unit files that work with ambient capabilities where possible but
2490 automatically fall back to traditional privilege dropping mechanisms
2491 on systems where this is not supported.
2493 * ListenNetlink= settings in socket units now support RDMA netlink
2496 * A new unit file setting LockPersonality= has been added which permits
2497 locking down the chosen execution domain ("personality") of a service
2500 * A new special target "getty-pre.target" has been added, which is
2501 ordered before all text logins, and may be used to order services
2502 before textual logins acquire access to the console.
2504 * systemd will now attempt to load the virtio-rng.ko kernel module very
2505 early on if a VM environment supporting this is detected. This should
2506 improve entropy during early boot in virtualized environments.
2508 * A _netdev option is now supported in /etc/crypttab that operates in a
2509 similar way as the same option in /etc/fstab: it permits configuring
2510 encrypted devices that need to be ordered after the network is up.
2511 Following this logic, two new special targets
2512 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
2513 added that are to cryptsetup.target what remote-fs.target and
2514 remote-fs-pre.target are to local-fs.target.
2516 * Service units gained a new UnsetEnvironment= setting which permits
2517 unsetting specific environment variables for services that are
2518 normally passed to it (for example in order to mask out locale
2519 settings for specific services that can't deal with it).
2521 * Units acquired a new boolean option IPAccounting=. When turned on, IP
2522 traffic accounting (packet count as well as byte count) is done for
2523 the service, and shown as part of "systemctl status" or "systemd-run
2526 * Service units acquired two new options IPAddressAllow= and
2527 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
2528 for configuring a simple IP access control list for all sockets of
2529 the unit. These options are available also on .slice and .socket
2530 units, permitting flexible access list configuration for individual
2531 services as well as groups of services (as defined by a slice unit),
2532 including system-wide. Note that IP ACLs configured this way are
2533 enforced on every single IPv4 and IPv6 socket created by any process
2534 of the service unit, and apply to ingress as well as egress traffic.
2536 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
2537 structured log message is generated each time the unit is stopped,
2538 containing information about the consumed resources of this
2541 * A new setting KeyringMode= has been added to unit files, which may be
2542 used to control how the kernel keyring is set up for executed
2545 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
2546 "systemctl kexec" and "systemctl exit" are now always asynchronous in
2547 behaviour (that is: these commands return immediately after the
2548 operation was enqueued instead of waiting for the operation to
2549 complete). Previously, "systemctl poweroff" and "systemctl reboot"
2550 were asynchronous on systems using systemd-logind (i.e. almost
2551 always, and like they were on sysvinit), and the other three commands
2552 were unconditionally synchronous. With this release this is cleaned
2553 up, and callers will see the same asynchronous behaviour on all
2554 systems for all five operations.
2556 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
2559 * .timer units now accept calendar specifications in other timezones
2560 than UTC or the local timezone.
2562 * The tmpfiles snippet var.conf has been changed to create
2563 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
2564 the "utmp" group already, and it appears to be generally understood
2565 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
2566 databases. Previously this was implemented correctly for all these
2567 databases excepts btmp, which has been opened up like this now
2568 too. Note that while the other databases are world-readable
2569 (i.e. 0644), btmp is not and remains more restrictive.
2571 * The systemd-resolve tool gained a new --reset-server-features
2572 switch. When invoked like this systemd-resolved will forget
2573 everything it learnt about the features supported by the configured
2574 upstream DNS servers, and restarts the feature probing logic on the
2575 next resolver look-up for them at the highest feature level
2578 * The status dump systemd-resolved sends to the logs upon receiving
2579 SIGUSR1 now also includes information about all DNS servers it is
2580 configured to use, and the features levels it probed for them.
2582 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
2583 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
2584 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
2585 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
2586 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
2587 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
2588 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
2589 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
2590 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
2591 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
2592 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
2593 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
2594 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
2595 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
2596 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
2597 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
2598 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
2599 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
2600 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
2601 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
2603 — Berlin, 2017-10-06
2607 * Meson is now supported as build system in addition to Automake. It is
2608 our plan to remove Automake in one of our next releases, so that
2609 Meson becomes our exclusive build system. Hence, please start using
2610 the Meson build system in your downstream packaging. There's plenty
2611 of documentation around how to use Meson, the extremely brief
2614 ./autogen.sh && ./configure && make && sudo make install
2618 meson build && ninja -C build && sudo ninja -C build install
2620 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
2621 which permits configuring a timeout on the time a job is
2622 running. This is particularly useful for setting timeouts on jobs for
2625 * Unit files gained two new options ConditionUser= and ConditionGroup=
2626 for conditionalizing units based on the identity of the user/group
2627 running a systemd user instance.
2629 * systemd-networkd now understands a new FlowLabel= setting in the
2630 [VXLAN] section of .network files, as well as a Priority= in
2631 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
2632 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
2633 gained support for configuration of GENEVE links, and IPv6 address
2634 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
2636 * .link files now understand a new Port= setting.
2638 * systemd-networkd's DHCP support gained support for DHCP option 119
2639 (domain search list).
2641 * systemd-networkd gained support for serving IPv6 address ranges using
2642 the Router Advertisement protocol. The new .network configuration
2643 section [IPv6Prefix] may be used to configure the ranges to
2644 serve. This is implemented based on a new, minimal, native server
2645 implementation of RA.
2647 * journalctl's --output= switch gained support for a new parameter
2648 "short-iso-precise" for a mode where timestamps are shown as precise
2651 * systemd-udevd's "net_id" builtin may now generate stable network
2652 interface names from IBM PowerVM VIO devices as well as ACPI platform
2655 * MulticastDNS support in systemd-resolved may now be explicitly
2656 enabled/disabled using the new MulticastDNS= configuration file
2659 * systemd-resolved may now optionally use libidn2 instead of the libidn
2660 for processing internationalized domain names. Support for libidn2
2661 should be considered experimental and should not be enabled by
2664 * "machinectl pull-tar" and related call may now do verification of
2665 downloaded images using SUSE-style .sha256 checksum files in addition
2666 to the already existing support for validating using Ubuntu-style
2669 * sd-bus gained support for a new sd_bus_message_appendv() call which
2670 is va_list equivalent of sd_bus_message_append().
2672 * sd-boot gained support for validating images using SHIM/MOK.
2674 * The SMACK code learnt support for "onlycap".
2676 * systemd-mount --umount is now much smarter in figuring out how to
2677 properly unmount a device given its mount or device path.
2679 * The code to call libnss_dns as a fallback from libnss_resolve when
2680 the communication with systemd-resolved fails was removed. This
2681 fallback was redundant and interfered with the [!UNAVAIL=return]
2682 suffix. See nss-resolve(8) for the recommended configuration.
2684 * systemd-logind may now be restarted without losing state. It stores
2685 the file descriptors for devices it manages in the system manager
2686 using the FDSTORE= mechanism. Please note that further changes in
2687 other components may be required to make use of this (for example
2688 Xorg has code to listen for stops of systemd-logind and terminate
2689 itself when logind is stopped or restarted, in order to avoid using
2690 stale file descriptors for graphical devices, which is now
2691 counterproductive and must be reverted in order for restarts of
2692 systemd-logind to be safe. See
2693 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
2695 * All kernel-install plugins are called with the environment variable
2696 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
2697 /etc/machine-id. If the machine ID could not be determined,
2698 $KERNEL_INSTALL_MACHINE_ID will be empty. Plugins should not put
2699 anything in the entry directory (passed as the second argument) if
2700 $KERNEL_INSTALL_MACHINE_ID is empty. For backwards compatibility, a
2701 temporary directory is passed as the entry directory and removed
2702 after all the plugins exit.
2704 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
2705 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
2706 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
2707 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
2708 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
2709 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
2710 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
2711 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
2712 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
2713 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
2714 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
2715 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
2716 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
2717 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
2718 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
2719 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
2720 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
2721 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
2722 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
2723 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
2724 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
2725 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
2726 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
2727 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
2728 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
2729 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
2730 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
2731 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
2732 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
2735 — Berlin, 2017-07-12
2739 * The "hybrid" control group mode has been modified to improve
2740 compatibility with "legacy" cgroups-v1 setups. Specifically, the
2741 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
2742 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
2743 cgroups-v1 hierarchy), the only externally visible change being that
2744 the cgroups-v2 hierarchy is also mounted, to
2745 /sys/fs/cgroup/unified. This should provide a large degree of
2746 compatibility with "legacy" cgroups-v1, while taking benefit of the
2747 better management capabilities of cgroups-v2.
2749 * The default control group setup mode may be selected both a boot-time
2750 via a set of kernel command line parameters (specifically:
2751 systemd.unified_cgroup_hierarchy= and
2752 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
2753 default selected on the configure command line
2754 (--with-default-hierarchy=). The upstream default is "hybrid"
2755 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
2756 this will change in a future systemd version to be "unified" (pure
2757 cgroups-v2 mode). The third option for the compile time option is
2758 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
2759 distributions to default to "hybrid" mode for release distributions,
2760 starting with v233. We recommend "unified" for development
2761 distributions (specifically: distributions such as Fedora's rawhide)
2762 as that's where things are headed in the long run. Use "legacy" for
2763 greatest stability and compatibility only.
2765 * Note one current limitation of "unified" and "hybrid" control group
2766 setup modes: the kernel currently does not permit the systemd --user
2767 instance (i.e. unprivileged code) to migrate processes between two
2768 disconnected cgroup subtrees, even if both are managed and owned by
2769 the user. This effectively means "systemd-run --user --scope" doesn't
2770 work when invoked from outside of any "systemd --user" service or
2771 scope. Specifically, it is not supported from session scopes. We are
2772 working on fixing this in a future systemd version. (See #3388 for
2773 further details about this.)
2775 * DBus policy files are now installed into /usr rather than /etc. Make
2776 sure your system has dbus >= 1.9.18 running before upgrading to this
2777 version, or override the install path with --with-dbuspolicydir= .
2779 * All python scripts shipped with systemd (specifically: the various
2780 tests written in Python) now require Python 3.
2782 * systemd unit tests can now run standalone (without the source or
2783 build directories), and can be installed into /usr/lib/systemd/tests/
2784 with 'make install-tests'.
2786 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
2787 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
2790 * Support for the %c, %r, %R specifiers in unit files has been
2791 removed. Specifiers are not supposed to be dependent on configuration
2792 in the unit file itself (so that they resolve the same regardless
2793 where used in the unit files), but these specifiers were influenced
2794 by the Slice= option.
2796 * The shell invoked by debug-shell.service now defaults to /bin/sh in
2797 all cases. If distributions want to use a different shell for this
2798 purpose (for example Fedora's /sbin/sushell) they need to specify
2799 this explicitly at configure time using --with-debug-shell=.
2801 * The confirmation spawn prompt has been reworked to offer the
2804 (c)ontinue, proceed without asking anymore
2805 (D)ump, show the state of the unit
2806 (f)ail, don't execute the command and pretend it failed
2808 (i)nfo, show a short summary of the unit
2809 (j)obs, show jobs that are in progress
2810 (s)kip, don't execute the command and pretend it succeeded
2811 (y)es, execute the command
2813 The 'n' choice for the confirmation spawn prompt has been removed,
2814 because its meaning was confusing.
2816 The prompt may now also be redirected to an alternative console by
2817 specifying the console as parameter to systemd.confirm_spawn=.
2819 * Services of Type=notify require a READY=1 notification to be sent
2820 during startup. If no such message is sent, the service now fails,
2821 even if the main process exited with a successful exit code.
2823 * Services that fail to start up correctly now always have their
2824 ExecStopPost= commands executed. Previously, they'd enter "failed"
2825 state directly, without executing these commands.
2827 * The option MulticastDNS= of network configuration files has acquired
2828 an actual implementation. With MulticastDNS=yes a host can resolve
2829 names of remote hosts and reply to mDNS A and AAAA requests.
2831 * When units are about to be started an additional check is now done to
2832 ensure that all dependencies of type BindsTo= (when used in
2833 combination with After=) have been started.
2835 * systemd-analyze gained a new verb "syscall-filter" which shows which
2836 system call groups are defined for the SystemCallFilter= unit file
2837 setting, and which system calls they contain.
2839 * A new system call filter group "@filesystem" has been added,
2840 consisting of various file system related system calls. Group
2841 "@reboot" has been added, covering reboot, kexec and shutdown related
2842 calls. Finally, group "@swap" has been added covering swap
2843 configuration related calls.
2845 * A new unit file option RestrictNamespaces= has been added that may be
2846 used to restrict access to the various process namespace types the
2847 Linux kernel provides. Specifically, it may be used to take away the
2848 right for a service unit to create additional file system, network,
2849 user, and other namespaces. This sandboxing option is particularly
2850 relevant due to the high amount of recently discovered namespacing
2851 related vulnerabilities in the kernel.
2853 * systemd-udev's .link files gained support for a new AutoNegotiation=
2854 setting for configuring Ethernet auto-negotiation.
2856 * systemd-networkd's .network files gained support for a new
2857 ListenPort= setting in the [DHCP] section to explicitly configure the
2858 UDP client port the DHCP client shall listen on.
2860 * .network files gained a new Unmanaged= boolean setting for explicitly
2861 excluding one or more interfaces from management by systemd-networkd.
2863 * The systemd-networkd ProxyARP= option has been renamed to
2864 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
2865 renamed to ReduceARPProxy=. The old names continue to be available
2868 * systemd-networkd gained support for configuring IPv6 Proxy NDP
2869 addresses via the new IPv6ProxyNDPAddress= .network file setting.
2871 * systemd-networkd's bonding device support gained support for two new
2872 configuration options ActiveSlave= and PrimarySlave=.
2874 * The various options in the [Match] section of .network files gained
2875 support for negative matching.
2877 * New systemd-specific mount options are now understood in /etc/fstab:
2879 x-systemd.mount-timeout= may be used to configure the maximum
2880 permitted runtime of the mount command.
2882 x-systemd.device-bound may be set to bind a mount point to its
2883 backing device unit, in order to automatically remove a mount point
2884 if its backing device is unplugged. This option may also be
2885 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
2886 on the block device, which is now automatically set for all CDROM
2887 drives, so that mounted CDs are automatically unmounted when they are
2888 removed from the drive.
2890 x-systemd.after= and x-systemd.before= may be used to explicitly
2891 order a mount after or before another unit or mount point.
2893 * Enqueued start jobs for device units are now automatically garbage
2894 collected if there are no jobs waiting for them anymore.
2896 * systemctl list-jobs gained two new switches: with --after, for every
2897 queued job the jobs it's waiting for are shown; with --before the
2898 jobs which it's blocking are shown.
2900 * systemd-nspawn gained support for ephemeral boots from disk images
2901 (or in other words: --ephemeral and --image= may now be
2902 combined). Moreover, ephemeral boots are now supported for normal
2903 directories, even if the backing file system is not btrfs. Of course,
2904 if the file system does not support file system snapshots or
2905 reflinks, the initial copy operation will be relatively expensive, but
2906 this should still be suitable for many use cases.
2908 * Calendar time specifications in .timer units now support
2909 specifications relative to the end of a month by using "~" instead of
2910 "-" as separator between month and day. For example, "*-02~03" means
2911 "the third last day in February". In addition a new syntax for
2912 repeated events has been added using the "/" character. For example,
2913 "9..17/2:00" means "every two hours from 9am to 5pm".
2915 * systemd-socket-proxyd gained a new parameter --connections-max= for
2916 configuring the maximum number of concurrent connections.
2918 * sd-id128 gained a new API for generating unique IDs for the host in a
2919 way that does not leak the machine ID. Specifically,
2920 sd_id128_get_machine_app_specific() derives an ID based on the
2921 machine ID a in well-defined, non-reversible, stable way. This is
2922 useful whenever an identifier for the host is needed but where the
2923 identifier shall not be useful to identify the system beyond the
2924 scope of the application itself. (Internally this uses HMAC-SHA256 as
2925 keyed hash function using the machine ID as input.)
2927 * NotifyAccess= gained a new supported value "exec". When set
2928 notifications are accepted from all processes systemd itself invoked,
2929 including all control processes.
2931 * .nspawn files gained support for defining overlay mounts using the
2932 Overlay= and OverlayReadOnly= options. Previously this functionality
2933 was only available on the systemd-nspawn command line.
2935 * systemd-nspawn's --bind= and --overlay= options gained support for
2936 bind/overlay mounts whose source lies within the container tree by
2937 prefixing the source path with "+".
2939 * systemd-nspawn's --bind= and --overlay= options gained support for
2940 automatically allocating a temporary source directory in /var/tmp
2941 that is removed when the container dies. Specifically, if the source
2942 directory is specified as empty string this mechanism is selected. An
2943 example usage is --overlay=+/var::/var, which creates an overlay
2944 mount based on the original /var contained in the image, overlayed
2945 with a temporary directory in the host's /var/tmp. This way changes
2946 to /var are automatically flushed when the container shuts down.
2948 * systemd-nspawn --image= option does now permit raw file system block
2949 devices (in addition to images containing partition tables, as
2952 * The disk image dissection logic in systemd-nspawn gained support for
2953 automatically setting up LUKS encrypted as well as Verity protected
2954 partitions. When a container is booted from an encrypted image the
2955 passphrase is queried at start-up time. When a container with Verity
2956 data is started, the root hash is search in a ".roothash" file
2957 accompanying the disk image (alternatively, pass the root hash via
2958 the new --root-hash= command line option).
2960 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
2961 be used to dissect disk images the same way as systemd-nspawn does
2962 it, following the Bootable Partition Specification. It may even be
2963 used to mount disk images with complex partition setups (including
2964 LUKS and Verity partitions) to a local host directory, in order to
2965 inspect them. This tool is not considered public API (yet), and is
2966 thus not installed into /usr/bin. Please do not rely on its
2967 existence, since it might go away or be changed in later systemd
2970 * A new generator "systemd-verity-generator" has been added, similar in
2971 style to "systemd-cryptsetup-generator", permitting automatic setup of
2972 Verity root partitions when systemd boots up. In order to make use of
2973 this your partition setup should follow the Discoverable Partitions
2974 Specification, and the GPT partition ID of the root file system
2975 partition should be identical to the upper 128bit of the Verity root
2976 hash. The GPT partition ID of the Verity partition protecting it
2977 should be the lower 128bit of the Verity root hash. If the partition
2978 image follows this model it is sufficient to specify a single
2979 "roothash=" kernel command line argument to both configure which root
2980 image and verity partition to use as well as the root hash for
2981 it. Note that systemd-nspawn's Verity support follows the same
2982 semantics, meaning that disk images with proper Verity data in place
2983 may be booted in containers with systemd-nspawn as well as on
2984 physical systems via the verity generator. Also note that the "mkosi"
2985 tool available at https://github.com/systemd/mkosi has been updated
2986 to generate Verity protected disk images following this scheme. In
2987 fact, it has been updated to generate disk images that optionally
2988 implement a complete UEFI SecureBoot trust chain, involving a signed
2989 kernel and initrd image that incorporates such a root hash as well as
2990 a Verity-enabled root partition.
2992 * The hardware database (hwdb) udev supports has been updated to carry
2993 accelerometer quirks.
2995 * All system services are now run with a fresh kernel keyring set up
2996 for them. The invocation ID is stored by default in it, thus
2997 providing a safe, non-overridable way to determine the invocation
3000 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
3001 options for bind mounting arbitrary paths in a service-specific
3002 way. When these options are used, arbitrary host or service files and
3003 directories may be mounted to arbitrary locations in the service's
3006 * Documentation has been added that lists all of systemd's low-level
3007 environment variables:
3009 https://github.com/systemd/systemd/blob/master/docs/ENVIRONMENT.md
3011 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
3012 whether a specific socket file descriptor matches a specified socket
3015 * systemd-firstboot has been updated to check for the
3016 systemd.firstboot= kernel command line option. It accepts a boolean
3017 and when set to false the first boot questions are skipped.
3019 * systemd-fstab-generator has been updated to check for the
3020 systemd.volatile= kernel command line option, which either takes an
3021 optional boolean parameter or the special value "state". If used the
3022 system may be booted in a "volatile" boot mode. Specifically,
3023 "systemd.volatile" is used, the root directory will be mounted as
3024 tmpfs, and only /usr is mounted from the actual root file system. If
3025 "systemd.volatile=state" is used, the root directory will be mounted
3026 as usual, but /var is mounted as tmpfs. This concept provides similar
3027 functionality as systemd-nspawn's --volatile= option, but provides it
3028 on physical boots. Use this option for implementing stateless
3029 systems, or testing systems with all state and/or configuration reset
3030 to the defaults. (Note though that many distributions are not
3031 prepared to boot up without a populated /etc or /var, though.)
3033 * systemd-gpt-auto-generator gained support for LUKS encrypted root
3034 partitions. Previously it only supported LUKS encrypted partitions
3035 for all other uses, except for the root partition itself.
3037 * Socket units gained support for listening on AF_VSOCK sockets for
3038 communication in virtualized QEMU environments.
3040 * The "configure" script gained a new option --with-fallback-hostname=
3041 for specifying the fallback hostname to use if none is configured in
3042 /etc/hostname. For example, by specifying
3043 --with-fallback-hostname=fedora it is possible to default to a
3044 hostname of "fedora" on pristine installations.
3046 * systemd-cgls gained support for a new --unit= switch for listing only
3047 the control groups of a specific unit. Similar --user-unit= has been
3048 added for listing only the control groups of a specific user unit.
3050 * systemd-mount gained a new --umount switch for unmounting a mount or
3051 automount point (and all mount/automount points below it).
3053 * systemd will now refuse full configuration reloads (via systemctl
3054 daemon-reload and related calls) unless at least 16MiB of free space
3055 are available in /run. This is a safety precaution in order to ensure
3056 that generators can safely operate after the reload completed.
3058 * A new unit file option RootImage= has been added, which has a similar
3059 effect as RootDirectory= but mounts the service's root directory from
3060 a disk image instead of plain directory. This logic reuses the same
3061 image dissection and mount logic that systemd-nspawn already uses,
3062 and hence supports any disk images systemd-nspawn supports, including
3063 those following the Discoverable Partition Specification, as well as
3064 Verity enabled images. This option enables systemd to run system
3065 services directly off disk images acting as resource bundles,
3066 possibly even including full integrity data.
3068 * A new MountAPIVFS= unit file option has been added, taking a boolean
3069 argument. If enabled /proc, /sys and /dev (collectively called the
3070 "API VFS") will be mounted for the service. This is only relevant if
3071 RootDirectory= or RootImage= is used for the service, as these mounts
3072 are of course in place in the host mount namespace anyway.
3074 * systemd-nspawn gained support for a new --pivot-root= switch. If
3075 specified the root directory within the container image is pivoted to
3076 the specified mount point, while the original root disk is moved to a
3077 different place. This option enables booting of ostree images
3078 directly with systemd-nspawn.
3080 * The systemd build scripts will no longer complain if the NTP server
3081 addresses are not changed from the defaults. Google now supports
3082 these NTP servers officially. We still recommend downstreams to
3083 properly register an NTP pool with the NTP pool project though.
3085 * coredumpctl gained a new "--reverse" option for printing the list
3086 of coredumps in reverse order.
3088 * coredumpctl will now show additional information about truncated and
3089 inaccessible coredumps, as well as coredumps that are still being
3090 processed. It also gained a new --quiet switch for suppressing
3091 additional informational message in its output.
3093 * coredumpctl gained support for only showing coredumps newer and/or
3094 older than specific timestamps, using the new --since= and --until=
3095 options, reminiscent of journalctl's options by the same name.
3097 * The systemd-coredump logic has been improved so that it may be reused
3098 to collect backtraces in non-compiled languages, for example in
3099 scripting languages such as Python.
3101 * machinectl will now show the UID shift of local containers, if user
3102 namespacing is enabled for them.
3104 * systemd will now optionally run "environment generator" binaries at
3105 configuration load time. They may be used to add environment
3106 variables to the environment block passed to services invoked. One
3107 user environment generator is shipped by default that sets up
3108 environment variables based on files dropped into /etc/environment.d
3109 and ~/.config/environment.d/.
3111 * systemd-resolved now includes the new, recently published 2017 DNSSEC
3114 * hostnamed has been updated to report a new chassis type of
3115 "convertible" to cover "foldable" laptops that can both act as a
3116 tablet and as a laptop, such as various Lenovo Yoga devices.
3118 Contributions from: Adrián López, Alexander Galanin, Alexander
3119 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
3120 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
3121 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
3122 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
3123 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
3124 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
3125 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
3126 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
3127 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
3128 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
3129 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
3130 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
3131 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
3132 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
3133 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
3134 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
3135 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
3136 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
3137 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
3138 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
3139 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
3140 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
3141 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
3142 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
3143 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
3144 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
3145 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
3148 — Berlin, 2017-03-01
3152 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
3153 RestrictAddressFamilies= enabled. These sandboxing options should
3154 generally be compatible with the various external udev call-out
3155 binaries we are aware of, however there may be exceptions, in
3156 particular when exotic languages for these call-outs are used. In
3157 this case, consider turning off these settings locally.
3159 * The new RemoveIPC= option can be used to remove IPC objects owned by
3160 the user or group of a service when that service exits.
3162 * The new ProtectKernelModules= option can be used to disable explicit
3163 load and unload operations of kernel modules by a service. In
3164 addition access to /usr/lib/modules is removed if this option is set.
3166 * ProtectSystem= option gained a new value "strict", which causes the
3167 whole file system tree with the exception of /dev, /proc, and /sys,
3168 to be remounted read-only for a service.
3170 * The new ProtectKernelTunables= option can be used to disable
3171 modification of configuration files in /sys and /proc by a service.
3172 Various directories and files are remounted read-only, so access is
3173 restricted even if the file permissions would allow it.
3175 * The new ProtectControlGroups= option can be used to disable write
3176 access by a service to /sys/fs/cgroup.
3178 * Various systemd services have been hardened with
3179 ProtectKernelTunables=yes, ProtectControlGroups=yes,
3180 RestrictAddressFamilies=.
3182 * Support for dynamically creating users for the lifetime of a service
3183 has been added. If DynamicUser=yes is specified, user and group IDs
3184 will be allocated from the range 61184..65519 for the lifetime of the
3185 service. They can be resolved using the new nss-systemd.so NSS
3186 module. The module must be enabled in /etc/nsswitch.conf. Services
3187 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
3188 any resources allocated by the service will be cleaned up when the
3189 service exits. They also have ProtectHome=read-only and
3190 ProtectSystem=strict enabled, so they are not able to make any
3191 permanent modifications to the system.
3193 * The nss-systemd module also always resolves root and nobody, making
3194 it possible to have no /etc/passwd or /etc/group files in minimal
3195 container or chroot environments.
3197 * Services may be started with their own user namespace using the new
3198 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
3199 under which the service is running are mapped. All other users are
3202 * Support for the cgroup namespace has been added to systemd-nspawn. If
3203 supported by kernel, the container system started by systemd-nspawn
3204 will have its own view of the cgroup hierarchy. This new behaviour
3205 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
3207 * The new MemorySwapMax= option can be used to limit the maximum swap
3208 usage under the unified cgroup hierarchy.
3210 * Support for the CPU controller in the unified cgroup hierarchy has
3211 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
3212 options. This controller requires out-of-tree patches for the kernel
3213 and the support is provisional.
3215 * Mount and automount units may now be created transiently
3216 (i.e. dynamically at runtime via the bus API, instead of requiring
3217 unit files in the file system).
3219 * systemd-mount is a new tool which may mount file systems – much like
3220 mount(8), optionally pulling in additional dependencies through
3221 transient .mount and .automount units. For example, this tool
3222 automatically runs fsck on a backing block device before mounting,
3223 and allows the automount logic to be used dynamically from the
3224 command line for establishing mount points. This tool is particularly
3225 useful when dealing with removable media, as it will ensure fsck is
3226 run – if necessary – before the first access and that the file system
3227 is quickly unmounted after each access by utilizing the automount
3228 logic. This maximizes the chance that the file system on the
3229 removable media stays in a clean state, and if it isn't in a clean
3230 state is fixed automatically.
3232 * LazyUnmount=yes option for mount units has been added to expose the
3233 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
3236 * /efi will be used as the mount point of the EFI boot partition, if
3237 the directory is present, and the mount point was not configured
3238 through other means (e.g. fstab). If /efi directory does not exist,
3239 /boot will be used as before. This makes it easier to automatically
3240 mount the EFI partition on systems where /boot is used for something
3243 * When operating on GPT disk images for containers, systemd-nspawn will
3244 now mount the ESP to /boot or /efi according to the same rules as PID
3245 1 running on a host. This allows tools like "bootctl" to operate
3246 correctly within such containers, in order to make container images
3247 bootable on physical systems.
3249 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
3251 * Two new user session targets have been added to support running
3252 graphical sessions under the systemd --user instance:
3253 graphical-session.target and graphical-session-pre.target. See
3254 systemd.special(7) for a description of how those targets should be
3257 * The vconsole initialization code has been significantly reworked to
3258 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
3259 support unicode keymaps. Font and keymap configuration will now be
3260 copied to all allocated virtual consoles.
3262 * FreeBSD's bhyve virtualization is now detected.
3264 * Information recorded in the journal for core dumps now includes the
3265 contents of /proc/mountinfo and the command line of the process at
3266 the top of the process hierarchy (which is usually the init process
3269 * systemd-journal-gatewayd learned the --directory= option to serve
3270 files from the specified location.
3272 * journalctl --root=… can be used to peruse the journal in the
3273 /var/log/ directories inside of a container tree. This is similar to
3274 the existing --machine= option, but does not require the container to
3277 * The hardware database has been extended to support
3278 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
3281 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
3282 specify the click rate for mice which include a horizontal wheel with
3283 a click rate that is different than the one for the vertical wheel.
3285 * systemd-run gained a new --wait option that makes service execution
3286 synchronous. (Specifically, the command will not return until the
3287 specified service binary exited.)
3289 * systemctl gained a new --wait option that causes the start command to
3290 wait until the units being started have terminated again.
3292 * A new journal output mode "short-full" has been added which displays
3293 timestamps with abbreviated English day names and adds a timezone
3294 suffix. Those timestamps include more information than the default
3295 "short" output mode, and can be passed directly to journalctl's
3296 --since= and --until= options.
3298 * /etc/resolv.conf will be bind-mounted into containers started by
3299 systemd-nspawn, if possible, so any changes to resolv.conf contents
3300 are automatically propagated to the container.
3302 * The number of instances for socket-activated services originating
3303 from a single IP address can be limited with
3304 MaxConnectionsPerSource=, extending the existing setting of
3307 * systemd-networkd gained support for vcan ("Virtual CAN") interface
3310 * .netdev and .network configuration can now be extended through
3313 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
3314 Segmentation Offload, Generic Receive Offload, Large Receive Offload
3315 can be enabled and disabled using the new UDPSegmentationOffload=,
3316 TCPSegmentationOffload=, GenericSegmentationOffload=,
3317 GenericReceiveOffload=, LargeReceiveOffload= options in the
3318 [Link] section of .link files.
3320 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
3321 Port VLAN ID can be configured for bridge devices using the new STP=,
3322 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
3323 section of .netdev files.
3325 * The route table to which routes received over DHCP or RA should be
3326 added can be configured with the new RouteTable= option in the [DHCP]
3327 and [IPv6AcceptRA] sections of .network files.
3329 * The Address Resolution Protocol can be disabled on links managed by
3330 systemd-networkd using the ARP=no setting in the [Link] section of
3333 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
3334 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
3335 encode information about the result and exit codes of the current
3336 service runtime cycle.
3338 * systemd-sysctl will now configure kernel parameters in the order
3339 they occur in the configuration files. This matches what sysctl
3340 has been traditionally doing.
3342 * kernel-install "plugins" that are executed to perform various
3343 tasks after a new kernel is added and before an old one is removed
3344 can now return a special value to terminate the procedure and
3345 prevent any later plugins from running.
3347 * Journald's SplitMode=login setting has been deprecated. It has been
3348 removed from documentation, and its use is discouraged. In a future
3349 release it will be completely removed, and made equivalent to current
3350 default of SplitMode=uid.
3352 * Storage=both option setting in /etc/systemd/coredump.conf has been
3353 removed. With fast LZ4 compression storing the core dump twice is not
3356 * The --share-system systemd-nspawn option has been replaced with an
3357 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
3358 this functionality is discouraged. In addition the variables
3359 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
3360 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
3361 individual namespaces.
3363 * "machinectl list" now shows the IP address of running containers in
3364 the output, as well as OS release information.
3366 * "loginctl list" now shows the TTY of each session in the output.
3368 * sd-bus gained new API calls sd_bus_track_set_recursive(),
3369 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
3370 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
3371 tracking objects in a "recursive" mode, where a single client can be
3372 counted multiple times, if it takes multiple references.
3374 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
3375 sd_bus_get_exit_on_disconnect(). They may be used to make a
3376 process using sd-bus automatically exit if the bus connection is
3379 * Bus clients of the service manager may now "pin" loaded units into
3380 memory, by taking an explicit reference on them. This is useful to
3381 ensure the client can retrieve runtime data about the service even
3382 after the service completed execution. Taking such a reference is
3383 available only for privileged clients and should be helpful to watch
3384 running services in a race-free manner, and in particular collect
3385 information about exit statuses and results.
3387 * The nss-resolve module has been changed to strictly return UNAVAIL
3388 when communication via D-Bus with resolved failed, and NOTFOUND when
3389 a lookup completed but was negative. This means it is now possible to
3390 neatly configure fallbacks using nsswitch.conf result checking
3391 expressions. Taking benefit of this, the new recommended
3392 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
3394 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
3396 * A new setting CtrlAltDelBurstAction= has been added to
3397 /etc/systemd/system.conf which may be used to configure the precise
3398 behaviour if the user on the console presses Ctrl-Alt-Del more often
3399 than 7 times in 2s. Previously this would unconditionally result in
3400 an expedited, immediate reboot. With this new setting the precise
3401 operation may be configured in more detail, and also turned off
3404 * In .netdev files two new settings RemoteChecksumTx= and
3405 RemoteChecksumRx= are now understood that permit configuring the
3406 remote checksumming logic for VXLAN networks.
3408 * The service manager learnt a new "invocation ID" concept for invoked
3409 services. Each runtime cycle of a service will get a new invocation
3410 ID (a 128bit random UUID) assigned that identifies the current
3411 run of the service uniquely and globally. A new invocation ID
3412 is generated each time a service starts up. The journal will store
3413 the invocation ID of a service along with any logged messages, thus
3414 making the invocation ID useful for matching the online runtime of a
3415 service with the offline log data it generated in a safe way without
3416 relying on synchronized timestamps. In many ways this new service
3417 invocation ID concept is similar to the kernel's boot ID concept that
3418 uniquely and globally identifies the runtime of each boot. The
3419 invocation ID of a service is passed to the service itself via an
3420 environment variable ($INVOCATION_ID). A new bus call
3421 GetUnitByInvocationID() has been added that is similar to GetUnit()
3422 but instead of retrieving the bus path for a unit by its name
3423 retrieves it by its invocation ID. The returned path is valid only as
3424 long as the passed invocation ID is current.
3426 * systemd-resolved gained a new "DNSStubListener" setting in
3427 resolved.conf. It either takes a boolean value or the special values
3428 "udp" and "tcp", and configures whether to enable the stub DNS
3429 listener on 127.0.0.53:53.
3431 * IP addresses configured via networkd may now carry additional
3432 configuration settings supported by the kernel. New options include:
3433 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
3434 PrefixRoute=, AutoJoin=.
3436 * The PAM configuration fragment file for "user@.service" shipped with
3437 systemd (i.e. the --user instance of systemd) has been stripped to
3438 the minimum necessary to make the system boot. Previously, it
3439 contained Fedora-specific stanzas that did not apply to other
3440 distributions. It is expected that downstream distributions add
3441 additional configuration lines, matching their needs to this file,
3442 using it only as rough template of what systemd itself needs. Note
3443 that this reduced fragment does not even include an invocation of
3444 pam_limits which most distributions probably want to add, even though
3445 systemd itself does not need it. (There's also the new build time
3446 option --with-pamconfdir=no to disable installation of the PAM
3449 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
3450 capability is now also dropped from its set (in addition to
3451 CAP_SYS_MKNOD as before).
3453 * In service unit files it is now possible to connect a specific named
3454 file descriptor with stdin/stdout/stdout of an executed service. The
3455 name may be specified in matching .socket units using the
3456 FileDescriptorName= setting.
3458 * A number of journal settings may now be configured on the kernel
3459 command line. Specifically, the following options are now understood:
3460 systemd.journald.max_level_console=,
3461 systemd.journald.max_level_store=,
3462 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
3463 systemd.journald.max_level_wall=.
3465 * "systemctl is-enabled --full" will now show by which symlinks a unit
3466 file is enabled in the unit dependency tree.
3468 * Support for VeraCrypt encrypted partitions has been added to the
3469 "cryptsetup" logic and /etc/crypttab.
3471 * systemd-detect-virt gained support for a new --private-users switch
3472 that checks whether the invoking processes are running inside a user
3473 namespace. Similar, a new special value "private-users" for the
3474 existing ConditionVirtualization= setting has been added, permitting
3475 skipping of specific units in user namespace environments.
3477 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
3478 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
3479 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
3480 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
3481 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
3482 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
3483 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
3484 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
3485 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
3486 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
3487 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
3488 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
3489 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
3490 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
3491 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
3492 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
3493 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
3494 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
3495 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
3496 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
3497 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
3498 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
3499 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
3500 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
3501 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
3502 Jędrzejewski-Szmek, Zeal Jagannatha
3504 — Santa Fe, 2016-11-03
3508 * In service units the various ExecXYZ= settings have been extended
3509 with an additional special character as first argument of the
3510 assigned value: if the character '+' is used the specified command
3511 line it will be run with full privileges, regardless of User=,
3512 Group=, CapabilityBoundingSet= and similar options. The effect is
3513 similar to the existing PermissionsStartOnly= option, but allows
3514 configuration of this concept for each executed command line
3517 * Services may now alter the service watchdog timeout at runtime by
3518 sending a WATCHDOG_USEC= message via sd_notify().
3520 * MemoryLimit= and related unit settings now optionally take percentage
3521 specifications. The percentage is taken relative to the amount of
3522 physical memory in the system (or in case of containers, the assigned
3523 amount of memory). This allows scaling service resources neatly with
3524 the amount of RAM available on the system. Similarly, systemd-logind's
3525 RuntimeDirectorySize= option now also optionally takes percentage
3528 * In similar fashion TasksMax= takes percentage values now, too. The
3529 value is taken relative to the configured maximum number of processes
3530 on the system. The per-service task maximum has been changed to 15%
3531 using this functionality. (Effectively this is an increase of 512 →
3532 4915 for service units, given the kernel's default pid_max setting.)
3534 * Calendar time specifications in .timer units now understand a ".."
3535 syntax for time ranges. Example: "4..7:10" may now be used for
3536 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
3539 * The InaccessableDirectories=, ReadOnlyDirectories= and
3540 ReadWriteDirectories= unit file settings have been renamed to
3541 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
3542 applied to all kinds of file nodes, and not just directories, with
3543 the exception of symlinks. Specifically these settings may now be
3544 used on block and character device nodes, UNIX sockets and FIFOS as
3545 well as regular files. The old names of these settings remain
3546 available for compatibility.
3548 * systemd will now log about all service processes it kills forcibly
3549 (using SIGKILL) because they remained after the clean shutdown phase
3550 of the service completed. This should help identifying services that
3551 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
3552 systemd-logind's configuration a similar log message is generated for
3553 processes killed at the end of each session due to this setting.
3555 * systemd will now set the $JOURNAL_STREAM environment variable for all
3556 services whose stdout/stderr are connected to the Journal (which
3557 effectively means by default: all services). The variable contains
3558 the device and inode number of the file descriptor used for
3559 stdout/stderr. This may be used by invoked programs to detect whether
3560 their stdout/stderr is connected to the Journal, in which case they
3561 can switch over to direct Journal communication, thus being able to
3562 pass extended, structured metadata along with their log messages. As
3563 one example, this is now used by glib's logging primitives.
3565 * When using systemd's default tmp.mount unit for /tmp, the mount point
3566 will now be established with the "nosuid" and "nodev" options. This
3567 avoids privilege escalation attacks that put traps and exploits into
3568 /tmp. However, this might cause problems if you e. g. put container
3569 images or overlays into /tmp; if you need this, override tmp.mount's
3570 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
3573 * systemd now supports the "memory" cgroup controller also on
3576 * The systemd-cgtop tool now optionally takes a control group path as
3577 command line argument. If specified, the control group list shown is
3578 limited to subgroups of that group.
3580 * The SystemCallFilter= unit file setting gained support for
3581 pre-defined, named system call filter sets. For example
3582 SystemCallFilter=@clock is now an effective way to make all clock
3583 changing-related system calls unavailable to a service. A number of
3584 similar pre-defined groups are defined. Writing system call filters
3585 for system services is simplified substantially with this new
3586 concept. Accordingly, all of systemd's own, long-running services now
3587 enable system call filtering based on this, by default.
3589 * A new service setting MemoryDenyWriteExecute= has been added, taking
3590 a boolean value. If turned on, a service may no longer create memory
3591 mappings that are writable and executable at the same time. This
3592 enhances security for services where this is enabled as it becomes
3593 harder to dynamically write and then execute memory in exploited
3594 service processes. This option has been enabled for all of systemd's
3595 own long-running services.
3597 * A new RestrictRealtime= service setting has been added, taking a
3598 boolean argument. If set the service's processes may no longer
3599 acquire realtime scheduling. This improves security as realtime
3600 scheduling may otherwise be used to easily freeze the system.
3602 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
3603 value. This may be used for requesting that the system manager inside
3604 of the container reports start-up completion to nspawn which then
3605 propagates this notification further to the service manager
3606 supervising nspawn itself. A related option NotifyReady= in .nspawn
3607 files has been added too. This functionality allows ordering of the
3608 start-up of multiple containers using the usual systemd ordering
3611 * machinectl gained a new command "stop" that is an alias for
3614 * systemd-resolved gained support for contacting DNS servers on
3615 link-local IPv6 addresses.
3617 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
3618 its caches. A method call for requesting the same operation has been
3619 added to the bus API too, and is made available via "systemd-resolve
3622 * systemd-resolve gained a new --status switch. If passed a brief
3623 summary of the used DNS configuration with per-interface information
3626 * resolved.conf gained a new Cache= boolean option, defaulting to
3627 on. If turned off local DNS caching is disabled. This comes with a
3628 performance penalty in particular when DNSSEC is enabled. Note that
3629 resolved disables its internal caching implicitly anyway, when the
3630 configured DNS server is on a host-local IP address such as ::1 or
3631 127.0.0.1, thus automatically avoiding double local caching.
3633 * systemd-resolved now listens on the local IP address 127.0.0.53:53
3634 for DNS requests. This improves compatibility with local programs
3635 that do not use the libc NSS or systemd-resolved's bus APIs for name
3636 resolution. This minimal DNS service is only available to local
3637 programs and does not implement the full DNS protocol, but enough to
3638 cover local DNS clients. A new, static resolv.conf file, listing just
3639 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
3640 now recommended to make /etc/resolv.conf a symlink to this file in
3641 order to route all DNS lookups to systemd-resolved, regardless if
3642 done via NSS, the bus API or raw DNS packets. Note that this local
3643 DNS service is not as fully featured as the libc NSS or
3644 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
3645 used to deliver link-local address information (as this implies
3646 sending a local interface index along), LLMNR/mDNS support via this
3647 interface is severely restricted. It is thus strongly recommended for
3648 all applications to use the libc NSS API or native systemd-resolved
3651 * systemd-networkd's bridge support learned a new setting
3652 VLANFiltering= for controlling VLAN filtering. Moreover a new section
3653 in .network files has been added for configuring VLAN bridging in
3654 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
3656 * systemd-networkd's IPv6 Router Advertisement code now makes use of
3657 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
3658 now be acquired without relying on DHCPv6. Two new options
3659 UseDomains= and UseDNS= have been added to configure this behaviour.
3661 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
3662 renamed IPv6AcceptRA=, without altering its behaviour. The old
3663 setting name remains available for compatibility reasons.
3665 * The systemd-networkd VTI/VTI6 tunneling support gained new options
3666 Key=, InputKey= and OutputKey=.
3668 * systemd-networkd gained support for VRF ("Virtual Routing Function")
3669 interface configuration.
3671 * "systemctl edit" may now be used to create new unit files by
3672 specifying the --force switch.
3674 * sd-event gained a new function sd_event_get_iteration() for
3675 requesting the current iteration counter of the event loop. It starts
3676 at zero and is increased by one with each event loop iteration.
3678 * A new rpm macro %systemd_ordering is provided by the macros.systemd
3679 file. It can be used in lieu of %systemd_requires in packages which
3680 don't use any systemd functionality and are intended to be installed
3681 in minimal containers without systemd present. This macro provides
3682 ordering dependencies to ensure that if the package is installed in
3683 the same rpm transaction as systemd, systemd will be installed before
3684 the scriptlets for the package are executed, allowing unit presets
3687 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
3688 been added to simplify packaging of generators.
3690 * The os-release file gained VERSION_CODENAME field for the
3691 distribution nickname (e.g. VERSION_CODENAME=woody).
3693 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
3694 can be set to disable parsing of metadata and the creation
3695 of persistent symlinks for that device.
3697 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
3698 to make them available to logged-in users has been reverted.
3700 * Much of the common code of the various systemd components is now
3701 built into an internal shared library libsystemd-shared-231.so
3702 (incorporating the systemd version number in the name, to be updated
3703 with future releases) that the components link to. This should
3704 decrease systemd footprint both in memory during runtime and on
3705 disk. Note that the shared library is not for public use, and is
3706 neither API nor ABI stable, but is likely to change with every new
3707 released update. Packagers need to make sure that binaries
3708 linking to libsystemd-shared.so are updated in step with the
3711 * Configuration for "mkosi" is now part of the systemd
3712 repository. mkosi is a tool to easily build legacy-free OS images,
3713 and is available on github: https://github.com/systemd/mkosi. If
3714 "mkosi" is invoked in the build tree a new raw OS image is generated
3715 incorporating the systemd sources currently being worked on and a
3716 clean, fresh distribution installation. The generated OS image may be
3717 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
3718 UEFI PC. This functionality is particularly useful to easily test
3719 local changes made to systemd in a pristine, defined environment. See
3720 doc/HACKING for details.
3722 * configure learned the --with-support-url= option to specify the
3723 distribution's bugtracker.
3725 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
3726 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
3727 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
3728 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
3729 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
3730 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
3731 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
3732 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
3733 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
3734 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
3735 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
3736 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
3737 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
3738 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
3739 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
3740 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
3741 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
3742 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
3743 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
3745 — Berlin, 2016-07-25
3749 * DNSSEC is now turned on by default in systemd-resolved (in
3750 "allow-downgrade" mode), but may be turned off during compile time by
3751 passing "--with-default-dnssec=no" to "configure" (and of course,
3752 during runtime with DNSSEC= in resolved.conf). We recommend
3753 downstreams to leave this on at least during development cycles and
3754 report any issues with the DNSSEC logic upstream. We are very
3755 interested in collecting feedback about the DNSSEC validator and its
3756 limitations in the wild. Note however, that DNSSEC support is
3757 probably nothing downstreams should turn on in stable distros just
3758 yet, as it might create incompatibilities with a few DNS servers and
3759 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
3760 automatically whenever we detect such incompatible setups, but there
3761 might be systems we do not cover yet. Hence: please help us testing
3762 the DNSSEC code, leave this on where you can, report back, but then
3763 again don't consider turning this on in your stable, LTS or
3764 production release just yet. (Note that you have to enable
3765 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
3766 and its DNSSEC mode for host name resolution from local
3769 * systemd-resolve conveniently resolves DANE records with the --tlsa
3770 option and OPENPGPKEY records with the --openpgp option. It also
3771 supports dumping raw DNS record data via the new --raw= switch.
3773 * systemd-logind will now by default terminate user processes that are
3774 part of the user session scope unit (session-XX.scope) when the user
3775 logs out. This behavior is controlled by the KillUserProcesses=
3776 setting in logind.conf, and the previous default of "no" is now
3777 changed to "yes". This means that user sessions will be properly
3778 cleaned up after, but additional steps are necessary to allow
3779 intentionally long-running processes to survive logout.
3781 While the user is logged in at least once, user@.service is running,
3782 and any service that should survive the end of any individual login
3783 session can be started at a user service or scope using systemd-run.
3784 systemd-run(1) man page has been extended with an example which shows
3785 how to run screen in a scope unit underneath user@.service. The same
3786 command works for tmux.
3788 After the user logs out of all sessions, user@.service will be
3789 terminated too, by default, unless the user has "lingering" enabled.
3790 To effectively allow users to run long-term tasks even if they are
3791 logged out, lingering must be enabled for them. See loginctl(1) for
3792 details. The default polkit policy was modified to allow users to
3793 set lingering for themselves without authentication.
3795 Previous defaults can be restored at compile time by the
3796 --without-kill-user-processes option to "configure".
3798 * systemd-logind gained new configuration settings SessionsMax= and
3799 InhibitorsMax=, both with a default of 8192. It will not register new
3800 user sessions or inhibitors above this limit.
3802 * systemd-logind will now reload configuration on SIGHUP.
3804 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
3805 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
3806 enable. Also, support for the "io" cgroup controller in the unified
3807 hierarchy has been added, so that the "memory", "pids" and "io" are
3808 now the controllers that are supported on the unified hierarchy.
3810 WARNING: it is not possible to use previous systemd versions with
3811 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
3812 is necessary to also update systemd in the initramfs if using the
3813 unified hierarchy. An updated SELinux policy is also required.
3815 * LLDP support has been extended, and both passive (receive-only) and
3816 active (sender) modes are supported. Passive mode ("routers-only") is
3817 enabled by default in systemd-networkd. Active LLDP mode is enabled
3818 by default for containers on the internal network. The "networkctl
3819 lldp" command may be used to list information gathered. "networkctl
3820 status" will also show basic LLDP information on connected peers now.
3822 * The IAID and DUID unique identifier sent in DHCP requests may now be
3823 configured for the system and each .network file managed by
3824 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
3826 * systemd-networkd gained support for configuring proxy ARP support for
3827 each interface, via the ProxyArp= setting in .network files. It also
3828 gained support for configuring the multicast querier feature of
3829 bridge devices, via the new MulticastQuerier= setting in .netdev
3830 files. Similarly, snooping on the IGMP traffic can be controlled
3831 via the new setting MulticastSnooping=.
3833 A new setting PreferredLifetime= has been added for addresses
3834 configured in .network file to configure the lifetime intended for an
3837 The systemd-networkd DHCP server gained the option EmitRouter=, which
3838 defaults to yes, to configure whether the DHCP Option 3 (Router)
3841 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
3842 systemd-socket-activate and installed into /usr/bin. It is now fully
3845 * systemd-journald now uses separate threads to flush changes to disk
3846 when closing journal files, thus reducing impact of slow disk I/O on
3847 logging performance.
3849 * The sd-journal API gained two new calls
3850 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
3851 can be used to open journal files using file descriptors instead of
3852 file or directory paths. sd_journal_open_container() has been
3853 deprecated, sd_journal_open_directory_fd() should be used instead
3854 with the flag SD_JOURNAL_OS_ROOT.
3856 * journalctl learned a new output mode "-o short-unix" that outputs log
3857 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
3858 UTC). It also gained support for a new --no-hostname setting to
3859 suppress the hostname column in the family of "short" output modes.
3861 * systemd-ask-password now optionally skips printing of the password to
3862 stdout with --no-output which can be useful in scripts.
3864 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
3865 (devices tagged with ID_MAKER_TOOL) are now tagged with
3866 "uaccess" and are available to logged in users.
3868 * The DeviceAllow= unit setting now supports specifiers (with "%").
3870 * "systemctl show" gained a new --value switch, which allows print a
3871 only the contents of a specific unit property, without also printing
3872 the property's name. Similar support was added to "show*" verbs
3873 of loginctl and machinectl that output "key=value" lists.
3875 * A new unit type "generated" was added for files dynamically generated
3876 by generator tools. Similarly, a new unit type "transient" is used
3877 for unit files created using the runtime API. "systemctl enable" will
3878 refuse to operate on such files.
3880 * A new command "systemctl revert" has been added that may be used to
3881 revert to the vendor version of a unit file, in case local changes
3882 have been made by adding drop-ins or overriding the unit file.
3884 * "machinectl clean" gained a new verb to automatically remove all or
3885 just hidden container images.
3887 * systemd-tmpfiles gained support for a new line type "e" for emptying
3888 directories, if they exist, without creating them if they don't.
3890 * systemd-nspawn gained support for automatically patching the UID/GIDs
3891 of the owners and the ACLs of all files and directories in a
3892 container tree to match the UID/GID user namespacing range selected
3893 for the container invocation. This mode is enabled via the new
3894 --private-users-chown switch. It also gained support for
3895 automatically choosing a free, previously unused UID/GID range when
3896 starting a container, via the new --private-users=pick setting (which
3897 implies --private-users-chown). Together, these options for the first
3898 time make user namespacing for nspawn containers fully automatic and
3899 thus deployable. The systemd-nspawn@.service template unit file has
3900 been changed to use this functionality by default.
3902 * systemd-nspawn gained a new --network-zone= switch, that allows
3903 creating ad-hoc virtual Ethernet links between multiple containers,
3904 that only exist as long as at least one container referencing them is
3905 running. This allows easy connecting of multiple containers with a
3906 common link that implements an Ethernet broadcast domain. Each of
3907 these network "zones" may be named relatively freely by the user, and
3908 may be referenced by any number of containers, but each container may
3909 only reference one of these "zones". On the lower level, this is
3910 implemented by an automatically managed bridge network interface for
3911 each zone, that is created when the first container referencing its
3912 zone is created and removed when the last one referencing its zone
3915 * The default start timeout may now be configured on the kernel command
3916 line via systemd.default_timeout_start_sec=. It was already
3917 configurable via the DefaultTimeoutStartSec= option in
3918 /etc/systemd/system.conf.
3920 * Socket units gained a new TriggerLimitIntervalSec= and
3921 TriggerLimitBurst= setting to configure a limit on the activation
3922 rate of the socket unit.
3924 * The LimitNICE= setting now optionally takes normal UNIX nice values
3925 in addition to the raw integer limit value. If the specified
3926 parameter is prefixed with "+" or "-" and is in the range -20..19 the
3927 value is understood as UNIX nice value. If not prefixed like this it
3928 is understood as raw RLIMIT_NICE limit.
3930 * Note that the effect of the PrivateDevices= unit file setting changed
3931 slightly with this release: the per-device /dev file system will be
3932 mounted read-only from this version on, and will have "noexec"
3933 set. This (minor) change of behavior might cause some (exceptional)
3934 legacy software to break, when PrivateDevices=yes is set for its
3935 service. Please leave PrivateDevices= off if you run into problems
3938 * systemd-bootchart has been split out to a separate repository:
3939 https://github.com/systemd/systemd-bootchart
3941 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
3942 merged into the kernel in its current form.
3944 * The compatibility libraries libsystemd-daemon.so,
3945 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
3946 which have been deprecated since systemd-209 have been removed along
3947 with the corresponding pkg-config files. All symbols provided by
3948 those libraries are provided by libsystemd.so.
3950 * The Capabilities= unit file setting has been removed (it is ignored
3951 for backwards compatibility). AmbientCapabilities= and
3952 CapabilityBoundingSet= should be used instead.
3954 * A new special target has been added, initrd-root-device.target,
3955 which creates a synchronization point for dependencies of the root
3956 device in early userspace. Initramfs builders must ensure that this
3957 target is now included in early userspace.
3959 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
3960 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
3961 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
3962 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
3963 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
3964 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
3965 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
3966 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
3967 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
3968 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
3969 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
3970 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
3971 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
3972 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
3973 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
3974 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
3975 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
3976 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
3977 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
3978 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
3979 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
3980 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
3981 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
3982 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
3983 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
3986 — Fairfax, 2016-05-21
3990 * The systemd-resolved DNS resolver service has gained a substantial
3991 set of new features, most prominently it may now act as a DNSSEC
3992 validating stub resolver. DNSSEC mode is currently turned off by
3993 default, but is expected to be turned on by default in one of the
3994 next releases. For now, we invite everybody to test the DNSSEC logic
3995 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
3996 service also gained a full set of D-Bus interfaces, including calls
3997 to configure DNS and DNSSEC settings per link (for use by external
3998 network management software). systemd-resolved and systemd-networkd
3999 now distinguish between "search" and "routing" domains. The former
4000 are used to qualify single-label names, the latter are used purely
4001 for routing lookups within certain domains to specific links.
4002 resolved now also synthesizes RRs for all entries from /etc/hosts.
4004 * The systemd-resolve tool (which is a client utility for
4005 systemd-resolved) has been improved considerably and is now fully
4006 supported and documented. Hence it has moved from /usr/lib/systemd to
4009 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
4012 * The coredump collection logic has been reworked: when a coredump is
4013 collected it is now written to disk, compressed and processed
4014 (including stacktrace extraction) from a new instantiated service
4015 systemd-coredump@.service, instead of directly from the
4016 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
4017 processing large coredumps can take up a substantial amount of
4018 resources and time, and this previously happened entirely outside of
4019 systemd's service supervision. With the new logic the core_pattern
4020 hook only does minimal metadata collection before passing off control
4021 to the new instantiated service, which is configured with a time
4022 limit, a nice level and other settings to minimize negative impact on
4023 the rest of the system. Also note that the new logic will honour the
4024 RLIMIT_CORE setting of the crashed process, which now allows users
4025 and processes to turn off coredumping for their processes by setting
4028 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
4029 and all forked processes by default. Previously, PID 1 would leave
4030 the setting at "0" for all processes, as set by the kernel. Note that
4031 the resource limit traditionally has no effect on the generated
4032 coredumps on the system if the /proc/sys/kernel/core_pattern hook
4033 logic is used. Since the limit is now honoured (see above) its
4034 default has been changed so that the coredumping logic is enabled by
4035 default for all processes, while allowing specific opt-out.
4037 * When the stacktrace is extracted from processes of system users, this
4038 is now done as "systemd-coredump" user, in order to sandbox this
4039 potentially security sensitive parsing operation. (Note that when
4040 processing coredumps of normal users this is done under the user ID
4041 of process that crashed, as before.) Packagers should take notice
4042 that it is now necessary to create the "systemd-coredump" system user
4043 and group at package installation time.
4045 * The systemd-activate socket activation testing tool gained support
4046 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
4047 and --seqpacket switches. It also has been extended to support both
4048 new-style and inetd-style file descriptor passing. Use the new
4049 --inetd switch to request inetd-style file descriptor passing.
4051 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
4052 variable, which takes a boolean value. If set to false, ANSI color
4053 output is disabled in the tools even when run on a terminal that
4056 * The VXLAN support in networkd now supports two new settings
4057 DestinationPort= and PortRange=.
4059 * A new systemd.machine_id= kernel command line switch has been added,
4060 that may be used to set the machine ID in /etc/machine-id if it is
4061 not initialized yet. This command line option has no effect if the
4062 file is already initialized.
4064 * systemd-nspawn gained a new --as-pid2 switch that invokes any
4065 specified command line as PID 2 rather than PID 1 in the
4066 container. In this mode PID 1 is a minimal stub init process that
4067 implements the special POSIX and Linux semantics of PID 1 regarding
4068 signal and child process management. Note that this stub init process
4069 is implemented in nspawn itself and requires no support from the
4070 container image. This new logic is useful to support running
4071 arbitrary commands in the container, as normal processes are
4072 generally not prepared to run as PID 1.
4074 * systemd-nspawn gained a new --chdir= switch for setting the current
4075 working directory for the process started in the container.
4077 * "journalctl /dev/sda" will now output all kernel log messages for
4078 specified device from the current boot, in addition to all devices
4079 that are parents of it. This should make log output about devices
4080 pretty useful, as long as kernel drivers attach enough metadata to
4081 the log messages. (The usual SATA drivers do.)
4083 * The sd-journal API gained two new calls
4084 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
4085 that report whether log data from /run or /var has been found.
4087 * journalctl gained a new switch "--fields" that prints all journal
4088 record field names currently in use in the journal. This is backed
4089 by two new sd-journal API calls sd_journal_enumerate_fields() and
4090 sd_journal_restart_fields().
4092 * Most configurable timeouts in systemd now expect an argument of
4093 "infinity" to turn them off, instead of "0" as before. The semantics
4094 from now on is that a timeout of "0" means "now", and "infinity"
4095 means "never". To maintain backwards compatibility, "0" continues to
4096 turn off previously existing timeout settings.
4098 * "systemctl reload-or-try-restart" has been renamed to "systemctl
4099 try-reload-or-restart" to clarify what it actually does: the "try"
4100 logic applies to both reloading and restarting, not just restarting.
4101 The old name continues to be accepted for compatibility.
4103 * On boot-up, when PID 1 detects that the system clock is behind the
4104 release date of the systemd version in use, the clock is now set
4105 to the latter. Previously, this was already done in timesyncd, in order
4106 to avoid running with clocks set to the various clock epochs such as
4107 1902, 1938 or 1970. With this change the logic is now done in PID 1
4108 in addition to timesyncd during early boot-up, so that it is enforced
4109 before the first process is spawned by systemd. Note that the logic
4110 in timesyncd remains, as it is more comprehensive and ensures
4111 clock monotonicity by maintaining a persistent timestamp file in
4112 /var. Since /var is generally not available in earliest boot or the
4113 initrd, this part of the logic remains in timesyncd, and is not done
4116 * Support for tweaking details in net_cls.class_id through the
4117 NetClass= configuration directive has been removed, as the kernel
4118 people have decided to deprecate that controller in cgroup v2.
4119 Userspace tools such as nftables are moving over to setting rules
4120 that are specific to the full cgroup path of a task, which obsoletes
4121 these controllers anyway. The NetClass= directive is kept around for
4122 legacy compatibility reasons. For a more in-depth description of the
4123 kernel change, please refer to the respective upstream commit:
4125 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
4127 * A new service setting RuntimeMaxSec= has been added that may be used
4128 to specify a maximum runtime for a service. If the timeout is hit, the
4129 service is terminated and put into a failure state.
4131 * A new service setting AmbientCapabilities= has been added. It allows
4132 configuration of additional Linux process capabilities that are
4133 passed to the activated processes. This is only available on very
4136 * The process resource limit settings in service units may now be used
4137 to configure hard and soft limits individually.
4139 * The various libsystemd APIs such as sd-bus or sd-event now publicly
4140 expose support for gcc's __attribute__((cleanup())) C extension.
4141 Specifically, for many object destructor functions alternative
4142 versions have been added that have names suffixed with "p" and take a
4143 pointer to a pointer to the object to destroy, instead of just a
4144 pointer to the object itself. This is useful because these destructor
4145 functions may be used directly as parameters to the cleanup
4146 construct. Internally, systemd has been a heavy user of this GCC
4147 extension for a long time, and with this change similar support is
4148 now available to consumers of the library outside of systemd. Note
4149 that by using this extension in your sources compatibility with old
4150 and strictly ANSI compatible C compilers is lost. However, all gcc or
4151 LLVM versions of recent years support this extension.
4153 * Timer units gained support for a new setting RandomizedDelaySec= that
4154 allows configuring some additional randomized delay to the configured
4155 time. This is useful to spread out timer events to avoid load peaks in
4156 clusters or larger setups.
4158 * Calendar time specifications now support sub-second accuracy.
4160 * Socket units now support listening on SCTP and UDP-lite protocol
4163 * The sd-event API now comes with a full set of man pages.
4165 * Older versions of systemd contained experimental support for
4166 compressing journal files and coredumps with the LZ4 compressor that
4167 was not compatible with the lz4 binary (due to API limitations of the
4168 lz4 library). This support has been removed; only support for files
4169 compatible with the lz4 binary remains. This LZ4 logic is now
4170 officially supported and no longer considered experimental.
4172 * The dkr image import logic has been removed again from importd. dkr's
4173 micro-services focus doesn't fit into the machine image focus of
4174 importd, and quickly got out of date with the upstream dkr API.
4176 * Creation of the /run/lock/lockdev/ directory was dropped from
4177 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
4178 been available for many years. If you still need this, you need to
4179 create your own tmpfiles.d config file with:
4181 d /run/lock/lockdev 0775 root lock -
4183 * The settings StartLimitBurst=, StartLimitInterval=, StartLimitAction=
4184 and RebootArgument= have been moved from the [Service] section of
4185 unit files to [Unit], and they are now supported on all unit types,
4186 not just service units. Of course, systemd will continue to
4187 understand these settings also at the old location, in order to
4188 maintain compatibility.
4190 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
4191 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
4192 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
4193 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
4194 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
4195 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
4196 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
4197 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
4198 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
4199 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
4200 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
4201 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
4202 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
4203 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
4204 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
4205 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
4206 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
4207 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
4208 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
4210 — Berlin, 2016-02-11
4214 * A number of properties previously only settable in unit
4215 files are now also available as properties to set when
4216 creating transient units programmatically via the bus, as it
4217 is exposed with systemd-run's --property=
4218 setting. Specifically, these are: SyslogIdentifier=,
4219 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
4220 EnvironmentFile=, ReadWriteDirectories=,
4221 ReadOnlyDirectories=, InaccessibleDirectories=,
4222 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
4224 * When creating transient services via the bus API it is now
4225 possible to pass in a set of file descriptors to use as
4226 STDIN/STDOUT/STDERR for the invoked process.
4228 * Slice units may now be created transiently via the bus APIs,
4229 similar to the way service and scope units may already be
4230 created transiently.
4232 * Wherever systemd expects a calendar timestamp specification
4233 (like in journalctl's --since= and --until= switches) UTC
4234 timestamps are now supported. Timestamps suffixed with "UTC"
4235 are now considered to be in Universal Time Coordinated
4236 instead of the local timezone. Also, timestamps may now
4237 optionally be specified with sub-second accuracy. Both of
4238 these additions also apply to recurring calendar event
4239 specification, such as OnCalendar= in timer units.
4241 * journalctl gained a new "--sync" switch that asks the
4242 journal daemon to write all so far unwritten log messages to
4243 disk and sync the files, before returning.
4245 * systemd-tmpfiles learned two new line types "q" and "Q" that
4246 operate like "v", but also set up a basic btrfs quota
4247 hierarchy when used on a btrfs file system with quota
4250 * tmpfiles' "v", "q" and "Q" will now create a plain directory
4251 instead of a subvolume (even on a btrfs file system) if the
4252 root directory is a plain directory, and not a
4253 subvolume. This should simplify things with certain chroot()
4254 environments which are not aware of the concept of btrfs
4257 * systemd-detect-virt gained a new --chroot switch to detect
4258 whether execution takes place in a chroot() environment.
4260 * CPUAffinity= now takes CPU index ranges in addition to
4263 * The various memory-related resource limit settings (such as
4264 LimitAS=) now understand the usual K, M, G, ... suffixes to
4265 the base of 1024 (IEC). Similar, the time-related resource
4266 limit settings understand the usual min, h, day, ...
4269 * There's a new system.conf setting DefaultTasksMax= to
4270 control the default TasksMax= setting for services and
4271 scopes running on the system. (TasksMax= is the primary
4272 setting that exposes the "pids" cgroup controller on systemd
4273 and was introduced in the previous systemd release.) The
4274 setting now defaults to 512, which means services that are
4275 not explicitly configured otherwise will only be able to
4276 create 512 processes or threads at maximum, from this
4277 version on. Note that this means that thread- or
4278 process-heavy services might need to be reconfigured to set
4279 TasksMax= to a higher value. It is sufficient to set
4280 TasksMax= in these specific unit files to a higher value, or
4281 even "infinity". Similar, there's now a logind.conf setting
4282 UserTasksMax= that defaults to 4096 and limits the total
4283 number of processes or tasks each user may own
4284 concurrently. nspawn containers also have the TasksMax=
4285 value set by default now, to 8192. Note that all of this
4286 only has an effect if the "pids" cgroup controller is
4287 enabled in the kernel. The general benefit of these changes
4288 should be a more robust and safer system, that provides a
4289 certain amount of per-service fork() bomb protection.
4291 * systemd-nspawn gained the new --network-veth-extra= switch
4292 to define additional and arbitrarily-named virtual Ethernet
4293 links between the host and the container.
4295 * A new service execution setting PassEnvironment= has been
4296 added that allows importing select environment variables
4297 from PID1's environment block into the environment block of
4300 * Timer units gained support for a new RemainAfterElapse=
4301 setting which takes a boolean argument. It defaults to on,
4302 exposing behaviour unchanged to previous releases. If set to
4303 off, timer units are unloaded after they elapsed if they
4304 cannot elapse again. This is particularly useful for
4305 transient timer units, which shall not stay around longer
4306 than until they first elapse.
4308 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
4309 default now (the kernel default is 16). This is beneficial
4310 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
4311 allows substantially larger numbers of queued
4312 datagrams. This should increase the capability of systemd to
4313 parallelize boot-up, as logging and sd_notify() are unlikely
4314 to stall execution anymore. If you need to change the value
4315 from the new defaults, use the usual sysctl.d/ snippets.
4317 * The compression framing format used by the journal or
4318 coredump processing has changed to be in line with what the
4319 official LZ4 tools generate. LZ4 compression support in
4320 systemd was considered unsupported previously, as the format
4321 was not compatible with the normal tools. With this release
4322 this has changed now, and it is hence safe for downstream
4323 distributions to turn it on. While not compressing as well
4324 as the XZ, LZ4 is substantially faster, which makes
4325 it a good default choice for the compression logic in the
4326 journal and in coredump handling.
4328 * Any reference to /etc/mtab has been dropped from
4329 systemd. The file has been obsolete since a while, but
4330 systemd refused to work on systems where it was incorrectly
4331 set up (it should be a symlink or non-existent). Please make
4332 sure to update to util-linux 2.27.1 or newer in conjunction
4333 with this systemd release, which also drops any reference to
4334 /etc/mtab. If you maintain a distribution make sure that no
4335 software you package still references it, as this is a
4336 likely source of bugs. There's also a glibc bug pending,
4337 asking for removal of any reference to this obsolete file:
4339 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
4341 Note that only util-linux versions built with
4342 --enable-libmount-force-mountinfo are supported.
4344 * Support for the ".snapshot" unit type has been removed. This
4345 feature turned out to be little useful and little used, and
4346 has now been removed from the core and from systemctl.
4348 * The dependency types RequiresOverridable= and
4349 RequisiteOverridable= have been removed from systemd. They
4350 have been used only very sparingly to our knowledge and
4351 other options that provide a similar effect (such as
4352 systemctl --mode=ignore-dependencies) are much more useful
4353 and commonly used. Moreover, they were only half-way
4354 implemented as the option to control behaviour regarding
4355 these dependencies was never added to systemctl. By removing
4356 these dependency types the execution engine becomes a bit
4357 simpler. Unit files that use these dependencies should be
4358 changed to use the non-Overridable dependency types
4359 instead. In fact, when parsing unit files with these
4360 options, that's what systemd will automatically convert them
4361 too, but it will also warn, asking users to fix the unit
4362 files accordingly. Removal of these dependency types should
4363 only affect a negligible number of unit files in the wild.
4365 * Behaviour of networkd's IPForward= option changed
4366 (again). It will no longer maintain a per-interface setting,
4367 but propagate one way from interfaces where this is enabled
4368 to the global kernel setting. The global setting will be
4369 enabled when requested by a network that is set up, but
4370 never be disabled again. This change was made to make sure
4371 IPv4 and IPv6 behaviour regarding packet forwarding is
4372 similar (as the Linux IPv6 stack does not support
4373 per-interface control of this setting) and to minimize
4376 * In unit files the behaviour of %u, %U, %h, %s has
4377 changed. These specifiers will now unconditionally resolve
4378 to the various user database fields of the user that the
4379 systemd instance is running as, instead of the user
4380 configured in the specific unit via User=. Note that this
4381 effectively doesn't change much, as resolving of these
4382 specifiers was already turned off in the --system instance
4383 of systemd, as we cannot do NSS lookups from PID 1. In the
4384 --user instance of systemd these specifiers where correctly
4385 resolved, but hardly made any sense, since the user instance
4386 lacks privileges to do user switches anyway, and User= is
4387 hence useless. Moreover, even in the --user instance of
4388 systemd behaviour was awkward as it would only take settings
4389 from User= assignment placed before the specifier into
4390 account. In order to unify and simplify the logic around
4391 this the specifiers will now always resolve to the
4392 credentials of the user invoking the manager (which in case
4393 of PID 1 is the root user).
4395 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
4396 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
4397 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
4398 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
4399 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
4400 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
4401 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
4402 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
4403 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
4404 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
4405 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
4406 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
4407 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
4408 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
4411 — Berlin, 2015-11-18
4415 * systemd now depends on util-linux v2.27. More specifically,
4416 the newly added mount monitor feature in libmount now
4417 replaces systemd's former own implementation.
4419 * libmount mandates /etc/mtab not to be regular file, and
4420 systemd now enforces this condition at early boot.
4421 /etc/mtab has been deprecated and warned about for a very
4422 long time, so systems running systemd should already have
4423 stopped having this file around as anything else than a
4424 symlink to /proc/self/mounts.
4426 * Support for the "pids" cgroup controller has been added. It
4427 allows accounting the number of tasks in a cgroup and
4428 enforcing limits on it. This adds two new setting
4429 TasksAccounting= and TasksMax= to each unit, as well as a
4430 global option DefaultTasksAccounting=.
4432 * Support for the "net_cls" cgroup controller has been added.
4433 It allows assigning a net class ID to each task in the
4434 cgroup, which can then be used in firewall rules and traffic
4435 shaping configurations. Note that the kernel netfilter net
4436 class code does not currently work reliably for ingress
4437 packets on unestablished sockets.
4439 This adds a new config directive called NetClass= to CGroup
4440 enabled units. Allowed values are positive numbers for fixed
4441 assignments and "auto" for picking a free value
4444 * 'systemctl is-system-running' now returns 'offline' if the
4445 system is not booted with systemd. This command can now be
4446 used as a substitute for 'systemd-notify --booted'.
4448 * Watchdog timeouts have been increased to 3 minutes for all
4449 in-tree service files. Apparently, disk IO issues are more
4450 frequent than we hoped, and user reported >1 minute waiting
4453 * 'machine-id-commit' functionality has been merged into
4454 'machine-id-setup --commit'. The separate binary has been
4457 * The WorkingDirectory= directive in unit files may now be set
4458 to the special value '~'. In this case, the working
4459 directory is set to the home directory of the user
4460 configured in User=.
4462 * "machinectl shell" will now open the shell in the home
4463 directory of the selected user by default.
4465 * The CrashChVT= configuration file setting is renamed to
4466 CrashChangeVT=, following our usual logic of not
4467 abbreviating unnecessarily. The old directive is still
4468 supported for compat reasons. Also, this directive now takes
4469 an integer value between 1 and 63, or a boolean value. The
4470 formerly supported '-1' value for disabling stays around for
4473 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
4474 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
4475 RootDirectory= properties can now be set for transient
4478 * The systemd-analyze tool gained a new "set-log-target" verb
4479 to change the logging target the system manager logs to
4480 dynamically during runtime. This is similar to how
4481 "systemd-analyze set-log-level" already changes the log
4484 * In nspawn /sys is now mounted as tmpfs, with only a selected
4485 set of subdirectories mounted in from the real sysfs. This
4486 enhances security slightly, and is useful for ensuring user
4487 namespaces work correctly.
4489 * Support for USB FunctionFS activation has been added. This
4490 allows implementation of USB gadget services that are
4491 activated as soon as they are requested, so that they don't
4492 have to run continuously, similar to classic socket
4495 * The "systemctl exit" command now optionally takes an
4496 additional parameter that sets the exit code to return from
4497 the systemd manager when exiting. This is only relevant when
4498 running the systemd user instance, or when running the
4499 system instance in a container.
4501 * sd-bus gained the new API calls sd_bus_path_encode_many()
4502 and sd_bus_path_decode_many() that allow easy encoding and
4503 decoding of multiple identifier strings inside a D-Bus
4504 object path. Another new call sd_bus_default_flush_close()
4505 has been added to flush and close per-thread default
4508 * systemd-cgtop gained support for a -M/--machine= switch to
4509 show the control groups within a certain container only.
4511 * "systemctl kill" gained support for an optional --fail
4512 switch. If specified the requested operation will fail of no
4513 processes have been killed, because the unit had no
4514 processes attached, or similar.
4516 * A new systemd.crash_reboot=1 kernel command line option has
4517 been added that triggers a reboot after crashing. This can
4518 also be set through CrashReboot= in systemd.conf.
4520 * The RuntimeDirectory= setting now understands unit
4521 specifiers like %i or %f.
4523 * A new (still internal) library API sd-ipv4acd has been added,
4524 that implements address conflict detection for IPv4. It's
4525 based on code from sd-ipv4ll, and will be useful for
4526 detecting DHCP address conflicts.
4528 * File descriptors passed during socket activation may now be
4529 named. A new API sd_listen_fds_with_names() is added to
4530 access the names. The default names may be overridden,
4531 either in the .socket file using the FileDescriptorName=
4532 parameter, or by passing FDNAME= when storing the file
4533 descriptors using sd_notify().
4535 * systemd-networkd gained support for:
4537 - Setting the IPv6 Router Advertisement settings via
4538 IPv6AcceptRouterAdvertisements= in .network files.
4540 - Configuring the HelloTimeSec=, MaxAgeSec= and
4541 ForwardDelaySec= bridge parameters in .netdev files.
4543 - Configuring PreferredSource= for static routes in
4546 * The "ask-password" framework used to query for LUKS harddisk
4547 passwords or SSL passwords during boot gained support for
4548 caching passwords in the kernel keyring, if it is
4549 available. This makes sure that the user only has to type in
4550 a passphrase once if there are multiple objects to unlock
4551 with the same one. Previously, such password caching was
4552 available only when Plymouth was used; this moves the
4553 caching logic into the systemd codebase itself. The
4554 "systemd-ask-password" utility gained a new --keyname=
4555 switch to control which kernel keyring key to use for
4556 caching a password in. This functionality is also useful for
4557 enabling display managers such as gdm to automatically
4558 unlock the user's GNOME keyring if its passphrase, the
4559 user's password and the harddisk password are the same, if
4560 gdm-autologin is used.
4562 * When downloading tar or raw images using "machinectl
4563 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
4564 file is now also downloaded, if it is available and stored
4565 next to the image file.
4567 * Units of type ".socket" gained a new boolean setting
4568 Writable= which is only useful in conjunction with
4569 ListenSpecial=. If true, enables opening the specified
4570 special file in O_RDWR mode rather than O_RDONLY mode.
4572 * systemd-rfkill has been reworked to become a singleton
4573 service that is activated through /dev/rfkill on each rfkill
4574 state change and saves the settings to disk. This way,
4575 systemd-rfkill is now compatible with devices that exist
4576 only intermittendly, and even restores state if the previous
4577 system shutdown was abrupt rather than clean.
4579 * The journal daemon gained support for vacuuming old journal
4580 files controlled by the number of files that shall remain,
4581 in addition to the already existing control by size and by
4582 date. This is useful as journal interleaving performance
4583 degrades with too many separate journal files, and allows
4584 putting an effective limit on them. The new setting defaults
4585 to 100, but this may be changed by setting SystemMaxFiles=
4586 and RuntimeMaxFiles= in journald.conf. Also, the
4587 "journalctl" tool gained the new --vacuum-files= switch to
4588 manually vacuum journal files to leave only the specified
4589 number of files in place.
4591 * udev will now create /dev/disk/by-path links for ATA devices
4592 on kernels where that is supported.
4594 * Galician, Serbian, Turkish and Korean translations were added.
4596 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
4597 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
4598 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
4599 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
4600 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
4601 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
4602 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
4603 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
4604 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
4605 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
4606 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
4607 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
4608 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
4609 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
4610 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
4611 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
4612 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
4613 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
4615 — Berlin, 2015-10-07
4619 * The DHCP implementation of systemd-networkd gained a set of
4622 - The DHCP server now supports emitting DNS and NTP
4623 information. It may be enabled and configured via
4624 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
4625 and NTP information is enabled, but no servers are
4626 configured, the corresponding uplink information (if there
4627 is any) is propagated.
4629 - Server and client now support transmission and reception
4630 of timezone information. It can be configured via the
4631 newly introduced network options UseTimezone=,
4632 EmitTimezone=, and Timezone=. Transmission of timezone
4633 information is enabled between host and containers by
4634 default now: the container will change its local timezone
4635 to what the host has set.
4637 - Lease timeouts can now be configured via
4638 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
4640 - The DHCP server improved on the stability of
4641 leases. Clients are more likely to get the same lease
4642 information back, even if the server loses state.
4644 - The DHCP server supports two new configuration options to
4645 control the lease address pool metrics, PoolOffset= and
4648 * The encapsulation limit of tunnels in systemd-networkd may
4649 now be configured via 'EncapsulationLimit='. It allows
4650 modifying the maximum additional levels of encapsulation
4651 that are permitted to be prepended to a packet.
4653 * systemd now supports the concept of user buses replacing
4654 session buses, if used with dbus-1.10 (and enabled via dbus
4655 --enable-user-session). It previously only supported this on
4656 kdbus-enabled systems, and this release expands this to
4657 'dbus-daemon' systems.
4659 * systemd-networkd now supports predictable interface names
4662 * systemd now optionally supports the new Linux kernel
4663 "unified" control group hierarchy. If enabled via the kernel
4664 command-line option 'systemd.unified_cgroup_hierarchy=1',
4665 systemd will try to mount the unified cgroup hierarchy
4666 directly on /sys/fs/cgroup. If not enabled, or not
4667 available, systemd will fall back to the legacy cgroup
4668 hierarchy setup, as before. Host system and containers can
4669 mix and match legacy and unified hierarchies as they
4670 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
4671 environment variable to individually select the hierarchy to
4672 use for executed containers. By default, nspawn will use the
4673 unified hierarchy for the containers if the host uses the
4674 unified hierarchy, and the legacy hierarchy otherwise.
4675 Please note that at this point the unified hierarchy is an
4676 experimental kernel feature and is likely to change in one
4677 of the next kernel releases. Therefore, it should not be
4678 enabled by default in downstream distributions yet. The
4679 minimum required kernel version for the unified hierarchy to
4680 work is 4.2. Note that when the unified hierarchy is used
4681 for the first time delegated access to controllers is
4682 safe. Because of this systemd-nspawn containers will get
4683 access to controllers now, as will systemd user
4684 sessions. This means containers and user sessions may now
4685 manage their own resources, partitioning up what the system
4688 * A new special scope unit "init.scope" has been introduced
4689 that encapsulates PID 1 of the system. It may be used to
4690 determine resource usage and enforce resource limits on PID
4691 1 itself. PID 1 hence moved out of the root of the control
4694 * The cgtop tool gained support for filtering out kernel
4695 threads when counting tasks in a control group. Also, the
4696 count of processes is now recursively summed up by
4697 default. Two options -k and --recursive= have been added to
4698 revert to old behaviour. The tool has also been updated to
4699 work correctly in containers now.
4701 * systemd-nspawn's --bind= and --bind-ro= options have been
4702 extended to allow creation of non-recursive bind mounts.
4704 * libsystemd gained two new calls sd_pid_get_cgroup() and
4705 sd_peer_get_cgroup() which return the control group path of
4706 a process or peer of a connected AF_UNIX socket. This
4707 function call is particularly useful when implementing
4708 delegated subtrees support in the control group hierarchy.
4710 * The "sd-event" event loop API of libsystemd now supports
4711 correct dequeuing of real-time signals, without losing
4714 * When systemd requests a polkit decision when managing units it
4715 will now add additional fields to the request, including unit
4716 name and desired operation. This enables more powerful polkit
4717 policies, that make decisions depending on these parameters.
4719 * nspawn learnt support for .nspawn settings files, that may
4720 accompany the image files or directories of containers, and
4721 may contain additional settings for the container. This is
4722 an alternative to configuring container parameters via the
4723 nspawn command line.
4725 Contributions from: Cristian Rodríguez, Daniel Mack, David
4726 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
4727 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
4728 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
4729 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
4730 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
4731 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
4732 Andersen, Tom Gundersen, Torstein Husebø
4734 — Berlin, 2015-09-08
4738 * machinectl gained a new verb 'shell' which opens a fresh
4739 shell on the target container or the host. It is similar to
4740 the existing 'login' command of machinectl, but spawns the
4741 shell directly without prompting for username or
4742 password. The pseudo machine '.host' now refers to the local
4743 host and is used by default. Hence, 'machinectl shell' can
4744 be used as replacement for 'su -' which spawns a session as
4745 a fresh systemd unit in a way that is fully isolated from
4746 the originating session.
4748 * systemd-networkd learned to cope with private-zone DHCP
4749 options and allows other programs to query the values.
4751 * SELinux access control when enabling/disabling units is no
4752 longer enforced with this release. The previous implementation
4753 was incorrect, and a new corrected implementation is not yet
4754 available. As unit file operations are still protected via
4755 polkit and D-Bus policy this is not a security problem. Yet,
4756 distributions which care about optimal SELinux support should
4757 probably not stabilize on this release.
4759 * sd-bus gained support for matches of type "arg0has=", that
4760 test for membership of strings in string arrays sent in bus
4763 * systemd-resolved now dumps the contents of its DNS and LLMNR
4764 caches to the logs on reception of the SIGUSR1 signal. This
4765 is useful to debug DNS behaviour.
4767 * The coredumpctl tool gained a new --directory= option to
4768 operate on journal files in a specific directory.
4770 * "systemctl reboot" and related commands gained a new
4771 "--message=" option which may be used to set a free-text
4772 wall message when shutting down or rebooting the
4773 system. This message is also logged, which is useful for
4774 figuring out the reason for a reboot or shutdown a
4777 * The "systemd-resolve-host" tool's -i switch now takes
4778 network interface numbers as alternative to interface names.
4780 * A new unit file setting for services has been introduced:
4781 UtmpMode= allows configuration of how precisely systemd
4782 handles utmp and wtmp entries for the service if this is
4783 enabled. This allows writing services that appear similar to
4784 user sessions in the output of the "w", "who", "last" and
4787 * systemd-resolved will now locally synthesize DNS resource
4788 records for the "localhost" and "gateway" domains as well as
4789 the local hostname. This should ensure that clients querying
4790 RRs via resolved will get similar results as those going via
4791 NSS, if nss-myhostname is enabled.
4793 Contributions from: Alastair Hughes, Alex Crawford, Daniel
4794 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
4795 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
4796 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
4797 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
4798 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
4799 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
4800 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
4801 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
4802 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
4803 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
4804 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
4806 — Berlin, 2015-08-27
4810 * The systemd-efi-boot-generator functionality was merged into
4811 systemd-gpt-auto-generator.
4813 * systemd-networkd now supports Group Policy for vxlan
4814 devices. It can be enabled via the new boolean configuration
4815 option called 'GroupPolicyExtension='.
4817 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
4818 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
4819 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
4821 — Berlin, 2015-07-31
4825 * The python-systemd code has been removed from the systemd repository.
4826 A new repository has been created which accommodates the code from
4827 now on, and we kindly ask distributions to create a separate package
4828 for this: https://github.com/systemd/python-systemd
4830 * The systemd daemon will now reload its main configuration
4831 (/etc/systemd/system.conf) on daemon-reload.
4833 * sd-dhcp now exposes vendor specific extensions via
4834 sd_dhcp_lease_get_vendor_specific().
4836 * systemd-networkd gained a number of new configuration options.
4838 - A new boolean configuration option for TAP devices called
4839 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
4840 device, thus allowing to send and receive GSO packets.
4842 - A new tunnel configuration option called 'CopyDSCP='.
4843 If enabled, the DSCP field of ip6 tunnels is copied into the
4844 decapsulated packet.
4846 - A set of boolean bridge configuration options were added.
4847 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
4848 and 'UnicastFlood=' are now parsed by networkd and applied to the
4849 respective bridge link device via the respective IFLA_BRPORT_*
4852 - A new string configuration option to override the hostname sent
4853 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
4854 is true, networkd will use the configured hostname instead of the
4855 system hostname when sending DHCP requests.
4857 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
4858 networkd will configure the IPv6 flow-label of the tunnel device
4859 according to RFC2460.
4861 - The 'macvtap' virtual network devices are now supported, similar to
4862 the already supported 'macvlan' devices.
4864 * systemd-resolved now implements RFC5452 to improve resilience against
4865 cache poisoning. Additionally, source port randomization is enabled
4866 by default to further protect against DNS spoofing attacks.
4868 * nss-mymachines now supports translating UIDs and GIDs of running
4869 containers with user-namespaces enabled. If a container 'foo'
4870 translates a host uid 'UID' to the container uid 'TUID', then
4871 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
4872 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
4873 mapped as 'vg-foo-TGID'.
4875 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
4876 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
4877 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
4878 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
4879 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
4880 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
4881 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
4882 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
4883 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
4884 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
4886 — Berlin, 2015-07-29
4890 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
4891 There are no known issues with current sysfs, and udev does not need
4892 or should be used to work around such bugs.
4894 * udev does no longer enable USB HID power management. Several reports
4895 indicate, that some devices cannot handle that setting.
4897 * The udev accelerometer helper was removed. The functionality
4898 is now fully included in iio-sensor-proxy. But this means,
4899 older iio-sensor-proxy versions will no longer provide
4900 accelerometer/orientation data with this systemd version.
4901 Please upgrade iio-sensor-proxy to version 1.0.
4903 * networkd gained a new configuration option IPv6PrivacyExtensions=
4904 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
4905 for Stateless Address") on selected networks.
4907 * For the sake of fewer build-time dependencies and less code in the
4908 main repository, the python bindings are about to be removed in the
4909 next release. A new repository has been created which accommodates
4910 the code from now on, and we kindly ask distributions to create a
4911 separate package for this. The removal will take place in v223.
4913 https://github.com/systemd/python-systemd
4915 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
4916 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
4917 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
4918 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
4919 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
4920 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
4921 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
4922 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
4923 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
4924 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
4926 — Berlin, 2015-07-07
4930 * The sd-bus.h and sd-event.h APIs have now been declared
4931 stable and have been added to the official interface of
4932 libsystemd.so. sd-bus implements an alternative D-Bus client
4933 library, that is relatively easy to use, very efficient and
4934 supports both classic D-Bus as well as kdbus as transport
4935 backend. sd-event is a generic event loop abstraction that
4936 is built around Linux epoll, but adds features such as event
4937 prioritization or efficient timer handling. Both APIs are good
4938 choices for C programs looking for a bus and/or event loop
4939 implementation that is minimal and does not have to be
4940 portable to other kernels.
4942 * kdbus support is no longer compile-time optional. It is now
4943 always built-in. However, it can still be disabled at
4944 runtime using the kdbus=0 kernel command line setting, and
4945 that setting may be changed to default to off, by specifying
4946 --disable-kdbus at build-time. Note though that the kernel
4947 command line setting has no effect if the kdbus.ko kernel
4948 module is not installed, in which case kdbus is (obviously)
4949 also disabled. We encourage all downstream distributions to
4950 begin testing kdbus by adding it to the kernel images in the
4951 development distributions, and leaving kdbus support in
4954 * The minimal required util-linux version has been bumped to
4957 * Support for chkconfig (--enable-chkconfig) was removed in
4958 favor of calling an abstraction tool
4959 /lib/systemd/systemd-sysv-install. This needs to be
4960 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
4961 in README for details.
4963 * If there's a systemd unit and a SysV init script for the
4964 same service name, and the user executes "systemctl enable"
4965 for it (or a related call), then this will now enable both
4966 (or execute the related operation on both), not just the
4969 * The libudev API documentation has been converted from gtkdoc
4972 * gudev has been removed from the systemd tree, it is now an
4975 * The systemd-cgtop tool learnt a new --raw switch to generate
4976 "raw" (machine parsable) output.
4978 * networkd's IPForwarding= .network file setting learnt the
4979 new setting "kernel", which ensures that networkd does not
4980 change the IP forwarding sysctl from the default kernel
4983 * The systemd-logind bus API now exposes a new boolean
4984 property "Docked" that reports whether logind considers the
4985 system "docked", i.e. connected to a docking station or not.
4987 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
4988 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
4989 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
4990 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
4991 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
4992 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
4993 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
4994 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
4995 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
4996 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
4997 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
4998 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
4999 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
5000 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
5001 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
5002 Fink, Zbigniew Jędrzejewski-Szmek
5004 — Berlin, 2015-06-19
5008 * The gudev library has been extracted into a separate repository
5009 available at: https://git.gnome.org/browse/libgudev/
5010 It is now managed as part of the Gnome project. Distributions
5011 are recommended to pass --disable-gudev to systemd and use
5012 gudev from the Gnome project instead. gudev is still included
5013 in systemd, for now. It will be removed soon, though. Please
5014 also see the announcement-thread on systemd-devel:
5015 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
5017 * systemd now exposes a CPUUsageNSec= property for each
5018 service unit on the bus, that contains the overall consumed
5019 CPU time of a service (the sum of what each process of the
5020 service consumed). This value is only available if
5021 CPUAccounting= is turned on for a service, and is then shown
5022 in the "systemctl status" output.
5024 * Support for configuring alternative mappings of the old SysV
5025 runlevels to systemd targets has been removed. They are now
5026 hardcoded in a way that runlevels 2, 3, 4 all map to
5027 multi-user.target and 5 to graphical.target (which
5028 previously was already the default behaviour).
5030 * The auto-mounter logic gained support for mount point
5031 expiry, using a new TimeoutIdleSec= setting in .automount
5032 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
5034 * The EFI System Partition (ESP) as mounted to /boot by
5035 systemd-efi-boot-generator will now be unmounted
5036 automatically after 2 minutes of not being used. This should
5037 minimize the risk of ESP corruptions.
5039 * New /etc/fstab options x-systemd.requires= and
5040 x-systemd.requires-mounts-for= are now supported to express
5041 additional dependencies for mounts. This is useful for
5042 journalling file systems that support external journal
5043 devices or overlay file systems that require underlying file
5044 systems to be mounted.
5046 * systemd does not support direct live-upgrades (via systemctl
5047 daemon-reexec) from versions older than v44 anymore. As no
5048 distribution we are aware of shipped such old versions in a
5049 stable release this should not be problematic.
5051 * When systemd forks off a new per-connection service instance
5052 it will now set the $REMOTE_ADDR environment variable to the
5053 remote IP address, and $REMOTE_PORT environment variable to
5054 the remote IP port. This behaviour is similar to the
5055 corresponding environment variables defined by CGI.
5057 * systemd-networkd gained support for uplink failure
5058 detection. The BindCarrier= option allows binding interface
5059 configuration dynamically to the link sense of other
5060 interfaces. This is useful to achieve behaviour like in
5063 * systemd-networkd gained support for configuring the DHCP
5064 client identifier to use when requesting leases.
5066 * systemd-networkd now has a per-network UseNTP= option to
5067 configure whether NTP server information acquired via DHCP
5068 is passed on to services like systemd-timesyncd.
5070 * systemd-networkd gained support for vti6 tunnels.
5072 * Note that systemd-networkd manages the sysctl variable
5073 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
5074 it is configured for since v219. The variable controls IP
5075 forwarding, and is a per-interface alternative to the global
5076 /proc/sys/net/ipv[46]/ip_forward. This setting is
5077 configurable in the IPForward= option, which defaults to
5078 "no". This means if networkd is used for an interface it is
5079 no longer sufficient to set the global sysctl option to turn
5080 on IP forwarding! Instead, the .network file option
5081 IPForward= needs to be turned on! Note that the
5082 implementation of this behaviour was broken in v219 and has
5085 * Many bonding and vxlan options are now configurable in
5088 * systemd-nspawn gained a new --property= setting to set unit
5089 properties for the container scope. This is useful for
5090 setting resource parameters (e.g. "CPUShares=500") on
5091 containers started from the command line.
5093 * systemd-nspawn gained a new --private-users= switch to make
5094 use of user namespacing available on recent Linux kernels.
5096 * systemd-nspawn may now be called as part of a shell pipeline
5097 in which case the pipes used for stdin and stdout are passed
5098 directly to the process invoked in the container, without
5099 indirection via a pseudo tty.
5101 * systemd-nspawn gained a new switch to control the UNIX
5102 signal to use when killing the init process of the container
5105 * systemd-nspawn gained a new --overlay= switch for mounting
5106 overlay file systems into the container using the new kernel
5109 * When a container image is imported via systemd-importd and
5110 the host file system is not btrfs, a loopback block device
5111 file is created in /var/lib/machines.raw with a btrfs file
5112 system inside. It is then mounted to /var/lib/machines to
5113 enable btrfs features for container management. The loopback
5114 file and btrfs file system is grown as needed when container
5115 images are imported via systemd-importd.
5117 * systemd-machined/systemd-importd gained support for btrfs
5118 quota, to enforce container disk space limits on disk. This
5119 is exposed in "machinectl set-limit".
5121 * systemd-importd now can import containers from local .tar,
5122 .raw and .qcow2 images, and export them to .tar and .raw. It
5123 can also import dkr v2 images now from the network (on top
5126 * systemd-importd gained support for verifying downloaded
5127 images with gpg2 (previously only gpg1 was supported).
5129 * systemd-machined, systemd-logind, systemd: most bus calls are
5130 now accessible to unprivileged processes via polkit. Also,
5131 systemd-logind will now allow users to kill their own sessions
5132 without further privileges or authorization.
5134 * systemd-shutdownd has been removed. This service was
5135 previously responsible for implementing scheduled shutdowns
5136 as exposed in /usr/bin/shutdown's time parameter. This
5137 functionality has now been moved into systemd-logind and is
5138 accessible via a bus interface.
5140 * "systemctl reboot" gained a new switch --firmware-setup that
5141 can be used to reboot into the EFI firmware setup, if that
5142 is available. systemd-logind now exposes an API on the bus
5143 to trigger such reboots, in case graphical desktop UIs want
5144 to cover this functionality.
5146 * "systemctl enable", "systemctl disable" and "systemctl mask"
5147 now support a new "--now" switch. If specified the units
5148 that are enabled will also be started, and the ones
5149 disabled/masked also stopped.
5151 * The Gummiboot EFI boot loader tool has been merged into
5152 systemd, and renamed to "systemd-boot". The bootctl tool has been
5153 updated to support systemd-boot.
5155 * An EFI kernel stub has been added that may be used to create
5156 kernel EFI binaries that contain not only the actual kernel,
5157 but also an initrd, boot splash, command line and OS release
5158 information. This combined binary can then be signed as a
5159 single image, so that the firmware can verify it all in one
5160 step. systemd-boot has special support for EFI binaries created
5161 like this and can extract OS release information from them
5162 and show them in the boot menu. This functionality is useful
5163 to implement cryptographically verified boot schemes.
5165 * Optional support has been added to systemd-fsck to pass
5166 fsck's progress report to an AF_UNIX socket in the file
5169 * udev will no longer create device symlinks for all block
5170 devices by default. A blacklist for excluding special block
5171 devices from this logic has been turned into a whitelist
5172 that requires picking block devices explicitly that require
5175 * A new (currently still internal) API sd-device.h has been
5176 added to libsystemd. This modernized API is supposed to
5177 replace libudev eventually. In fact, already much of libudev
5178 is now just a wrapper around sd-device.h.
5180 * A new hwdb database for storing metadata about pointing
5181 stick devices has been added.
5183 * systemd-tmpfiles gained support for setting file attributes
5184 similar to the "chattr" tool with new 'h' and 'H' lines.
5186 * systemd-journald will no longer unconditionally set the
5187 btrfs NOCOW flag on new journal files. This is instead done
5188 with tmpfiles snippet using the new 'h' line type. This
5189 allows easy disabling of this logic, by masking the
5190 journal-nocow.conf tmpfiles file.
5192 * systemd-journald will now translate audit message types to
5193 human readable identifiers when writing them to the
5194 journal. This should improve readability of audit messages.
5196 * The LUKS logic gained support for the offset= and skip=
5197 options in /etc/crypttab, as previously implemented by
5200 * /usr/lib/os-release gained a new optional field VARIANT= for
5201 distributions that support multiple variants (such as a
5202 desktop edition, a server edition, ...)
5204 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
5205 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
5206 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
5207 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
5208 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
5209 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
5210 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
5211 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
5212 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
5213 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
5214 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
5215 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
5216 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
5217 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
5218 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
5219 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
5220 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
5221 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
5222 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
5223 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
5224 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
5225 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
5226 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
5227 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
5228 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
5229 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
5230 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
5232 — Berlin, 2015-05-22
5236 * Introduce a new API "sd-hwdb.h" for querying the hardware
5237 metadata database. With this minimal interface one can query
5238 and enumerate the udev hwdb, decoupled from the old libudev
5239 library. libudev's interface for this is now only a wrapper
5240 around sd-hwdb. A new tool systemd-hwdb has been added to
5241 interface with and update the database.
5243 * When any of systemd's tools copies files (for example due to
5244 tmpfiles' C lines) a btrfs reflink will attempted first,
5245 before bytewise copying is done.
5247 * systemd-nspawn gained a new --ephemeral switch. When
5248 specified a btrfs snapshot is taken of the container's root
5249 directory, and immediately removed when the container
5250 terminates again. Thus, a container can be started whose
5251 changes never alter the container's root directory, and are
5252 lost on container termination. This switch can also be used
5253 for starting a container off the root file system of the
5254 host without affecting the host OS. This switch is only
5255 available on btrfs file systems.
5257 * systemd-nspawn gained a new --template= switch. It takes the
5258 path to a container tree to use as template for the tree
5259 specified via --directory=, should that directory be
5260 missing. This allows instantiating containers dynamically,
5261 on first run. This switch is only available on btrfs file
5264 * When a .mount unit refers to a mount point on which multiple
5265 mounts are stacked, and the .mount unit is stopped all of
5266 the stacked mount points will now be unmounted until no
5267 mount point remains.
5269 * systemd now has an explicit notion of supported and
5270 unsupported unit types. Jobs enqueued for unsupported unit
5271 types will now fail with an "unsupported" error code. More
5272 specifically .swap, .automount and .device units are not
5273 supported in containers, .busname units are not supported on
5274 non-kdbus systems. .swap and .automount are also not
5275 supported if their respective kernel compile time options
5278 * machinectl gained support for two new "copy-from" and
5279 "copy-to" commands for copying files from a running
5280 container to the host or vice versa.
5282 * machinectl gained support for a new "bind" command to bind
5283 mount host directories into local containers. This is
5284 currently only supported for nspawn containers.
5286 * networkd gained support for configuring bridge forwarding
5287 database entries (fdb) from .network files.
5289 * A new tiny daemon "systemd-importd" has been added that can
5290 download container images in tar, raw, qcow2 or dkr formats,
5291 and make them available locally in /var/lib/machines, so
5292 that they can run as nspawn containers. The daemon can GPG
5293 verify the downloads (not supported for dkr, since it has no
5294 provisions for verifying downloads). It will transparently
5295 decompress bz2, xz, gzip compressed downloads if necessary,
5296 and restore sparse files on disk. The daemon uses privilege
5297 separation to ensure the actual download logic runs with
5298 fewer privileges than the daemon itself. machinectl has
5299 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
5300 make the functionality of importd available to the
5301 user. With this in place the Fedora and Ubuntu "Cloud"
5302 images can be downloaded and booted as containers unmodified
5303 (the Fedora images lack the appropriate GPG signature files
5304 currently, so they cannot be verified, but this will change
5305 soon, hopefully). Note that downloading images is currently
5306 only fully supported on btrfs.
5308 * machinectl is now able to list container images found in
5309 /var/lib/machines, along with some metadata about sizes of
5310 disk and similar. If the directory is located on btrfs and
5311 quota is enabled, this includes quota display. A new command
5312 "image-status" has been added that shows additional
5313 information about images.
5315 * machinectl is now able to clone container images
5316 efficiently, if the underlying file system (btrfs) supports
5317 it, with the new "machinectl clone" command. It also
5318 gained commands for renaming and removing images, as well as
5319 marking them read-only or read-write (supported also on
5320 legacy file systems).
5322 * networkd gained support for collecting LLDP network
5323 announcements, from hardware that supports this. This is
5324 shown in networkctl output.
5326 * systemd-run gained support for a new -t (--pty) switch for
5327 invoking a binary on a pty whose input and output is
5328 connected to the invoking terminal. This allows executing
5329 processes as system services while interactively
5330 communicating with them via the terminal. Most interestingly
5331 this is supported across container boundaries. Invoking
5332 "systemd-run -t /bin/bash" is an alternative to running a
5333 full login session, the difference being that the former
5334 will not register a session, nor go through the PAM session
5337 * tmpfiles gained support for a new "v" line type for creating
5338 btrfs subvolumes. If the underlying file system is a legacy
5339 file system, this automatically degrades to creating a
5340 normal directory. Among others /var/lib/machines is now
5341 created like this at boot, should it be missing.
5343 * The directory /var/lib/containers/ has been deprecated and
5344 been replaced by /var/lib/machines. The term "machines" has
5345 been used in the systemd context as generic term for both
5346 VMs and containers, and hence appears more appropriate for
5347 this, as the directory can also contain raw images bootable
5350 * systemd-nspawn when invoked with -M but without --directory=
5351 or --image= is now capable of searching for the container
5352 root directory, subvolume or disk image automatically, in
5353 /var/lib/machines. systemd-nspawn@.service has been updated
5354 to make use of this, thus allowing it to be used for raw
5357 * A new machines.target unit has been introduced that is
5358 supposed to group all containers/VMs invoked as services on
5359 the system. systemd-nspawn@.service has been updated to
5360 integrate with that.
5362 * machinectl gained a new "start" command, for invoking a
5363 container as a service. "machinectl start foo" is mostly
5364 equivalent to "systemctl start systemd-nspawn@foo.service",
5365 but handles escaping in a nicer way.
5367 * systemd-nspawn will now mount most of the cgroupfs tree
5368 read-only into each container, with the exception of the
5369 container's own subtree in the name=systemd hierarchy.
5371 * journald now sets the special FS_NOCOW file flag for its
5372 journal files. This should improve performance on btrfs, by
5373 avoiding heavy fragmentation when journald's write-pattern
5374 is used on COW file systems. It degrades btrfs' data
5375 integrity guarantees for the files to the same levels as for
5376 ext3/ext4 however. This should be OK though as journald does
5377 its own data integrity checks and all its objects are
5378 checksummed on disk. Also, journald should handle btrfs disk
5379 full events a lot more gracefully now, by processing SIGBUS
5380 errors, and not relying on fallocate() anymore.
5382 * When journald detects that journal files it is writing to
5383 have been deleted it will immediately start new journal
5386 * systemd now provides a way to store file descriptors
5387 per-service in PID 1. This is useful for daemons to ensure
5388 that fds they require are not lost during a daemon
5389 restart. The fds are passed to the daemon on the next
5390 invocation in the same way socket activation fds are
5391 passed. This is now used by journald to ensure that the
5392 various sockets connected to all the system's stdout/stderr
5393 are not lost when journald is restarted. File descriptors
5394 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
5395 an extension to sd_notify(). Note that a limit is enforced
5396 on the number of fds a service can store in PID 1, and it
5397 defaults to 0, so that no fds may be stored, unless this is
5398 explicitly turned on.
5400 * The default TERM variable to use for units connected to a
5401 terminal, when no other value is explicitly is set is now
5402 vt220 rather than vt102. This should be fairly safe still,
5403 but allows PgUp/PgDn work.
5405 * The /etc/crypttab option header= as known from Debian is now
5408 * "loginctl user-status" and "loginctl session-status" will
5409 now show the last 10 lines of log messages of the
5410 user/session following the status output. Similar,
5411 "machinectl status" will show the last 10 log lines
5412 associated with a virtual machine or container
5413 service. (Note that this is usually not the log messages
5414 done in the VM/container itself, but simply what the
5415 container manager logs. For nspawn this includes all console
5418 * "loginctl session-status" without further argument will now
5419 show the status of the session of the caller. Similar,
5420 "lock-session", "unlock-session", "activate",
5421 "enable-linger", "disable-linger" may now be called without
5422 session/user parameter in which case they apply to the
5423 caller's session/user.
5425 * An X11 session scriptlet is now shipped that uploads
5426 $DISPLAY and $XAUTHORITY into the environment of the systemd
5427 --user daemon if a session begins. This should improve
5428 compatibility with X11 enabled applications run as systemd
5431 * Generators are now subject to masking via /etc and /run, the
5432 same way as unit files.
5434 * networkd .network files gained support for configuring
5435 per-link IPv4/IPv6 packet forwarding as well as IPv4
5436 masquerading. This is by default turned on for veth links to
5437 containers, as registered by systemd-nspawn. This means that
5438 nspawn containers run with --network-veth will now get
5439 automatic routed access to the host's networks without any
5440 further configuration or setup, as long as networkd runs on
5443 * systemd-nspawn gained the --port= (-p) switch to expose TCP
5444 or UDP posts of a container on the host. With this in place
5445 it is possible to run containers with private veth links
5446 (--network-veth), and have their functionality exposed on
5447 the host as if their services were running directly on the
5450 * systemd-nspawn's --network-veth switch now gained a short
5451 version "-n", since with the changes above it is now truly
5452 useful out-of-the-box. The systemd-nspawn@.service has been
5453 updated to make use of it too by default.
5455 * systemd-nspawn will now maintain a per-image R/W lock, to
5456 ensure that the same image is not started more than once
5457 writable. (It's OK to run an image multiple times
5458 simultaneously in read-only mode.)
5460 * systemd-nspawn's --image= option is now capable of
5461 dissecting and booting MBR and GPT disk images that contain
5462 only a single active Linux partition. Previously it
5463 supported only GPT disk images with proper GPT type
5464 IDs. This allows running cloud images from major
5465 distributions directly with systemd-nspawn, without
5468 * In addition to collecting mouse dpi data in the udev
5469 hardware database, there's now support for collecting angle
5470 information for mouse scroll wheels. The database is
5471 supposed to guarantee similar scrolling behavior on mice
5472 that it knows about. There's also support for collecting
5473 information about Touchpad types.
5475 * udev's input_id built-in will now also collect touch screen
5476 dimension data and attach it to probed devices.
5478 * /etc/os-release gained support for a Distribution Privacy
5481 * networkd gained support for creating "ipvlan", "gretap",
5482 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
5484 * systemd-tmpfiles gained support for "a" lines for setting
5487 * systemd-nspawn will now mount /tmp in the container to
5488 tmpfs, automatically.
5490 * systemd now exposes the memory.usage_in_bytes cgroup
5491 attribute and shows it for each service in the "systemctl
5492 status" output, if available.
5494 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
5495 immediate reboot is triggered. This useful if shutdown is
5496 hung and is unable to complete, to expedite the
5497 operation. Note that this kind of reboot will still unmount
5498 all file systems, and hence should not result in fsck being
5501 * A .device unit for an optical block device will now be
5502 considered active only when a medium is in the drive. Also,
5503 mount units are now bound to their backing devices thus
5504 triggering automatic unmounting when devices become
5505 unavailable. With this in place systemd will now
5506 automatically unmount left-over mounts when a CD-ROM is
5507 ejected or an USB stick is yanked from the system.
5509 * networkd-wait-online now has support for waiting for
5510 specific interfaces only (with globbing), and for giving up
5511 after a configurable timeout.
5513 * networkd now exits when idle. It will be automatically
5514 restarted as soon as interfaces show up, are removed or
5515 change state. networkd will stay around as long as there is
5516 at least one DHCP state machine or similar around, that keep
5519 * networkd may now configure IPv6 link-local addressing in
5520 addition to IPv4 link-local addressing.
5522 * The IPv6 "token" for use in SLAAC may now be configured for
5523 each .network interface in networkd.
5525 * Routes configured with networkd may now be assigned a scope
5528 * networkd's [Match] sections now support globbing and lists
5529 of multiple space-separated matches per item.
5531 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
5532 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
5533 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
5534 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
5535 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
5536 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
5537 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
5538 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
5539 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
5540 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
5541 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
5542 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
5543 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
5544 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
5545 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
5546 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
5547 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
5548 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
5549 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
5550 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
5551 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
5552 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
5553 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
5554 Hoffmann, Zbigniew Jędrzejewski-Szmek
5556 — Berlin, 2015-02-16
5560 * When querying unit file enablement status (for example via
5561 "systemctl is-enabled"), a new state "indirect" is now known
5562 which indicates that a unit might not be enabled itself, but
5563 another unit listed in its Also= setting might be.
5565 * Similar to the various existing ConditionXYZ= settings for
5566 units, there are now matching AssertXYZ= settings. While
5567 failing conditions cause a unit to be skipped, but its job
5568 to succeed, failing assertions declared like this will cause
5569 a unit start operation and its job to fail.
5571 * hostnamed now knows a new chassis type "embedded".
5573 * systemctl gained a new "edit" command. When used on a unit
5574 file, this allows extending unit files with .d/ drop-in
5575 configuration snippets or editing the full file (after
5576 copying it from /usr/lib to /etc). This will invoke the
5577 user's editor (as configured with $EDITOR), and reload the
5578 modified configuration after editing.
5580 * "systemctl status" now shows the suggested enablement state
5581 for a unit, as declared in the (usually vendor-supplied)
5582 system preset files.
5584 * nss-myhostname will now resolve the single-label host name
5585 "gateway" to the locally configured default IP routing
5586 gateways, ordered by their metrics. This assigns a stable
5587 name to the used gateways, regardless which ones are
5588 currently configured. Note that the name will only be
5589 resolved after all other name sources (if nss-myhostname is
5590 configured properly) and should hence not negatively impact
5591 systems that use the single-label host name "gateway" in
5594 * systemd-inhibit now allows filtering by mode when listing
5597 * Scope and service units gained a new "Delegate" boolean
5598 property, which, when set, allows processes running inside the
5599 unit to further partition resources. This is primarily
5600 useful for systemd user instances as well as container
5603 * journald will now pick up audit messages directly from
5604 the kernel, and log them like any other log message. The
5605 audit fields are split up and fully indexed. This means that
5606 journalctl in many ways is now a (nicer!) alternative to
5607 ausearch, the traditional audit client. Note that this
5608 implements only a minimal audit client. If you want the
5609 special audit modes like reboot-on-log-overflow, please use
5610 the traditional auditd instead, which can be used in
5611 parallel to journald.
5613 * The ConditionSecurity= unit file option now understands the
5614 special string "audit" to check whether auditing is
5617 * journalctl gained two new commands --vacuum-size= and
5618 --vacuum-time= to delete old journal files until the
5619 remaining ones take up no more than the specified size on disk,
5620 or are not older than the specified time.
5622 * A new, native PPPoE library has been added to sd-network,
5623 systemd's library of light-weight networking protocols. This
5624 library will be used in a future version of networkd to
5625 enable PPPoE communication without an external pppd daemon.
5627 * The busctl tool now understands a new "capture" verb that
5628 works similar to "monitor", but writes a packet capture
5629 trace to STDOUT that can be redirected to a file which is
5630 compatible with libcap's capture file format. This can then
5631 be loaded in Wireshark and similar tools to inspect bus
5634 * The busctl tool now understands a new "tree" verb that shows
5635 the object trees of a specific service on the bus, or of all
5638 * The busctl tool now understands a new "introspect" verb that
5639 shows all interfaces and members of objects on the bus,
5640 including their signature and values. This is particularly
5641 useful to get more information about bus objects shown by
5642 the new "busctl tree" command.
5644 * The busctl tool now understands new verbs "call",
5645 "set-property" and "get-property" for invoking bus method
5646 calls, setting and getting bus object properties in a
5649 * busctl gained a new --augment-creds= argument that controls
5650 whether the tool shall augment credential information it
5651 gets from the bus with data from /proc, in a possibly
5654 * nspawn's --link-journal= switch gained two new values
5655 "try-guest" and "try-host" that work like "guest" and
5656 "host", but do not fail if the host has no persistent
5657 journalling enabled. -j is now equivalent to
5658 --link-journal=try-guest.
5660 * macvlan network devices created by nspawn will now have
5661 stable MAC addresses.
5663 * A new SmackProcessLabel= unit setting has been added, which
5664 controls the SMACK security label processes forked off by
5665 the respective unit shall use.
5667 * If compiled with --enable-xkbcommon, systemd-localed will
5668 verify x11 keymap settings by compiling the given keymap. It
5669 will spew out warnings if the compilation fails. This
5670 requires libxkbcommon to be installed.
5672 * When a coredump is collected, a larger number of metadata
5673 fields is now collected and included in the journal records
5674 created for it. More specifically, control group membership,
5675 environment variables, memory maps, working directory,
5676 chroot directory, /proc/$PID/status, and a list of open file
5677 descriptors is now stored in the log entry.
5679 * The udev hwdb now contains DPI information for mice. For
5682 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
5684 * All systemd programs that read standalone configuration
5685 files in /etc now also support a corresponding series of
5686 .conf.d configuration directories in /etc/, /run/,
5687 /usr/local/lib/, /usr/lib/, and (if configured with
5688 --enable-split-usr) /lib/. In particular, the following
5689 configuration files now have corresponding configuration
5690 directories: system.conf user.conf, logind.conf,
5691 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
5692 resolved.conf, timesyncd.conf, journal-remote.conf, and
5693 journal-upload.conf. Note that distributions should use the
5694 configuration directories in /usr/lib/; the directories in
5695 /etc/ are reserved for the system administrator.
5697 * systemd-rfkill will no longer take the rfkill device name
5698 into account when storing rfkill state on disk, as the name
5699 might be dynamically assigned and not stable. Instead, the
5700 ID_PATH udev variable combined with the rfkill type (wlan,
5701 bluetooth, ...) is used.
5703 * A new service systemd-machine-id-commit.service has been
5704 added. When used on systems where /etc is read-only during
5705 boot, and /etc/machine-id is not initialized (but an empty
5706 file), this service will copy the temporary machine ID
5707 created as replacement into /etc after the system is fully
5708 booted up. This is useful for systems that are freshly
5709 installed with a non-initialized machine ID, but should get
5710 a fixed machine ID for subsequent boots.
5712 * networkd's .netdev files now provide a large set of
5713 configuration parameters for VXLAN devices. Similarly, the
5714 bridge port cost parameter is now configurable in .network
5715 files. There's also new support for configuring IP source
5716 routing. networkd .link files gained support for a new
5717 OriginalName= match that is useful to match against the
5718 original interface name the kernel assigned. .network files
5719 may include MTU= and MACAddress= fields for altering the MTU
5720 and MAC address while being connected to a specific network
5723 * The LUKS logic gained supported for configuring
5724 UUID-specific key files. There's also new support for naming
5725 LUKS device from the kernel command line, using the new
5726 luks.name= argument.
5728 * Timer units may now be transiently created via the bus API
5729 (this was previously already available for scope and service
5730 units). In addition it is now possible to create multiple
5731 transient units at the same time with a single bus call. The
5732 "systemd-run" tool has been updated to make use of this for
5733 running commands on a specified time, in at(1)-style.
5735 * tmpfiles gained support for "t" lines, for assigning
5736 extended attributes to files. Among other uses this may be
5737 used to assign SMACK labels to files.
5739 Contributions from: Alin Rauta, Alison Chaiken, Andrej
5740 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
5741 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
5742 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
5743 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
5744 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
5745 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
5746 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
5747 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
5748 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
5749 Michael Biebl, Michael Chapman, Michael Marineau, Michal
5750 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
5751 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
5752 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
5753 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
5754 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
5755 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
5756 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
5758 — Berlin, 2014-12-10
5762 * journalctl gained the new options -t/--identifier= to match
5763 on the syslog identifier (aka "tag"), as well as --utc to
5764 show log timestamps in the UTC timezone. journalctl now also
5765 accepts -n/--lines=all to disable line capping in a pager.
5767 * journalctl gained a new switch, --flush, that synchronously
5768 flushes logs from /run/log/journal to /var/log/journal if
5769 persistent storage is enabled. systemd-journal-flush.service
5770 now waits until the operation is complete.
5772 * Services can notify the manager before they start a reload
5773 (by sending RELOADING=1) or shutdown (by sending
5774 STOPPING=1). This allows the manager to track and show the
5775 internal state of daemons and closes a race condition when
5776 the process is still running but has closed its D-Bus
5779 * Services with Type=oneshot do not have to have any ExecStart
5782 * User units are now loaded also from
5783 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
5784 /run/systemd/user directory that was already previously
5785 supported, but is under the control of the user.
5787 * Job timeouts (i.e. timeouts on the time a job that is
5788 queued stays in the run queue) can now optionally result in
5789 immediate reboot or power-off actions (JobTimeoutAction= and
5790 JobTimeoutRebootArgument=). This is useful on ".target"
5791 units, to limit the maximum time a target remains
5792 undispatched in the run queue, and to trigger an emergency
5793 operation in such a case. This is now used by default to
5794 turn off the system if boot-up (as defined by everything in
5795 basic.target) hangs and does not complete for at least
5796 15min. Also, if power-off or reboot hang for at least 30min
5797 an immediate power-off/reboot operation is triggered. This
5798 functionality is particularly useful to increase reliability
5799 on embedded devices, but also on laptops which might
5800 accidentally get powered on when carried in a backpack and
5801 whose boot stays stuck in a hard disk encryption passphrase
5804 * systemd-logind can be configured to also handle lid switch
5805 events even when the machine is docked or multiple displays
5806 are attached (HandleLidSwitchDocked= option).
5808 * A helper binary and a service have been added which can be
5809 used to resume from hibernation in the initramfs. A
5810 generator will parse the resume= option on the kernel
5811 command line to trigger resume.
5813 * A user console daemon systemd-consoled has been
5814 added. Currently, it is a preview, and will so far open a
5815 single terminal on each session of the user marked as
5816 Desktop=systemd-console.
5818 * Route metrics can be specified for DHCP routes added by
5821 * The SELinux context of socket-activated services can be set
5822 from the information provided by the networking stack
5823 (SELinuxContextFromNet= option).
5825 * Userspace firmware loading support has been removed and
5826 the minimum supported kernel version is thus bumped to 3.7.
5828 * Timeout for udev workers has been increased from 1 to 3
5829 minutes, but a warning will be printed after 1 minute to
5830 help diagnose kernel modules that take a long time to load.
5832 * Udev rules can now remove tags on devices with TAG-="foobar".
5834 * systemd's readahead implementation has been removed. In many
5835 circumstances it didn't give expected benefits even for
5836 rotational disk drives and was becoming less relevant in the
5837 age of SSDs. As none of the developers has been using
5838 rotating media anymore, and nobody stepped up to actively
5839 maintain this component of systemd it has now been removed.
5841 * Swap units can use Options= to specify discard options.
5842 Discard options specified for swaps in /etc/fstab are now
5845 * Docker containers are now detected as a separate type of
5848 * The Password Agent protocol gained support for queries where
5849 the user input is shown, useful e.g. for user names.
5850 systemd-ask-password gained a new --echo option to turn that
5853 * The default sysctl.d/ snippets will now set:
5855 net.core.default_qdisc = fq_codel
5857 This selects Fair Queuing Controlled Delay as the default
5858 queuing discipline for network interfaces. fq_codel helps
5859 fight the network bufferbloat problem. It is believed to be
5860 a good default with no tuning required for most workloads.
5861 Downstream distributions may override this choice. On 10Gbit
5862 servers that do not do forwarding, "fq" may perform better.
5863 Systems without a good clocksource should use "pfifo_fast".
5865 * If kdbus is enabled during build a new option BusPolicy= is
5866 available for service units, that allows locking all service
5867 processes into a stricter bus policy, in order to limit
5868 access to various bus services, or even hide most of them
5869 from the service's view entirely.
5871 * networkctl will now show the .network and .link file
5872 networkd has applied to a specific interface.
5874 * sd-login gained a new API call sd_session_get_desktop() to
5875 query which desktop environment has been selected for a
5878 * UNIX utmp support is now compile-time optional to support
5879 legacy-free systems.
5881 * systemctl gained two new commands "add-wants" and
5882 "add-requires" for pulling in units from specific targets
5885 * If the word "rescue" is specified on the kernel command line
5886 the system will now boot into rescue mode (aka
5887 rescue.target), which was previously available only by
5888 specifying "1" or "systemd.unit=rescue.target" on the kernel
5889 command line. This new kernel command line option nicely
5890 mirrors the already existing "emergency" kernel command line
5893 * New kernel command line options mount.usr=, mount.usrflags=,
5894 mount.usrfstype= have been added that match root=, rootflags=,
5895 rootfstype= but allow mounting a specific file system to
5898 * The $NOTIFY_SOCKET is now also passed to control processes of
5899 services, not only the main process.
5901 * This version reenables support for fsck's -l switch. This
5902 means at least version v2.25 of util-linux is required for
5903 operation, otherwise dead-locks on device nodes may
5904 occur. Again: you need to update util-linux to at least
5905 v2.25 when updating systemd to v217.
5907 * The "multi-seat-x" tool has been removed from systemd, as
5908 its functionality has been integrated into X servers 1.16,
5909 and the tool is hence redundant. It is recommended to update
5910 display managers invoking this tool to simply invoke X
5911 directly from now on, again.
5913 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
5914 message flag has been added for all of systemd's polkit
5915 authenticated method calls has been added. In particular this
5916 now allows optional interactive authorization via polkit for
5917 many of PID1's privileged operations such as unit file
5918 enabling and disabling.
5920 * "udevadm hwdb --update" learnt a new switch "--usr" for
5921 placing the rebuilt hardware database in /usr instead of
5922 /etc. When used only hardware database entries stored in
5923 /usr will be used, and any user database entries in /etc are
5924 ignored. This functionality is useful for vendors to ship a
5925 pre-built database on systems where local configuration is
5926 unnecessary or unlikely.
5928 * Calendar time specifications in .timer units now also
5929 understand the strings "semi-annually", "quarterly" and
5930 "minutely" as shortcuts (in addition to the preexisting
5931 "anually", "hourly", ...).
5933 * systemd-tmpfiles will now correctly create files in /dev
5934 at boot which are marked for creation only at boot. It is
5935 recommended to always create static device nodes with 'c!'
5936 and 'b!', so that they are created only at boot and not
5937 overwritten at runtime.
5939 * When the watchdog logic is used for a service (WatchdogSec=)
5940 and the watchdog timeout is hit the service will now be
5941 terminated with SIGABRT (instead of just SIGTERM), in order
5942 to make sure a proper coredump and backtrace is
5943 generated. This ensures that hanging services will result in
5944 similar coredump/backtrace behaviour as services that hit a
5947 Contributions from: Andreas Henriksson, Andrei Borzenkov,
5948 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
5949 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
5950 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
5951 Herrmann, David Sommerseth, David Strauss, Emil Renner
5952 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
5953 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
5954 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
5955 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
5956 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
5957 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
5958 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
5959 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
5960 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
5961 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
5962 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
5963 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
5964 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
5965 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
5966 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
5967 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
5970 — Berlin, 2014-10-28
5974 * timedated no longer reads NTP implementation unit names from
5975 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
5976 implementations should add a
5978 Conflicts=systemd-timesyncd.service
5980 to their unit files to take over and replace systemd's NTP
5981 default functionality.
5983 * systemd-sysusers gained a new line type "r" for configuring
5984 which UID/GID ranges to allocate system users/groups
5985 from. Lines of type "u" may now add an additional column
5986 that specifies the home directory for the system user to be
5987 created. Also, systemd-sysusers may now optionally read user
5988 information from STDIN instead of a file. This is useful for
5989 invoking it from RPM preinst scriptlets that need to create
5990 users before the first RPM file is installed since these
5991 files might need to be owned by them. A new
5992 %sysusers_create_inline RPM macro has been introduced to do
5993 just that. systemd-sysusers now updates the shadow files as
5994 well as the user/group databases, which should enhance
5995 compatibility with certain tools like grpck.
5997 * A number of bus APIs of PID 1 now optionally consult polkit to
5998 permit access for otherwise unprivileged clients under certain
5999 conditions. Note that this currently doesn't support
6000 interactive authentication yet, but this is expected to be
6001 added eventually, too.
6003 * /etc/machine-info now has new fields for configuring the
6004 deployment environment of the machine, as well as the
6005 location of the machine. hostnamectl has been updated with
6006 new command to update these fields.
6008 * systemd-timesyncd has been updated to automatically acquire
6009 NTP server information from systemd-networkd, which might
6010 have been discovered via DHCP.
6012 * systemd-resolved now includes a caching DNS stub resolver
6013 and a complete LLMNR name resolution implementation. A new
6014 NSS module "nss-resolve" has been added which can be used
6015 instead of glibc's own "nss-dns" to resolve hostnames via
6016 systemd-resolved. Hostnames, addresses and arbitrary RRs may
6017 be resolved via systemd-resolved D-Bus APIs. In contrast to
6018 the glibc internal resolver systemd-resolved is aware of
6019 multi-homed system, and keeps DNS server and caches separate
6020 and per-interface. Queries are sent simultaneously on all
6021 interfaces that have DNS servers configured, in order to
6022 properly handle VPNs and local LANs which might resolve
6023 separate sets of domain names. systemd-resolved may acquire
6024 DNS server information from systemd-networkd automatically,
6025 which in turn might have discovered them via DHCP. A tool
6026 "systemd-resolve-host" has been added that may be used to
6027 query the DNS logic in resolved. systemd-resolved implements
6028 IDNA and automatically uses IDNA or UTF-8 encoding depending
6029 on whether classic DNS or LLMNR is used as transport. In the
6030 next releases we intend to add a DNSSEC and mDNS/DNS-SD
6031 implementation to systemd-resolved.
6033 * A new NSS module nss-mymachines has been added, that
6034 automatically resolves the names of all local registered
6035 containers to their respective IP addresses.
6037 * A new client tool "networkctl" for systemd-networkd has been
6038 added. It currently is entirely passive and will query
6039 networking configuration from udev, rtnetlink and networkd,
6040 and present it to the user in a very friendly
6041 way. Eventually, we hope to extend it to become a full
6042 control utility for networkd.
6044 * .socket units gained a new DeferAcceptSec= setting that
6045 controls the kernels' TCP_DEFER_ACCEPT sockopt for
6046 TCP. Similarly, support for controlling TCP keep-alive
6047 settings has been added (KeepAliveTimeSec=,
6048 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
6049 turning off Nagle's algorithm on TCP has been added
6052 * logind learned a new session type "web", for use in projects
6053 like Cockpit which register web clients as PAM sessions.
6055 * timer units with at least one OnCalendar= setting will now
6056 be started only after time-sync.target has been
6057 reached. This way they will not elapse before the system
6058 clock has been corrected by a local NTP client or
6059 similar. This is particular useful on RTC-less embedded
6060 machines, that come up with an invalid system clock.
6062 * systemd-nspawn's --network-veth= switch should now result in
6063 stable MAC addresses for both the outer and the inner side
6066 * systemd-nspawn gained a new --volatile= switch for running
6067 container instances with /etc or /var unpopulated.
6069 * The kdbus client code has been updated to use the new Linux
6070 3.17 memfd subsystem instead of the old kdbus-specific one.
6072 * systemd-networkd's DHCP client and server now support
6073 FORCERENEW. There are also new configuration options to
6074 configure the vendor client identifier and broadcast mode
6077 * systemd will no longer inform the kernel about the current
6078 timezone, as this is necessarily incorrect and racy as the
6079 kernel has no understanding of DST and similar
6080 concepts. This hence means FAT timestamps will be always
6081 considered UTC, similar to what Android is already
6082 doing. Also, when the RTC is configured to the local time
6083 (rather than UTC) systemd will never synchronize back to it,
6084 as this might confuse Windows at a later boot.
6086 * systemd-analyze gained a new command "verify" for offline
6087 validation of unit files.
6089 * systemd-networkd gained support for a couple of additional
6090 settings for bonding networking setups. Also, the metric for
6091 statically configured routes may now be configured. For
6092 network interfaces where this is appropriate the peer IP
6093 address may now be configured.
6095 * systemd-networkd's DHCP client will no longer request
6096 broadcasting by default, as this tripped up some networks.
6097 For hardware where broadcast is required the feature should
6098 be switched back on using RequestBroadcast=yes.
6100 * systemd-networkd will now set up IPv4LL addresses (when
6101 enabled) even if DHCP is configured successfully.
6103 * udev will now default to respect network device names given
6104 by the kernel when the kernel indicates that these are
6105 predictable. This behavior can be tweaked by changing
6106 NamePolicy= in the relevant .link file.
6108 * A new library systemd-terminal has been added that
6109 implements full TTY stream parsing and rendering. This
6110 library is supposed to be used later on for implementing a
6111 full userspace VT subsystem, replacing the current kernel
6114 * A new tool systemd-journal-upload has been added to push
6115 journal data to a remote system running
6116 systemd-journal-remote.
6118 * journald will no longer forward all local data to another
6119 running syslog daemon. This change has been made because
6120 rsyslog (which appears to be the most commonly used syslog
6121 implementation these days) no longer makes use of this, and
6122 instead pulls the data out of the journal on its own. Since
6123 forwarding the messages to a non-existent syslog server is
6124 more expensive than we assumed we have now turned this
6125 off. If you run a syslog server that is not a recent rsyslog
6126 version, you have to turn this option on again
6127 (ForwardToSyslog= in journald.conf).
6129 * journald now optionally supports the LZ4 compressor for
6130 larger journal fields. This compressor should perform much
6131 better than XZ which was the previous default.
6133 * machinectl now shows the IP addresses of local containers,
6134 if it knows them, plus the interface name of the container.
6136 * A new tool "systemd-escape" has been added that makes it
6137 easy to escape strings to build unit names and similar.
6139 * sd_notify() messages may now include a new ERRNO= field
6140 which is parsed and collected by systemd and shown among the
6141 "systemctl status" output for a service.
6143 * A new component "systemd-firstboot" has been added that
6144 queries the most basic systemd information (timezone,
6145 hostname, root password) interactively on first
6146 boot. Alternatively it may also be used to provision these
6147 things offline on OS images installed into directories.
6149 * The default sysctl.d/ snippets will now set
6151 net.ipv4.conf.default.promote_secondaries=1
6153 This has the benefit of no flushing secondary IP addresses
6154 when primary addresses are removed.
6156 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
6157 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
6158 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
6159 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
6160 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
6161 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
6162 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
6163 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
6164 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
6165 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
6166 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
6167 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
6168 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
6169 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
6170 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
6172 — Berlin, 2014-08-19
6176 * A new tool systemd-sysusers has been added. This tool
6177 creates system users and groups in /etc/passwd and
6178 /etc/group, based on static declarative system user/group
6179 definitions in /usr/lib/sysusers.d/. This is useful to
6180 enable factory resets and volatile systems that boot up with
6181 an empty /etc directory, and thus need system users and
6182 groups created during early boot. systemd now also ships
6183 with two default sysusers.d/ files for the most basic
6184 users and groups systemd and the core operating system
6187 * A new tmpfiles snippet has been added that rebuilds the
6188 essential files in /etc on boot, should they be missing.
6190 * A directive for ensuring automatic clean-up of
6191 /var/cache/man/ has been removed from the default
6192 configuration. This line should now be shipped by the man
6193 implementation. The necessary change has been made to the
6194 man-db implementation. Note that you need to update your man
6195 implementation to one that ships this line, otherwise no
6196 automatic clean-up of /var/cache/man will take place.
6198 * A new condition ConditionNeedsUpdate= has been added that
6199 may conditionalize services to only run when /etc or /var
6200 are "older" than the vendor operating system resources in
6201 /usr. This is useful for reconstructing or updating /etc
6202 after an offline update of /usr or a factory reset, on the
6203 next reboot. Services that want to run once after such an
6204 update or reset should use this condition and order
6205 themselves before the new systemd-update-done.service, which
6206 will mark the two directories as fully updated. A number of
6207 service files have been added making use of this, to rebuild
6208 the udev hardware database, the journald message catalog and
6209 dynamic loader cache (ldconfig). The systemd-sysusers tool
6210 described above also makes use of this now. With this in
6211 place it is now possible to start up a minimal operating
6212 system with /etc empty cleanly. For more information on the
6213 concepts involved see this recent blog story:
6215 http://0pointer.de/blog/projects/stateless.html
6217 * A new system group "input" has been introduced, and all
6218 input device nodes get this group assigned. This is useful
6219 for system-level software to get access to input devices. It
6220 complements what is already done for "audio" and "video".
6222 * systemd-networkd learnt minimal DHCPv4 server support in
6223 addition to the existing DHCPv4 client support. It also
6224 learnt DHCPv6 client and IPv6 Router Solicitation client
6225 support. The DHCPv4 client gained support for static routes
6226 passed in from the server. Note that the [DHCPv4] section
6227 known in older systemd-networkd versions has been renamed to
6228 [DHCP] and is now also used by the DHCPv6 client. Existing
6229 .network files using settings of this section should be
6230 updated, though compatibility is maintained. Optionally, the
6231 client hostname may now be sent to the DHCP server.
6233 * networkd gained support for vxlan virtual networks as well
6234 as tun/tap and dummy devices.
6236 * networkd gained support for automatic allocation of address
6237 ranges for interfaces from a system-wide pool of
6238 addresses. This is useful for dynamically managing a large
6239 number of interfaces with a single network configuration
6240 file. In particular this is useful to easily assign
6241 appropriate IP addresses to the veth links of a large number
6242 of nspawn instances.
6244 * RPM macros for processing sysusers, sysctl and binfmt
6245 drop-in snippets at package installation time have been
6248 * The /etc/os-release file should now be placed in
6249 /usr/lib/os-release. The old location is automatically
6250 created as symlink. /usr/lib is the more appropriate
6251 location of this file, since it shall actually describe the
6252 vendor operating system shipped in /usr, and not the
6253 configuration stored in /etc.
6255 * .mount units gained a new boolean SloppyOptions= setting
6256 that maps to mount(8)'s -s option which enables permissive
6257 parsing of unknown mount options.
6259 * tmpfiles learnt a new "L+" directive which creates a symlink
6260 but (unlike "L") deletes a pre-existing file first, should
6261 it already exist and not already be the correct
6262 symlink. Similarly, "b+", "c+" and "p+" directives have been
6263 added as well, which create block and character devices, as
6264 well as fifos in the filesystem, possibly removing any
6265 pre-existing files of different types.
6267 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
6268 'argument' field (which so far specified the source to
6269 symlink/copy the files from) is now optional. If omitted the
6270 same file os copied from /usr/share/factory/ suffixed by the
6271 full destination path. This is useful for populating /etc
6272 with essential files, by copying them from vendor defaults
6273 shipped in /usr/share/factory/etc.
6275 * A new command "systemctl preset-all" has been added that
6276 applies the service preset settings to all installed unit
6277 files. A new switch --preset-mode= has been added that
6278 controls whether only enable or only disable operations
6281 * A new command "systemctl is-system-running" has been added
6282 that allows checking the overall state of the system, for
6283 example whether it is fully up and running.
6285 * When the system boots up with an empty /etc, the equivalent
6286 to "systemctl preset-all" is executed during early boot, to
6287 make sure all default services are enabled after a factory
6290 * systemd now contains a minimal preset file that enables the
6291 most basic services systemd ships by default.
6293 * Unit files' [Install] section gained a new DefaultInstance=
6294 field for defining the default instance to create if a
6295 template unit is enabled with no instance specified.
6297 * A new passive target cryptsetup-pre.target has been added
6298 that may be used by services that need to make they run and
6299 finish before the first LUKS cryptographic device is set up.
6301 * The /dev/loop-control and /dev/btrfs-control device nodes
6302 are now owned by the "disk" group by default, opening up
6303 access to this group.
6305 * systemd-coredump will now automatically generate a
6306 stack trace of all core dumps taking place on the system,
6307 based on elfutils' libdw library. This stack trace is logged
6310 * systemd-coredump may now optionally store coredumps directly
6311 on disk (in /var/lib/systemd/coredump, possibly compressed),
6312 instead of storing them unconditionally in the journal. This
6313 mode is the new default. A new configuration file
6314 /etc/systemd/coredump.conf has been added to configure this
6315 and other parameters of systemd-coredump.
6317 * coredumpctl gained a new "info" verb to show details about a
6318 specific coredump. A new switch "-1" has also been added
6319 that makes sure to only show information about the most
6320 recent entry instead of all entries. Also, as the tool is
6321 generally useful now the "systemd-" prefix of the binary
6322 name has been removed. Distributions that want to maintain
6323 compatibility with the old name should add a symlink from
6324 the old name to the new name.
6326 * journald's SplitMode= now defaults to "uid". This makes sure
6327 that unprivileged users can access their own coredumps with
6328 coredumpctl without restrictions.
6330 * New kernel command line options "systemd.wants=" (for
6331 pulling an additional unit during boot), "systemd.mask="
6332 (for masking a specific unit for the boot), and
6333 "systemd.debug-shell" (for enabling the debug shell on tty9)
6334 have been added. This is implemented in the new generator
6335 "systemd-debug-generator".
6337 * systemd-nspawn will now by default filter a couple of
6338 syscalls for containers, among them those required for
6339 kernel module loading, direct x86 IO port access, swap
6340 management, and kexec. Most importantly though
6341 open_by_handle_at() is now prohibited for containers,
6342 closing a hole similar to a recently discussed vulnerability
6343 in docker regarding access to files on file hierarchies the
6344 container should normally not have access to. Note that, for
6345 nspawn, we generally make no security claims anyway (and
6346 this is explicitly documented in the man page), so this is
6347 just a fix for one of the most obvious problems.
6349 * A new man page file-hierarchy(7) has been added that
6350 contains a minimized, modernized version of the file system
6351 layout systemd expects, similar in style to the FHS
6352 specification or hier(5). A new tool systemd-path(1) has
6353 been added to query many of these paths for the local
6356 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
6357 longer done. Since the directory now has a per-user size
6358 limit, and is cleaned on logout this appears unnecessary,
6359 in particular since this now brings the lifecycle of this
6360 directory closer in line with how IPC objects are handled.
6362 * systemd.pc now exports a number of additional directories,
6363 including $libdir (which is useful to identify the library
6364 path for the primary architecture of the system), and a
6365 couple of drop-in directories.
6367 * udev's predictable network interface names now use the dev_port
6368 sysfs attribute, introduced in linux 3.15 instead of dev_id to
6369 distinguish between ports of the same PCI function. dev_id should
6370 only be used for ports using the same HW address, hence the need
6373 * machined has been updated to export the OS version of a
6374 container (read from /etc/os-release and
6375 /usr/lib/os-release) on the bus. This is now shown in
6376 "machinectl status" for a machine.
6378 * A new service setting RestartForceExitStatus= has been
6379 added. If configured to a set of exit signals or process
6380 return values, the service will be restarted when the main
6381 daemon process exits with any of them, regardless of the
6384 * systemctl's -H switch for connecting to remote systemd
6385 machines has been extended so that it may be used to
6386 directly connect to a specific container on the
6387 host. "systemctl -H root@foobar:waldi" will now connect as
6388 user "root" to host "foobar", and then proceed directly to
6389 the container named "waldi". Note that currently you have to
6390 authenticate as user "root" for this to work, as entering
6391 containers is a privileged operation.
6393 Contributions from: Andreas Henriksson, Benjamin Steinwender,
6394 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
6395 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
6396 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
6397 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
6398 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
6399 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
6400 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
6401 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
6402 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
6403 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
6404 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
6406 — Berlin, 2014-07-03
6410 * As an experimental feature, udev now tries to lock the
6411 disk device node (flock(LOCK_SH|LOCK_NB)) while it
6412 executes events for the disk or any of its partitions.
6413 Applications like partitioning programs can lock the
6414 disk device node (flock(LOCK_EX)) and claim temporary
6415 device ownership that way; udev will entirely skip all event
6416 handling for this disk and its partitions. If the disk
6417 was opened for writing, the close will trigger a partition
6418 table rescan in udev's "watch" facility, and if needed
6419 synthesize "change" events for the disk and all its partitions.
6420 This is now unconditionally enabled, and if it turns out to
6421 cause major problems, we might turn it on only for specific
6422 devices, or might need to disable it entirely. Device Mapper
6423 devices are excluded from this logic.
6425 * We temporarily dropped the "-l" switch for fsck invocations,
6426 since they collide with the flock() logic above. util-linux
6427 upstream has been changed already to avoid this conflict,
6428 and we will re-add "-l" as soon as util-linux with this
6429 change has been released.
6431 * The dependency on libattr has been removed. Since a long
6432 time, the extended attribute calls have moved to glibc, and
6433 libattr is thus unnecessary.
6435 * Virtualization detection works without privileges now. This
6436 means the systemd-detect-virt binary no longer requires
6437 CAP_SYS_PTRACE file capabilities, and our daemons can run
6438 with fewer privileges.
6440 * systemd-networkd now runs under its own "systemd-network"
6441 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
6442 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
6443 loses the ability to write to files owned by root this way.
6445 * Similarly, systemd-resolved now runs under its own
6446 "systemd-resolve" user with no capabilities remaining.
6448 * Similarly, systemd-bus-proxyd now runs under its own
6449 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
6451 * systemd-networkd gained support for setting up "veth"
6452 virtual Ethernet devices for container connectivity, as well
6453 as GRE and VTI tunnels.
6455 * systemd-networkd will no longer automatically attempt to
6456 manually load kernel modules necessary for certain tunnel
6457 transports. Instead, it is assumed the kernel loads them
6458 automatically when required. This only works correctly on
6459 very new kernels. On older kernels, please consider adding
6460 the kernel modules to /etc/modules-load.d/ as a work-around.
6462 * The resolv.conf file systemd-resolved generates has been
6463 moved to /run/systemd/resolve/. If you have a symlink from
6464 /etc/resolv.conf, it might be necessary to correct it.
6466 * Two new service settings, ProtectHome= and ProtectSystem=,
6467 have been added. When enabled, they will make the user data
6468 (such as /home) inaccessible or read-only and the system
6469 (such as /usr) read-only, for specific services. This allows
6470 very light-weight per-service sandboxing to avoid
6471 modifications of user data or system files from
6472 services. These two new switches have been enabled for all
6473 of systemd's long-running services, where appropriate.
6475 * Socket units gained new SocketUser= and SocketGroup=
6476 settings to set the owner user and group of AF_UNIX sockets
6477 and FIFOs in the file system.
6479 * Socket units gained a new RemoveOnStop= setting. If enabled,
6480 all FIFOS and sockets in the file system will be removed
6481 when the specific socket unit is stopped.
6483 * Socket units gained a new Symlinks= setting. It takes a list
6484 of symlinks to create to file system sockets or FIFOs
6485 created by the specific Unix sockets. This is useful to
6486 manage symlinks to socket nodes with the same lifecycle as
6489 * The /dev/log socket and /dev/initctl FIFO have been moved to
6490 /run, and have been replaced by symlinks. This allows
6491 connecting to these facilities even if PrivateDevices=yes is
6492 used for a service (which makes /dev/log itself unavailable,
6493 but /run is left). This also has the benefit of ensuring
6494 that /dev only contains device nodes, directories and
6495 symlinks, and nothing else.
6497 * sd-daemon gained two new calls sd_pid_notify() and
6498 sd_pid_notifyf(). They are similar to sd_notify() and
6499 sd_notifyf(), but allow overriding of the source PID of
6500 notification messages if permissions permit this. This is
6501 useful to send notify messages on behalf of a different
6502 process (for example, the parent process). The
6503 systemd-notify tool has been updated to make use of this
6504 when sending messages (so that notification messages now
6505 originate from the shell script invoking systemd-notify and
6506 not the systemd-notify process itself. This should minimize
6507 a race where systemd fails to associate notification
6508 messages to services when the originating process already
6511 * A new "on-abnormal" setting for Restart= has been added. If
6512 set, it will result in automatic restarts on all "abnormal"
6513 reasons for a process to exit, which includes unclean
6514 signals, core dumps, timeouts and watchdog timeouts, but
6515 does not include clean and unclean exit codes or clean
6516 signals. Restart=on-abnormal is an alternative for
6517 Restart=on-failure for services that shall be able to
6518 terminate and avoid restarts on certain errors, by
6519 indicating so with an unclean exit code. Restart=on-failure
6520 or Restart=on-abnormal is now the recommended setting for
6521 all long-running services.
6523 * If the InaccessibleDirectories= service setting points to a
6524 mount point (or if there are any submounts contained within
6525 it), it is now attempted to completely unmount it, to make
6526 the file systems truly unavailable for the respective
6529 * The ReadOnlyDirectories= service setting and
6530 systemd-nspawn's --read-only parameter are now recursively
6531 applied to all submounts, too.
6533 * Mount units may now be created transiently via the bus APIs.
6535 * The support for SysV and LSB init scripts has been removed
6536 from the systemd daemon itself. Instead, it is now
6537 implemented as a generator that creates native systemd units
6538 from these scripts when needed. This enables us to remove a
6539 substantial amount of legacy code from PID 1, following the
6540 fact that many distributions only ship a very small number
6541 of LSB/SysV init scripts nowadays.
6543 * Privileged Xen (dom0) domains are not considered
6544 virtualization anymore by the virtualization detection
6545 logic. After all, they generally have unrestricted access to
6546 the hardware and usually are used to manage the unprivileged
6549 * systemd-tmpfiles gained a new "C" line type, for copying
6550 files or entire directories.
6552 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
6553 lines. So far, they have been non-globbing versions of the
6554 latter, and have thus been redundant. In future, it is
6555 recommended to only use "z". "m" has hence been removed
6556 from the documentation, even though it stays supported.
6558 * A tmpfiles snippet to recreate the most basic structure in
6559 /var has been added. This is enough to create the /var/run →
6560 /run symlink and create a couple of structural
6561 directories. This allows systems to boot up with an empty or
6562 volatile /var. Of course, while with this change, the core OS
6563 now is capable with dealing with a volatile /var, not all
6564 user services are ready for it. However, we hope that sooner
6565 or later, many service daemons will be changed upstream so
6566 that they are able to automatically create their necessary
6567 directories in /var at boot, should they be missing. This is
6568 the first step to allow state-less systems that only require
6569 the vendor image for /usr to boot.
6571 * systemd-nspawn has gained a new --tmpfs= switch to mount an
6572 empty tmpfs instance to a specific directory. This is
6573 particularly useful for making use of the automatic
6574 reconstruction of /var (see above), by passing --tmpfs=/var.
6576 * Access modes specified in tmpfiles snippets may now be
6577 prefixed with "~", which indicates that they shall be masked
6578 by whether the existing file or directory is currently
6579 writable, readable or executable at all. Also, if specified,
6580 the sgid/suid/sticky bits will be masked for all
6583 * A new passive target unit "network-pre.target" has been
6584 added which is useful for services that shall run before any
6585 network is configured, for example firewall scripts.
6587 * The "floppy" group that previously owned the /dev/fd*
6588 devices is no longer used. The "disk" group is now used
6589 instead. Distributions should probably deprecate usage of
6592 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
6593 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
6594 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
6595 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
6596 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
6597 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
6600 — Berlin, 2014-06-11
6604 * A new "systemd-timesyncd" daemon has been added for
6605 synchronizing the system clock across the network. It
6606 implements an SNTP client. In contrast to NTP
6607 implementations such as chrony or the NTP reference server,
6608 this only implements a client side, and does not bother with
6609 the full NTP complexity, focusing only on querying time from
6610 one remote server and synchronizing the local clock to
6611 it. Unless you intend to serve NTP to networked clients or
6612 want to connect to local hardware clocks, this simple NTP
6613 client should be more than appropriate for most
6614 installations. The daemon runs with minimal privileges, and
6615 has been hooked up with networkd to only operate when
6616 network connectivity is available. The daemon saves the
6617 current clock to disk every time a new NTP sync has been
6618 acquired, and uses this to possibly correct the system clock
6619 early at bootup, in order to accommodate for systems that
6620 lack an RTC such as the Raspberry Pi and embedded devices,
6621 and to make sure that time monotonically progresses on these
6622 systems, even if it is not always correct. To make use of
6623 this daemon, a new system user and group "systemd-timesync"
6624 needs to be created on installation of systemd.
6626 * The queue "seqnum" interface of libudev has been disabled, as
6627 it was generally incompatible with device namespacing as
6628 sequence numbers of devices go "missing" if the devices are
6629 part of a different namespace.
6631 * "systemctl list-timers" and "systemctl list-sockets" gained
6632 a --recursive switch for showing units of these types also
6633 for all local containers, similar in style to the already
6634 supported --recursive switch for "systemctl list-units".
6636 * A new RebootArgument= setting has been added for service
6637 units, which may be used to specify a kernel reboot argument
6638 to use when triggering reboots with StartLimitAction=.
6640 * A new FailureAction= setting has been added for service
6641 units which may be used to specify an operation to trigger
6642 when a service fails. This works similarly to
6643 StartLimitAction=, but unlike it, controls what is done
6644 immediately rather than only after several attempts to
6645 restart the service in question.
6647 * hostnamed got updated to also expose the kernel name,
6648 release, and version on the bus. This is useful for
6649 executing commands like hostnamectl with the -H switch.
6650 systemd-analyze makes use of this to properly display
6651 details when running non-locally.
6653 * The bootchart tool can now show cgroup information in the
6654 graphs it generates.
6656 * The CFS CPU quota cgroup attribute is now exposed for
6657 services. The new CPUQuota= switch has been added for this
6658 which takes a percentage value. Setting this will have the
6659 result that a service may never get more CPU time than the
6660 specified percentage, even if the machine is otherwise idle.
6662 * systemd-networkd learned IPIP and SIT tunnel support.
6664 * LSB init scripts exposing a dependency on $network will now
6665 get a dependency on network-online.target rather than simply
6666 network.target. This should bring LSB handling closer to
6667 what it was on SysV systems.
6669 * A new fsck.repair= kernel option has been added to control
6670 how fsck shall deal with unclean file systems at boot.
6672 * The (.ini) configuration file parser will now silently
6673 ignore sections whose name begins with "X-". This may be
6674 used to maintain application-specific extension sections in unit
6677 * machined gained a new API to query the IP addresses of
6678 registered containers. "machinectl status" has been updated
6679 to show these addresses in its output.
6681 * A new call sd_uid_get_display() has been added to the
6682 sd-login APIs for querying the "primary" session of a
6683 user. The "primary" session of the user is elected from the
6684 user's sessions and generally a graphical session is
6685 preferred over a text one.
6687 * A minimal systemd-resolved daemon has been added. It
6688 currently simply acts as a companion to systemd-networkd and
6689 manages resolv.conf based on per-interface DNS
6690 configuration, possibly supplied via DHCP. In the long run
6691 we hope to extend this into a local DNSSEC enabled DNS and
6694 * The systemd-networkd-wait-online tool is now enabled by
6695 default. It will delay network-online.target until a network
6696 connection has been configured. The tool primarily integrates
6697 with networkd, but will also make a best effort to make sense
6698 of network configuration performed in some other way.
6700 * Two new service options StartupCPUShares= and
6701 StartupBlockIOWeight= have been added that work similarly to
6702 CPUShares= and BlockIOWeight= however only apply during
6703 system startup. This is useful to prioritize certain services
6704 differently during bootup than during normal runtime.
6706 * hostnamed has been changed to prefer the statically
6707 configured hostname in /etc/hostname (unless set to
6708 'localhost' or empty) over any dynamic one supplied by
6709 dhcp. With this change, the rules for picking the hostname
6710 match more closely the rules of other configuration settings
6711 where the local administrator's configuration in /etc always
6712 overrides any other settings.
6714 Contributions from: Ali H. Caliskan, Alison Chaiken, Bas van
6715 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
6716 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
6717 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
6718 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
6719 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
6720 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
6721 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
6722 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
6723 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
6724 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
6725 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
6726 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
6727 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
6728 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
6729 Lindskog, WaLyong Cho, Will Woods, Zbigniew
6732 — Beijing, 2014-05-28
6736 * When restoring the screen brightness at boot, stay away from
6737 the darkest setting or from the lowest 5% of the available
6738 range, depending on which is the larger value of both. This
6739 should effectively protect the user from rebooting into a
6740 black screen, should the brightness have been set to minimum
6743 * sd-login gained a new sd_machine_get_class() call to
6744 determine the class ("vm" or "container") of a machine
6745 registered with machined.
6747 * sd-login gained new calls
6748 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
6749 to query the identity of the peer of a local AF_UNIX
6750 connection. They operate similarly to their sd_pid_get_xyz()
6753 * PID 1 will now maintain a system-wide system state engine
6754 with the states "starting", "running", "degraded",
6755 "maintenance", "stopping". These states are bound to system
6756 startup, normal runtime, runtime with at least one failed
6757 service, rescue/emergency mode and system shutdown. This
6758 state is shown in the "systemctl status" output when no unit
6759 name is passed. It is useful to determine system state, in
6760 particularly when doing so for many systems or containers at
6763 * A new command "list-machines" has been added to "systemctl"
6764 that lists all local OS containers and shows their system
6765 state (see above), if systemd runs inside of them.
6767 * systemctl gained a new "-r" switch to recursively enumerate
6768 units on all local containers, when used with the
6769 "list-unit" command (which is the default one that is
6770 executed when no parameters are specified).
6772 * The GPT automatic partition discovery logic will now honour
6773 two GPT partition flags: one may be set on a partition to
6774 cause it to be mounted read-only, and the other may be set
6775 on a partition to ignore it during automatic discovery.
6777 * Two new GPT type UUIDs have been added for automatic root
6778 partition discovery, for 32-bit and 64-bit ARM. This is not
6779 particularly useful for discovering the root directory on
6780 these architectures during bare-metal boots (since UEFI is
6781 not common there), but still very useful to allow booting of
6782 ARM disk images in nspawn with the -i option.
6784 * MAC addresses of interfaces created with nspawn's
6785 --network-interface= switch will now be generated from the
6786 machine name, and thus be stable between multiple invocations
6789 * logind will now automatically remove all IPC objects owned
6790 by a user if she or he fully logs out. This makes sure that
6791 users who are logged out cannot continue to consume IPC
6792 resources. This covers SysV memory, semaphores and message
6793 queues as well as POSIX shared memory and message
6794 queues. Traditionally, SysV and POSIX IPC had no lifecycle
6795 limits. With this functionality, that is corrected. This may
6796 be turned off by using the RemoveIPC= switch of logind.conf.
6798 * The systemd-machine-id-setup and tmpfiles tools gained a
6799 --root= switch to operate on a specific root directory,
6802 * journald can now forward logged messages to the TTYs of all
6803 logged in users ("wall"). This is the default for all
6804 emergency messages now.
6806 * A new tool systemd-journal-remote has been added to stream
6807 journal log messages across the network.
6809 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
6810 controller trees are mounted into it. Note that the
6811 directories mounted beneath it are not read-only. This is a
6812 security measure and is particularly useful because glibc
6813 actually includes a search logic to pick any tmpfs it can
6814 find to implement shm_open() if /dev/shm is not available
6815 (which it might very well be in namespaced setups).
6817 * machinectl gained a new "poweroff" command to cleanly power
6818 down a local OS container.
6820 * The PrivateDevices= unit file setting will now also drop the
6821 CAP_MKNOD capability from the capability bound set, and
6822 imply DevicePolicy=closed.
6824 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
6825 comprehensively on all long-running systemd services where
6826 this is appropriate.
6828 * systemd-udevd will now run in a disassociated mount
6829 namespace. To mount directories from udev rules, make sure to
6830 pull in mount units via SYSTEMD_WANTS properties.
6832 * The kdbus support gained support for uploading policy into
6833 the kernel. sd-bus gained support for creating "monitoring"
6834 connections that can eavesdrop into all bus communication
6835 for debugging purposes.
6837 * Timestamps may now be specified in seconds since the UNIX
6838 epoch Jan 1st, 1970 by specifying "@" followed by the value
6841 * Native tcpwrap support in systemd has been removed. tcpwrap
6842 is old code, not really maintained anymore and has serious
6843 shortcomings, and better options such as firewalls
6844 exist. For setups that require tcpwrap usage, please
6845 consider invoking your socket-activated service via tcpd,
6846 like on traditional inetd.
6848 * A new system.conf configuration option
6849 DefaultTimerAccuracySec= has been added that controls the
6850 default AccuracySec= setting of .timer units.
6852 * Timer units gained a new WakeSystem= switch. If enabled,
6853 timers configured this way will cause the system to resume
6854 from system suspend (if the system supports that, which most
6857 * Timer units gained a new Persistent= switch. If enabled,
6858 timers configured this way will save to disk when they have
6859 been last triggered. This information is then used on next
6860 reboot to possible execute overdue timer events, that
6861 could not take place because the system was powered off.
6862 This enables simple anacron-like behaviour for timer units.
6864 * systemctl's "list-timers" will now also list the time a
6865 timer unit was last triggered in addition to the next time
6866 it will be triggered.
6868 * systemd-networkd will now assign predictable IPv4LL
6869 addresses to its local interfaces.
6871 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
6872 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
6873 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
6874 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
6875 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
6876 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
6877 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
6878 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
6881 — Berlin, 2014-03-25
6885 * A new unit file setting RestrictAddressFamilies= has been
6886 added to restrict which socket address families unit
6887 processes gain access to. This takes address family names
6888 like "AF_INET" or "AF_UNIX", and is useful to minimize the
6889 attack surface of services via exotic protocol stacks. This
6890 is built on seccomp system call filters.
6892 * Two new unit file settings RuntimeDirectory= and
6893 RuntimeDirectoryMode= have been added that may be used to
6894 manage a per-daemon runtime directories below /run. This is
6895 an alternative for setting up directory permissions with
6896 tmpfiles snippets, and has the advantage that the runtime
6897 directory's lifetime is bound to the daemon runtime and that
6898 the daemon starts up with an empty directory each time. This
6899 is particularly useful when writing services that drop
6900 privileges using the User= or Group= setting.
6902 * The DeviceAllow= unit setting now supports globbing for
6903 matching against device group names.
6905 * The systemd configuration file system.conf gained new
6906 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
6907 DefaultMemoryAccounting= to globally turn on/off accounting
6908 for specific resources (cgroups) for all units. These
6909 settings may still be overridden individually in each unit
6912 * systemd-gpt-auto-generator is now able to discover /srv and
6913 root partitions in addition to /home and swap partitions. It
6914 also supports LUKS-encrypted partitions now. With this in
6915 place, automatic discovery of partitions to mount following
6916 the Discoverable Partitions Specification
6917 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
6918 is now a lot more complete. This allows booting without
6919 /etc/fstab and without root= on the kernel command line on
6920 systems prepared appropriately.
6922 * systemd-nspawn gained a new --image= switch which allows
6923 booting up disk images and Linux installations on any block
6924 device that follow the Discoverable Partitions Specification
6925 (see above). This means that installations made with
6926 appropriately updated installers may now be started and
6927 deployed using container managers, completely
6928 unmodified. (We hope that libvirt-lxc will add support for
6929 this feature soon, too.)
6931 * systemd-nspawn gained a new --network-macvlan= setting to
6932 set up a private macvlan interface for the
6933 container. Similarly, systemd-networkd gained a new
6934 Kind=macvlan setting in .netdev files.
6936 * systemd-networkd now supports configuring local addresses
6939 * A new tool systemd-network-wait-online has been added to
6940 synchronously wait for network connectivity using
6943 * The sd-bus.h bus API gained a new sd_bus_track object for
6944 tracking the lifecycle of bus peers. Note that sd-bus.h is
6945 still not a public API though (unless you specify
6946 --enable-kdbus on the configure command line, which however
6947 voids your warranty and you get no API stability guarantee).
6949 * The $XDG_RUNTIME_DIR runtime directories for each user are
6950 now individual tmpfs instances, which has the benefit of
6951 introducing separate pools for each user, with individual
6952 size limits, and thus making sure that unprivileged clients
6953 can no longer negatively impact the system or other users by
6954 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
6955 RuntimeDirectorySize= has been introduced that allows
6956 controlling the default size limit for all users. It
6957 defaults to 10% of the available physical memory. This is no
6958 replacement for quotas on tmpfs though (which the kernel
6959 still does not support), as /dev/shm and /tmp are still
6960 shared resources used by both the system and unprivileged
6963 * logind will now automatically turn off automatic suspending
6964 on laptop lid close when more than one display is
6965 connected. This was previously expected to be implemented
6966 individually in desktop environments (such as GNOME),
6967 however has been added to logind now, in order to fix a
6968 boot-time race where a desktop environment might not have
6969 been started yet and thus not been able to take an inhibitor
6970 lock at the time where logind already suspends the system
6971 due to a closed lid.
6973 * logind will now wait at least 30s after each system
6974 suspend/resume cycle, and 3min after system boot before
6975 suspending the system due to a closed laptop lid. This
6976 should give USB docking stations and similar enough time to
6977 be probed and configured after system resume and boot in
6978 order to then act as suspend blocker.
6980 * systemd-run gained a new --property= setting which allows
6981 initialization of resource control properties (and others)
6982 for the created scope or service unit. Example: "systemd-run
6983 --property=BlockIOWeight=10 updatedb" may be used to run
6984 updatedb at a low block IO scheduling weight.
6986 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
6987 now also work in --scope mode.
6989 * When systemd is compiled with kdbus support, basic support
6990 for enforced policies is now in place. (Note that enabling
6991 kdbus still voids your warranty and no API compatibility
6994 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
6995 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
6996 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
6997 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
6998 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
6999 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
7000 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
7001 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
7002 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
7003 Zbigniew Jędrzejewski-Szmek
7005 — Berlin, 2014-03-12
7009 * systemd will now relabel /dev after loading the SMACK policy
7010 according to SMACK rules.
7012 * A new unit file option AppArmorProfile= has been added to
7013 set the AppArmor profile for the processes of a unit.
7015 * A new condition check ConditionArchitecture= has been added
7016 to conditionalize units based on the system architecture, as
7017 reported by uname()'s "machine" field.
7019 * systemd-networkd now supports matching on the system
7020 virtualization, architecture, kernel command line, host name
7023 * logind is now a lot more aggressive when suspending the
7024 machine due to a closed laptop lid. Instead of acting only
7025 on the lid close action, it will continuously watch the lid
7026 status and act on it. This is useful for laptops where the
7027 power button is on the outside of the chassis so that it can
7028 be reached without opening the lid (such as the Lenovo
7029 Yoga). On those machines, logind will now immediately
7030 re-suspend the machine if the power button has been
7031 accidentally pressed while the laptop was suspended and in a
7032 backpack or similar.
7034 * logind will now watch SW_DOCK switches and inhibit reaction
7035 to the lid switch if it is pressed. This means that logind
7036 will not suspend the machine anymore if the lid is closed
7037 and the system is docked, if the laptop supports SW_DOCK
7038 notifications via the input layer. Note that ACPI docking
7039 stations do not generate this currently. Also note that this
7040 logic is usually not fully sufficient and Desktop
7041 Environments should take a lid switch inhibitor lock when an
7042 external display is connected, as systemd will not watch
7045 * nspawn will now make use of the devices cgroup controller by
7046 default, and only permit creation of and access to the usual
7047 API device nodes like /dev/null or /dev/random, as well as
7048 access to (but not creation of) the pty devices.
7050 * We will now ship a default .network file for
7051 systemd-networkd that automatically configures DHCP for
7052 network interfaces created by nspawn's --network-veth or
7053 --network-bridge= switches.
7055 * systemd will now understand the usual M, K, G, T suffixes
7056 according to SI conventions (i.e. to the base 1000) when
7057 referring to throughput and hardware metrics. It will stay
7058 with IEC conventions (i.e. to the base 1024) for software
7059 metrics, according to what is customary according to
7060 Wikipedia. We explicitly document which base applies for
7061 each configuration option.
7063 * The DeviceAllow= setting in unit files now supports a syntax
7064 to whitelist an entire group of devices node majors at once,
7065 based on the /proc/devices listing. For example, with the
7066 string "char-pts", it is now possible to whitelist all
7067 current and future pseudo-TTYs at once.
7069 * sd-event learned a new "post" event source. Event sources of
7070 this type are triggered by the dispatching of any event
7071 source of a type that is not "post". This is useful for
7072 implementing clean-up and check event sources that are
7073 triggered by other work being done in the program.
7075 * systemd-networkd is no longer statically enabled, but uses
7076 the usual [Install] sections so that it can be
7077 enabled/disabled using systemctl. It still is enabled by
7080 * When creating a veth interface pair with systemd-nspawn, the
7081 host side will now be prefixed with "vb-" if
7082 --network-bridge= is used, and with "ve-" if --network-veth
7083 is used. This way, it is easy to distinguish these cases on
7084 the host, for example to apply different configuration to
7085 them with systemd-networkd.
7087 * The compatibility libraries for libsystemd-journal.so,
7088 libsystem-id128.so, libsystemd-login.so and
7089 libsystemd-daemon.so do not make use of IFUNC
7090 anymore. Instead, we now build libsystemd.so multiple times
7091 under these alternative names. This means that the footprint
7092 is drastically increased, but given that these are
7093 transitional compatibility libraries, this should not matter
7094 much. This change has been made necessary to support the ARM
7095 platform for these compatibility libraries, as the ARM
7096 toolchain is not really at the same level as the toolchain
7097 for other architectures like x86 and does not support
7098 IFUNC. Please make sure to use --enable-compat-libs only
7099 during a transitional period!
7101 * The .include syntax has been deprecated and is not documented
7102 anymore. Drop-in files in .d directories should be used instead.
7104 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
7105 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
7106 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
7107 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
7108 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
7109 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
7110 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
7111 Zbigniew Jędrzejewski-Szmek
7113 — Berlin, 2014-02-24
7117 * A new component "systemd-networkd" has been added that can
7118 be used to configure local network interfaces statically or
7119 via DHCP. It is capable of bringing up bridges, VLANs, and
7120 bonding. Currently, no hook-ups for interactive network
7121 configuration are provided. Use this for your initrd,
7122 container, embedded, or server setup if you need a simple,
7123 yet powerful, network configuration solution. This
7124 configuration subsystem is quite nifty, as it allows wildcard
7125 hotplug matching in interfaces. For example, with a single
7126 configuration snippet, you can configure that all Ethernet
7127 interfaces showing up are automatically added to a bridge,
7128 or similar. It supports link-sensing and more.
7130 * A new tool "systemd-socket-proxyd" has been added which can
7131 act as a bidirectional proxy for TCP sockets. This is
7132 useful for adding socket activation support to services that
7133 do not actually support socket activation, including virtual
7134 machines and the like.
7136 * Add a new tool to save/restore rfkill state on
7139 * Save/restore state of keyboard backlights in addition to
7140 display backlights on shutdown/boot.
7142 * udev learned a new SECLABEL{} construct to label device
7143 nodes with a specific security label when they appear. For
7144 now, only SECLABEL{selinux} is supported, but the syntax is
7145 prepared for additional security frameworks.
7147 * udev gained a new scheme to configure link-level attributes
7148 from files in /etc/systemd/network/*.link. These files can
7149 match against MAC address, device path, driver name and type,
7150 and will apply attributes like the naming policy, link speed,
7151 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
7152 address assignment policy (randomized, ...).
7154 * The configuration of network interface naming rules for
7155 "permanent interface names" has changed: a new NamePolicy=
7156 setting in the [Link] section of .link files determines the
7157 priority of possible naming schemes (onboard, slot, MAC,
7158 path). The default value of this setting is determined by
7159 /usr/lib/net/links/99-default.link. Old
7160 80-net-name-slot.rules udev configuration file has been
7161 removed, so local configuration overriding this file should
7162 be adapted to override 99-default.link instead.
7164 * When the User= switch is used in a unit file, also
7165 initialize $SHELL= based on the user database entry.
7167 * systemd no longer depends on libdbus. All communication is
7168 now done with sd-bus, systemd's low-level bus library
7171 * kdbus support has been added to PID 1 itself. When kdbus is
7172 enabled, this causes PID 1 to set up the system bus and
7173 enable support for a new ".busname" unit type that
7174 encapsulates bus name activation on kdbus. It works a little
7175 bit like ".socket" units, except for bus names. A new
7176 generator has been added that converts classic dbus1 service
7177 activation files automatically into native systemd .busname
7180 * sd-bus: add a light-weight vtable implementation that allows
7181 defining objects on the bus with a simple static const
7182 vtable array of its methods, signals and properties.
7184 * systemd will not generate or install static dbus
7185 introspection data anymore to /usr/share/dbus-1/interfaces,
7186 as the precise format of these files is unclear, and
7187 nothing makes use of it.
7189 * A proxy daemon is now provided to proxy clients connecting
7190 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
7191 compatibility with classic D-Bus.
7193 * A bus driver implementation has been added that supports the
7194 classic D-Bus bus driver calls on kdbus, also for
7195 compatibility purposes.
7197 * A new API "sd-event.h" has been added that implements a
7198 minimal event loop API built around epoll. It provides a
7199 couple of features that direct epoll usage is lacking:
7200 prioritization of events, scales to large numbers of timer
7201 events, per-event timer slack (accuracy), system-wide
7202 coalescing of timer events, exit handlers, watchdog
7203 supervision support using systemd's sd_notify() API, child
7206 * A new API "sd-rntl.h" has been added that provides an API
7207 around the route netlink interface of the kernel, similar in
7208 style to "sd-bus.h".
7210 * A new API "sd-dhcp-client.h" has been added that provides a
7211 small DHCPv4 client-side implementation. This is used by
7214 * There is a new kernel command line option
7215 "systemd.restore_state=0|1". When set to "0", none of the
7216 systemd tools will restore saved runtime state to hardware
7217 devices. More specifically, the rfkill and backlight states
7220 * The FsckPassNo= compatibility option in mount/service units
7221 has been removed. The fstab generator will now add the
7222 necessary dependencies automatically, and does not require
7223 PID1's support for that anymore.
7225 * journalctl gained a new switch, --list-boots, that lists
7226 recent boots with their times and boot IDs.
7228 * The various tools like systemctl, loginctl, timedatectl,
7229 busctl, systemd-run, ... have gained a new switch "-M" to
7230 connect to a specific, local OS container (as direct
7231 connection, without requiring SSH). This works on any
7232 container that is registered with machined, such as those
7233 created by libvirt-lxc or nspawn.
7235 * systemd-run and systemd-analyze also gained support for "-H"
7236 to connect to remote hosts via SSH. This is particularly
7237 useful for systemd-run because it enables queuing of jobs
7238 onto remote systems.
7240 * machinectl gained a new command "login" to open a getty
7241 login in any local container. This works with any container
7242 that is registered with machined (such as those created by
7243 libvirt-lxc or nspawn), and which runs systemd inside.
7245 * machinectl gained a new "reboot" command that may be used to
7246 trigger a reboot on a specific container that is registered
7247 with machined. This works on any container that runs an init
7248 system of some kind.
7250 * systemctl gained a new "list-timers" command to print a nice
7251 listing of installed timer units with the times they elapse
7254 * Alternative reboot() parameters may now be specified on the
7255 "systemctl reboot" command line and are passed to the
7256 reboot() system call.
7258 * systemctl gained a new --job-mode= switch to configure the
7259 mode to queue a job with. This is a more generic version of
7260 --fail, --irreversible, and --ignore-dependencies, which are
7261 still available but not advertised anymore.
7263 * /etc/systemd/system.conf gained new settings to configure
7264 various default timeouts of units, as well as the default
7265 start limit interval and burst. These may still be overridden
7268 * PID1 will now export on the bus profile data of the security
7269 policy upload process (such as the SELinux policy upload to
7272 * journald: when forwarding logs to the console, include
7273 timestamps (following the setting in
7274 /sys/module/printk/parameters/time).
7276 * OnCalendar= in timer units now understands the special
7277 strings "yearly" and "annually". (Both are equivalent)
7279 * The accuracy of timer units is now configurable with the new
7280 AccuracySec= setting. It defaults to 1min.
7282 * A new dependency type JoinsNamespaceOf= has been added that
7283 allows running two services within the same /tmp and network
7284 namespace, if PrivateNetwork= or PrivateTmp= are used.
7286 * A new command "cat" has been added to systemctl. It outputs
7287 the original unit file of a unit, and concatenates the
7288 contents of additional "drop-in" unit file snippets, so that
7289 the full configuration is shown.
7291 * systemctl now supports globbing on the various "list-xyz"
7292 commands, like "list-units" or "list-sockets", as well as on
7293 those commands which take multiple unit names.
7295 * journalctl's --unit= switch gained support for globbing.
7297 * All systemd daemons now make use of the watchdog logic so
7298 that systemd automatically notices when they hang.
7300 * If the $container_ttys environment variable is set,
7301 getty-generator will automatically spawn a getty for each
7302 listed tty. This is useful for container managers to request
7303 login gettys to be spawned on as many ttys as needed.
7305 * %h, %s, %U specifier support is not available anymore when
7306 used in unit files for PID 1. This is because NSS calls are
7307 not safe from PID 1. They stay available for --user
7308 instances of systemd, and as special case for the root user.
7310 * loginctl gained a new "--no-legend" switch to turn off output
7313 * The "sd-login.h" API gained three new calls:
7314 sd_session_is_remote(), sd_session_get_remote_user(),
7315 sd_session_get_remote_host() to query information about
7318 * The udev hardware database now also carries vendor/product
7319 information of SDIO devices.
7321 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
7322 determine whether watchdog notifications are requested by
7325 * Socket-activated per-connection services now include a
7326 short description of the connection parameters in the
7329 * tmpfiles gained a new "--boot" option. When this is not used,
7330 only lines where the command character is not suffixed with
7331 "!" are executed. When this option is specified, those
7332 options are executed too. This partitions tmpfiles
7333 directives into those that can be safely executed at any
7334 time, and those which should be run only at boot (for
7335 example, a line that creates /run/nologin).
7337 * A new API "sd-resolve.h" has been added which provides a simple
7338 asynchronous wrapper around glibc NSS host name resolution
7339 calls, such as getaddrinfo(). In contrast to glibc's
7340 getaddrinfo_a(), it does not use signals. In contrast to most
7341 other asynchronous name resolution libraries, this one does
7342 not reimplement DNS, but reuses NSS, so that alternate
7343 host name resolution systems continue to work, such as mDNS,
7344 LDAP, etc. This API is based on libasyncns, but it has been
7345 cleaned up for inclusion in systemd.
7347 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
7348 "sd-daemon.h" are no longer found in individual libraries
7349 libsystemd-journal.so, libsystemd-login.so,
7350 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
7351 merged them into a single library, libsystemd.so, which
7352 provides all symbols. The reason for this is cyclic
7353 dependencies, as these libraries tend to use each other's
7354 symbols. So far, we have managed to workaround that by linking
7355 a copy of a good part of our code into each of these
7356 libraries again and again, which, however, makes certain
7357 things hard to do, like sharing static variables. Also, it
7358 substantially increases footprint. With this change, there
7359 is only one library for the basic APIs systemd
7360 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
7361 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
7362 library as well, however are subject to the --enable-kdbus
7363 switch (see below). Note that "sd-dhcp-client.h" is not part
7364 of this library (this is because it only consumes, never
7365 provides, services of/to other APIs). To make the transition
7366 easy from the separate libraries to the unified one, we
7367 provide the --enable-compat-libs compile-time switch which
7368 will generate stub libraries that are compatible with the
7369 old ones but redirect all calls to the new one.
7371 * All of the kdbus logic and the new APIs "sd-bus.h",
7372 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
7373 and "sd-utf8.h" are compile-time optional via the
7374 "--enable-kdbus" switch, and they are not compiled in by
7375 default. To make use of kdbus, you have to explicitly enable
7376 the switch. Note however, that neither the kernel nor the
7377 userspace API for all of this is considered stable yet. We
7378 want to maintain the freedom to still change the APIs for
7379 now. By specifying this build-time switch, you acknowledge
7380 that you are aware of the instability of the current
7383 * Also, note that while kdbus is pretty much complete,
7384 it lacks one thing: proper policy support. This means you
7385 can build a fully working system with all features; however,
7386 it will be highly insecure. Policy support will be added in
7387 one of the next releases, at the same time that we will
7388 declare the APIs stable.
7390 * When the kernel command line argument "kdbus" is specified,
7391 systemd will automatically load the kdbus.ko kernel module. At
7392 this stage of development, it is only useful for testing kdbus
7393 and should not be used in production. Note: if "--enable-kdbus"
7394 is specified, and the kdbus.ko kernel module is available, and
7395 "kdbus" is added to the kernel command line, the entire system
7396 runs with kdbus instead of dbus-daemon, with the above mentioned
7397 problem of missing the system policy enforcement. Also a future
7398 version of kdbus.ko or a newer systemd will not be compatible with
7399 each other, and will unlikely be able to boot the machine if only
7400 one of them is updated.
7402 * systemctl gained a new "import-environment" command which
7403 uploads the caller's environment (or parts thereof) into the
7404 service manager so that it is inherited by services started
7405 by the manager. This is useful to upload variables like
7406 $DISPLAY into the user service manager.
7408 * A new PrivateDevices= switch has been added to service units
7409 which allows running a service with a namespaced /dev
7410 directory that does not contain any device nodes for
7411 physical devices. More specifically, it only includes devices
7412 such as /dev/null, /dev/urandom, and /dev/zero which are API
7415 * logind has been extended to support behaviour like VT
7416 switching on seats that do not support a VT. This makes
7417 multi-session available on seats that are not the first seat
7418 (seat0), and on systems where kernel support for VTs has
7419 been disabled at compile-time.
7421 * If a process holds a delay lock for system sleep or shutdown
7422 and fails to release it in time, we will now log its
7423 identity. This makes it easier to identify processes that
7424 cause slow suspends or power-offs.
7426 * When parsing /etc/crypttab, support for a new key-slot=
7427 option as supported by Debian is added. It allows indicating
7428 which LUKS slot to use on disk, speeding up key loading.
7430 * The sd_journal_sendv() API call has been checked and
7431 officially declared to be async-signal-safe so that it may
7432 be invoked from signal handlers for logging purposes.
7434 * Boot-time status output is now enabled automatically after a
7435 short timeout if boot does not progress, in order to give
7436 the user an indication what she or he is waiting for.
7438 * The boot-time output has been improved to show how much time
7439 remains until jobs expire.
7441 * The KillMode= switch in service units gained a new possible
7442 value "mixed". If set, and the unit is shut down, then the
7443 initial SIGTERM signal is sent only to the main daemon
7444 process, while the following SIGKILL signal is sent to
7445 all remaining processes of the service.
7447 * When a scope unit is registered, a new property "Controller"
7448 may be set. If set to a valid bus name, systemd will send a
7449 RequestStop() signal to this name when it would like to shut
7450 down the scope. This may be used to hook manager logic into
7451 the shutdown logic of scope units. Also, scope units may now
7452 be put in a special "abandoned" state, in which case the
7453 manager process which created them takes no further
7454 responsibilities for it.
7456 * When reading unit files, systemd will now verify
7457 the access mode of these files, and warn about certain
7458 suspicious combinations. This has been added to make it
7459 easier to track down packaging bugs where unit files are
7460 marked executable or world-writable.
7462 * systemd-nspawn gained a new "--setenv=" switch to set
7463 container-wide environment variables. The similar option in
7464 systemd-activate was renamed from "--environment=" to
7465 "--setenv=" for consistency.
7467 * systemd-nspawn has been updated to create a new kdbus domain
7468 for each container that is invoked, thus allowing each
7469 container to have its own set of system and user buses,
7470 independent of the host.
7472 * systemd-nspawn gained a new --drop-capability= switch to run
7473 the container with less capabilities than the default. Both
7474 --drop-capability= and --capability= now take the special
7475 string "all" for dropping or keeping all capabilities.
7477 * systemd-nspawn gained new switches for executing containers
7478 with specific SELinux labels set.
7480 * systemd-nspawn gained a new --quiet switch to not generate
7481 any additional output but the container's own console
7484 * systemd-nspawn gained a new --share-system switch to run a
7485 container without PID namespacing enabled.
7487 * systemd-nspawn gained a new --register= switch to control
7488 whether the container is registered with systemd-machined or
7489 not. This is useful for containers that do not run full
7490 OS images, but only specific apps.
7492 * systemd-nspawn gained a new --keep-unit which may be used
7493 when invoked as the only program from a service unit, and
7494 results in registration of the unit service itself in
7495 systemd-machined, instead of a newly opened scope unit.
7497 * systemd-nspawn gained a new --network-interface= switch for
7498 moving arbitrary interfaces to the container. The new
7499 --network-veth switch creates a virtual Ethernet connection
7500 between host and container. The new --network-bridge=
7501 switch then allows assigning the host side of this virtual
7502 Ethernet connection to a bridge device.
7504 * systemd-nspawn gained a new --personality= switch for
7505 setting the kernel personality for the container. This is
7506 useful when running a 32-bit container on a 64-bit host. A
7507 similar option Personality= is now also available for service
7510 * logind will now also track a "Desktop" identifier for each
7511 session which encodes the desktop environment of it. This is
7512 useful for desktop environments that want to identify
7513 multiple running sessions of itself easily.
7515 * A new SELinuxContext= setting for service units has been
7516 added that allows setting a specific SELinux execution
7517 context for a service.
7519 * Most systemd client tools will now honour $SYSTEMD_LESS for
7520 settings of the "less" pager. By default, these tools will
7521 override $LESS to allow certain operations to work, such as
7522 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
7523 influence this logic.
7525 * systemd's "seccomp" hook-up has been changed to make use of
7526 the libseccomp library instead of using its own
7527 implementation. This has benefits for portability among
7530 * For usage together with SystemCallFilter=, a new
7531 SystemCallErrorNumber= setting has been introduced that
7532 allows configuration of a system error number to be returned
7533 on filtered system calls, instead of immediately killing the
7534 process. Also, SystemCallArchitectures= has been added to
7535 limit access to system calls of a particular architecture
7536 (in order to turn off support for unused secondary
7537 architectures). There is also a global
7538 SystemCallArchitectures= setting in system.conf now to turn
7539 off support for non-native system calls system-wide.
7541 * systemd requires a kernel with a working name_to_handle_at(),
7542 please see the kernel config requirements in the README file.
7544 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
7545 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
7546 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
7547 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
7548 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
7549 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
7550 Elia Pinto, Florian Weimer, George McCollister, Goffredo
7551 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
7552 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
7553 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
7554 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
7555 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
7556 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
7557 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
7558 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
7559 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
7560 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
7561 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
7562 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
7563 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
7564 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
7565 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
7566 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
7567 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
7569 — Berlin, 2014-02-20
7573 * logind has gained support for facilitating privileged input
7574 and drm device access for unprivileged clients. This work is
7575 useful to allow Wayland display servers (and similar
7576 programs, such as kmscon) to run under the user's ID and
7577 access input and drm devices which are normally
7578 protected. When this is used (and the kernel is new enough)
7579 logind will "mute" IO on the file descriptors passed to
7580 Wayland as long as it is in the background and "unmute" it
7581 if it returns into the foreground. This allows secure
7582 session switching without allowing background sessions to
7583 eavesdrop on input and display data. This also introduces
7584 session switching support if VT support is turned off in the
7585 kernel, and on seats that are not seat0.
7587 * A new kernel command line option luks.options= is understood
7588 now which allows specifying LUKS options for usage for LUKS
7589 encrypted partitions specified with luks.uuid=.
7591 * tmpfiles.d(5) snippets may now use specifier expansion in
7592 path names. More specifically %m, %b, %H, %v, are now
7593 replaced by the local machine id, boot id, hostname, and
7594 kernel version number.
7596 * A new tmpfiles.d(5) command "m" has been introduced which
7597 may be used to change the owner/group/access mode of a file
7598 or directory if it exists, but do nothing if it does not.
7600 * This release removes high-level support for the
7601 MemorySoftLimit= cgroup setting. The underlying kernel
7602 cgroup attribute memory.soft_limit= is currently badly
7603 designed and likely to be removed from the kernel API in its
7604 current form, hence we should not expose it for now.
7606 * The memory.use_hierarchy cgroup attribute is now enabled for
7607 all cgroups systemd creates in the memory cgroup
7608 hierarchy. This option is likely to be come the built-in
7609 default in the kernel anyway, and the non-hierarchical mode
7610 never made much sense in the intrinsically hierarchical
7613 * A new field _SYSTEMD_SLICE= is logged along with all journal
7614 messages containing the slice a message was generated
7615 from. This is useful to allow easy per-customer filtering of
7616 logs among other things.
7618 * systemd-journald will no longer adjust the group of journal
7619 files it creates to the "systemd-journal" group. Instead we
7620 rely on the journal directory to be owned by the
7621 "systemd-journal" group, and its setgid bit set, so that the
7622 kernel file system layer will automatically enforce that
7623 journal files inherit this group assignment. The reason for
7624 this change is that we cannot allow NSS look-ups from
7625 journald which would be necessary to resolve
7626 "systemd-journal" to a numeric GID, because this might
7627 create deadlocks if NSS involves synchronous queries to
7628 other daemons (such as nscd, or sssd) which in turn are
7629 logging clients of journald and might block on it, which
7630 would then dead lock. A tmpfiles.d(5) snippet included in
7631 systemd will make sure the setgid bit and group are
7632 properly set on the journal directory if it exists on every
7633 boot. However, we recommend adjusting it manually after
7634 upgrades too (or from RPM scriptlets), so that the change is
7635 not delayed until next reboot.
7637 * Backlight and random seed files in /var/lib/ have moved into
7638 the /var/lib/systemd/ directory, in order to centralize all
7639 systemd generated files in one directory.
7641 * Boot time performance measurements (as displayed by
7642 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
7643 performance information if that's available to determine how
7644 much time BIOS and boot loader initialization required. With
7645 a sufficiently new BIOS you hence no longer need to boot
7646 with Gummiboot to get access to such information.
7648 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
7649 Cristian Rodríguez, Dave Reisner, David Herrmann, David
7650 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
7651 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
7652 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
7653 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
7654 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
7656 — Berlin, 2013-10-02
7660 * The Restart= option for services now understands a new
7661 on-watchdog setting, which will restart the service
7662 automatically if the service stops sending out watchdog keep
7663 alive messages (as configured with WatchdogSec=).
7665 * The getty generator (which is responsible for bringing up a
7666 getty on configured serial consoles) will no longer only
7667 start a getty on the primary kernel console but on all
7668 others, too. This makes the order in which console= is
7669 specified on the kernel command line less important.
7671 * libsystemd-logind gained a new sd_session_get_vt() call to
7672 retrieve the VT number of a session.
7674 * If the option "tries=0" is set for an entry of /etc/crypttab
7675 its passphrase is queried indefinitely instead of any
7676 maximum number of tries.
7678 * If a service with a configure PID file terminates its PID
7679 file will now be removed automatically if it still exists
7680 afterwards. This should put an end to stale PID files.
7682 * systemd-run will now also take relative binary path names
7683 for execution and no longer insists on absolute paths.
7685 * InaccessibleDirectories= and ReadOnlyDirectories= now take
7686 paths that are optionally prefixed with "-" to indicate that
7687 it should not be considered a failure if they do not exist.
7689 * journalctl -o (and similar commands) now understands a new
7690 output mode "short-precise", it is similar to "short" but
7691 shows timestamps with usec accuracy.
7693 * The option "discard" (as known from Debian) is now
7694 synonymous to "allow-discards" in /etc/crypttab. In fact,
7695 "discard" is preferred now (since it is easier to remember
7698 * Some licensing clean-ups were made, so that more code is now
7699 LGPL-2.1 licensed than before.
7701 * A minimal tool to save/restore the display backlight
7702 brightness across reboots has been added. It will store the
7703 backlight setting as late as possible at shutdown, and
7704 restore it as early as possible during reboot.
7706 * A logic to automatically discover and enable home and swap
7707 partitions on GPT disks has been added. With this in place
7708 /etc/fstab becomes optional for many setups as systemd can
7709 discover certain partitions located on the root disk
7710 automatically. Home partitions are recognized under their
7711 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
7712 partitions are recognized under their GPT type ID
7713 0657fd6da4ab43c484e50933c84b4f4f.
7715 * systemd will no longer pass any environment from the kernel
7716 or initrd to system services. If you want to set an
7717 environment for all services, do so via the kernel command
7718 line systemd.setenv= assignment.
7720 * The systemd-sysctl tool no longer natively reads the file
7721 /etc/sysctl.conf. If desired, the file should be symlinked
7722 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
7723 legacy support by a symlink rather than built-in code, it
7724 also makes the otherwise hidden order of application of the
7725 different files visible. (Note that this partly reverts to a
7726 pre-198 application order of sysctl knobs!)
7728 * The "systemctl set-log-level" and "systemctl dump" commands
7729 have been moved to systemd-analyze.
7731 * systemd-run learned the new --remain-after-exit switch,
7732 which causes the scope unit not to be cleaned up
7733 automatically after the process terminated.
7735 * tmpfiles learned a new --exclude-prefix= switch to exclude
7736 certain paths from operation.
7738 * journald will now automatically flush all messages to disk
7739 as soon as a message at the log level CRIT, ALERT or EMERG
7742 Contributions from: Andrew Cook, Brandon Philips, Christian
7743 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
7744 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
7745 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
7746 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
7747 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
7748 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
7749 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
7750 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
7751 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
7752 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
7753 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
7754 William Giokas, Zbigniew Jędrzejewski-Szmek
7756 — Berlin, 2013-09-13
7760 * The documentation has been updated to cover the various new
7761 concepts introduced with 205.
7763 * Unit files now understand the new %v specifier which
7764 resolves to the kernel version string as returned by "uname
7767 * systemctl now supports filtering the unit list output by
7768 load state, active state and sub state, using the new
7771 * "systemctl status" will now show the results of the
7772 condition checks (like ConditionPathExists= and similar) of
7773 the last start attempts of the unit. They are also logged to
7776 * "journalctl -b" may now be used to look for boot output of a
7777 specific boot. Try "journalctl -b -1" for the previous boot,
7778 but the syntax is substantially more powerful.
7780 * "journalctl --show-cursor" has been added which prints the
7781 cursor string the last shown log line. This may then be used
7782 with the new "journalctl --after-cursor=" switch to continue
7783 browsing logs from that point on.
7785 * "journalctl --force" may now be used to force regeneration
7788 * Creation of "dead" device nodes has been moved from udev
7789 into kmod and tmpfiles. Previously, udev would read the kmod
7790 databases to pre-generate dead device nodes based on meta
7791 information contained in kernel modules, so that these would
7792 be auto-loaded on access rather then at boot. As this
7793 does not really have much to do with the exposing actual
7794 kernel devices to userspace this has always been slightly
7795 alien in the udev codebase. Following the new scheme kmod
7796 will now generate a runtime snippet for tmpfiles from the
7797 module meta information and it now is tmpfiles' job to the
7798 create the nodes. This also allows overriding access and
7799 other parameters for the nodes using the usual tmpfiles
7800 facilities. As side effect this allows us to remove the
7801 CAP_SYS_MKNOD capability bit from udevd entirely.
7803 * logind's device ACLs may now be applied to these "dead"
7804 devices nodes too, thus finally allowing managed access to
7805 devices such as /dev/snd/sequencer without loading the
7806 backing module right-away.
7808 * A new RPM macro has been added that may be used to apply
7809 tmpfiles configuration during package installation.
7811 * systemd-detect-virt and ConditionVirtualization= now can
7812 detect User-Mode-Linux machines (UML).
7814 * journald will now implicitly log the effective capabilities
7815 set of processes in the message metadata.
7817 * systemd-cryptsetup has gained support for TrueCrypt volumes.
7819 * The initrd interface has been simplified (more specifically,
7820 support for passing performance data via environment
7821 variables and fsck results via files in /run has been
7822 removed). These features were non-essential, and are
7823 nowadays available in a much nicer way by having systemd in
7824 the initrd serialize its state and have the hosts systemd
7825 deserialize it again.
7827 * The udev "keymap" data files and tools to apply keyboard
7828 specific mappings of scan to key codes, and force-release
7829 scan code lists have been entirely replaced by a udev
7830 "keyboard" builtin and a hwdb data file.
7832 * systemd will now honour the kernel's "quiet" command line
7833 argument also during late shutdown, resulting in a
7834 completely silent shutdown when used.
7836 * There's now an option to control the SO_REUSEPORT socket
7837 option in .socket units.
7839 * Instance units will now automatically get a per-template
7840 subslice of system.slice unless something else is explicitly
7841 configured. For example, instances of sshd@.service will now
7842 implicitly be placed in system-sshd.slice rather than
7843 system.slice as before.
7845 * Test coverage support may now be enabled at build time.
7847 Contributions from: Dave Reisner, Frederic Crozat, Harald
7848 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
7849 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
7850 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
7851 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
7852 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
7853 Giokas, Zbigniew Jędrzejewski-Szmek
7855 — Berlin, 2013-07-23
7859 * Two new unit types have been introduced:
7861 Scope units are very similar to service units, however, are
7862 created out of pre-existing processes — instead of PID 1
7863 forking off the processes. By using scope units it is
7864 possible for system services and applications to group their
7865 own child processes (worker processes) in a powerful way
7866 which then maybe used to organize them, or kill them
7867 together, or apply resource limits on them.
7869 Slice units may be used to partition system resources in an
7870 hierarchical fashion and then assign other units to them. By
7871 default there are now three slices: system.slice (for all
7872 system services), user.slice (for all user sessions),
7873 machine.slice (for VMs and containers).
7875 Slices and scopes have been introduced primarily in
7876 context of the work to move cgroup handling to a
7877 single-writer scheme, where only PID 1
7878 creates/removes/manages cgroups.
7880 * There's a new concept of "transient" units. In contrast to
7881 normal units these units are created via an API at runtime,
7882 not from configuration from disk. More specifically this
7883 means it is now possible to run arbitrary programs as
7884 independent services, with all execution parameters passed
7885 in via bus APIs rather than read from disk. Transient units
7886 make systemd substantially more dynamic then it ever was,
7887 and useful as a general batch manager.
7889 * logind has been updated to make use of scope and slice units
7890 for managing user sessions. As a user logs in he will get
7891 his own private slice unit, to which all sessions are added
7892 as scope units. We also added support for automatically
7893 adding an instance of user@.service for the user into the
7894 slice. Effectively logind will no longer create cgroup
7895 hierarchies on its own now, it will defer entirely to PID 1
7896 for this by means of scope, service and slice units. Since
7897 user sessions this way become entities managed by PID 1
7898 the output of "systemctl" is now a lot more comprehensive.
7900 * A new mini-daemon "systemd-machined" has been added which
7901 may be used by virtualization managers to register local
7902 VMs/containers. nspawn has been updated accordingly, and
7903 libvirt will be updated shortly. machined will collect a bit
7904 of meta information about the VMs/containers, and assign
7905 them their own scope unit (see above). The collected
7906 meta-data is then made available via the "machinectl" tool,
7907 and exposed in "ps" and similar tools. machined/machinectl
7908 is compile-time optional.
7910 * As discussed earlier, the low-level cgroup configuration
7911 options ControlGroup=, ControlGroupModify=,
7912 ControlGroupPersistent=, ControlGroupAttribute= have been
7913 removed. Please use high-level attribute settings instead as
7914 well as slice units.
7916 * A new bus call SetUnitProperties() has been added to alter
7917 various runtime parameters of a unit. This is primarily
7918 useful to alter cgroup parameters dynamically in a nice way,
7919 but will be extended later on to make more properties
7920 modifiable at runtime. systemctl gained a new set-properties
7921 command that wraps this call.
7923 * A new tool "systemd-run" has been added which can be used to
7924 run arbitrary command lines as transient services or scopes,
7925 while configuring a number of settings via the command
7926 line. This tool is currently very basic, however already
7927 very useful. We plan to extend this tool to even allow
7928 queuing of execution jobs with time triggers from the
7929 command line, similar in fashion to "at".
7931 * nspawn will now inform the user explicitly that kernels with
7932 audit enabled break containers, and suggest the user to turn
7935 * Support for detecting the IMA and AppArmor security
7936 frameworks with ConditionSecurity= has been added.
7938 * journalctl gained a new "-k" switch for showing only kernel
7939 messages, mimicking dmesg output; in addition to "--user"
7940 and "--system" switches for showing only user's own logs
7943 * systemd-delta can now show information about drop-in
7944 snippets extending unit files.
7946 * libsystemd-bus has been substantially updated but is still
7947 not available as public API.
7949 * systemd will now look for the "debug" argument on the kernel
7950 command line and enable debug logging, similar to what
7951 "systemd.log_level=debug" already did before.
7953 * "systemctl set-default", "systemctl get-default" has been
7954 added to configure the default.target symlink, which
7955 controls what to boot into by default.
7957 * "systemctl set-log-level" has been added as a convenient
7958 way to raise and lower systemd logging threshold.
7960 * "systemd-analyze plot" will now show the time the various
7961 generators needed for execution, as well as information
7962 about the unit file loading.
7964 * libsystemd-journal gained a new sd_journal_open_files() call
7965 for opening specific journal files. journactl also gained a
7966 new switch to expose this new functionality. Previously we
7967 only supported opening all files from a directory, or all
7968 files from the system, as opening individual files only is
7969 racy due to journal file rotation.
7971 * systemd gained the new DefaultEnvironment= setting in
7972 /etc/systemd/system.conf to set environment variables for
7975 * If a privileged process logs a journal message with the
7976 OBJECT_PID= field set, then journald will automatically
7977 augment this with additional OBJECT_UID=, OBJECT_GID=,
7978 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
7979 system services want to log events about specific client
7980 processes. journactl/systemctl has been updated to make use
7981 of this information if all log messages regarding a specific
7984 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
7985 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
7986 Reisner, David Coppa, David King, David Strauss, Eelco
7987 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
7988 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
7989 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
7990 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
7991 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
7992 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
7993 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
7994 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
7995 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
7996 Łukasz Stelmach, 장동준
8000 * The Python bindings gained some minimal support for the APIs
8001 exposed by libsystemd-logind.
8003 * ConditionSecurity= gained support for detecting SMACK. Since
8004 this condition already supports SELinux and AppArmor we only
8005 miss IMA for this. Patches welcome!
8007 Contributions from: Karol Lewandowski, Lennart Poettering,
8008 Zbigniew Jędrzejewski-Szmek
8012 * systemd-nspawn will now create /etc/resolv.conf if
8013 necessary, before bind-mounting the host's file onto it.
8015 * systemd-nspawn will now store meta information about a
8016 container on the container's cgroup as extended attribute
8017 fields, including the root directory.
8019 * The cgroup hierarchy has been reworked in many ways. All
8020 objects any of the components systemd creates in the cgroup
8021 tree are now suffixed. More specifically, user sessions are
8022 now placed in cgroups suffixed with ".session", users in
8023 cgroups suffixed with ".user", and nspawn containers in
8024 cgroups suffixed with ".nspawn". Furthermore, all cgroup
8025 names are now escaped in a simple scheme to avoid collision
8026 of userspace object names with kernel filenames. This work
8027 is preparation for making these objects relocatable in the
8028 cgroup tree, in order to allow easy resource partitioning of
8029 these objects without causing naming conflicts.
8031 * systemctl list-dependencies gained the new switches
8032 --plain, --reverse, --after and --before.
8034 * systemd-inhibit now shows the process name of processes that
8035 have taken an inhibitor lock.
8037 * nss-myhostname will now also resolve "localhost"
8038 implicitly. This makes /etc/hosts an optional file and
8039 nicely handles that on IPv6 ::1 maps to both "localhost" and
8042 * libsystemd-logind.so gained a new call
8043 sd_get_machine_names() to enumerate running containers and
8044 VMs (currently only supported by very new libvirt and
8045 nspawn). sd_login_monitor can now be used to watch
8046 VMs/containers coming and going.
8048 * .include is not allowed recursively anymore, and only in
8049 unit files. Usually it is better to use drop-in snippets in
8050 .d/*.conf anyway, as introduced with systemd 198.
8052 * systemd-analyze gained a new "critical-chain" command that
8053 determines the slowest chain of units run during system
8054 boot-up. It is very useful for tracking down where
8055 optimizing boot time is the most beneficial.
8057 * systemd will no longer allow manipulating service paths in
8058 the name=systemd:/system cgroup tree using ControlGroup= in
8059 units. (But is still fine with it in all other dirs.)
8061 * There's a new systemd-nspawn@.service service file that may
8062 be used to easily run nspawn containers as system
8063 services. With the container's root directory in
8064 /var/lib/container/foobar it is now sufficient to run
8065 "systemctl start systemd-nspawn@foobar.service" to boot it.
8067 * systemd-cgls gained a new parameter "--machine" to list only
8068 the processes within a certain container.
8070 * ConditionSecurity= now can check for "apparmor". We still
8071 are lacking checks for SMACK and IMA for this condition
8072 check though. Patches welcome!
8074 * A new configuration file /etc/systemd/sleep.conf has been
8075 added that may be used to configure which kernel operation
8076 systemd is supposed to execute when "suspend", "hibernate"
8077 or "hybrid-sleep" is requested. This makes the new kernel
8078 "freeze" state accessible to the user.
8080 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
8081 the passed argument if applicable.
8083 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
8084 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
8085 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
8086 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
8087 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
8088 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
8089 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
8094 * The output of 'systemctl list-jobs' got some polishing. The
8095 '--type=' argument may now be passed more than once. A new
8096 command 'systemctl list-sockets' has been added which shows
8097 a list of kernel sockets systemd is listening on with the
8098 socket units they belong to, plus the units these socket
8101 * The experimental libsystemd-bus library got substantial
8102 updates to work in conjunction with the (also experimental)
8103 kdbus kernel project. It works well enough to exchange
8104 messages with some sophistication. Note that kdbus is not
8105 ready yet, and the library is mostly an elaborate test case
8106 for now, and not installable.
8108 * systemd gained a new unit 'systemd-static-nodes.service'
8109 that generates static device nodes earlier during boot, and
8110 can run in conjunction with udev.
8112 * libsystemd-login gained a new call sd_pid_get_user_unit()
8113 to retrieve the user systemd unit a process is running
8114 in. This is useful for systems where systemd is used as
8117 * systemd-nspawn now places all containers in the new /machine
8118 top-level cgroup directory in the name=systemd
8119 hierarchy. libvirt will soon do the same, so that we get a
8120 uniform separation of /system, /user and /machine for system
8121 services, user processes and containers/virtual
8122 machines. This new cgroup hierarchy is also useful to stick
8123 stable names to specific container instances, which can be
8124 recognized later this way (this name may be controlled
8125 via systemd-nspawn's new -M switch). libsystemd-login also
8126 gained a new call sd_pid_get_machine_name() to retrieve the
8127 name of the container/VM a specific process belongs to.
8129 * bootchart can now store its data in the journal.
8131 * libsystemd-journal gained a new call
8132 sd_journal_add_conjunction() for AND expressions to the
8133 matching logic. This can be used to express more complex
8134 logical expressions.
8136 * journactl can now take multiple --unit= and --user-unit=
8139 * The cryptsetup logic now understands the "luks.key=" kernel
8140 command line switch for specifying a file to read the
8141 decryption key from. Also, if a configured key file is not
8142 found the tool will now automatically fall back to prompting
8145 * Python systemd.journal module was updated to wrap recently
8146 added functions from libsystemd-journal. The interface was
8147 changed to bring the low level interface in s.j._Reader
8148 closer to the C API, and the high level interface in
8149 s.j.Reader was updated to wrap and convert all data about
8152 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
8153 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
8154 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
8155 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
8156 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
8157 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
8161 * journalctl --update-catalog now understands a new --root=
8162 option to operate on catalogs found in a different root
8165 * During shutdown after systemd has terminated all running
8166 services a final killing loop kills all remaining left-over
8167 processes. We will now print the name of these processes
8168 when we send SIGKILL to them, since this usually indicates a
8171 * If /etc/crypttab refers to password files stored on
8172 configured mount points automatic dependencies will now be
8173 generated to ensure the specific mount is established first
8174 before the key file is attempted to be read.
8176 * 'systemctl status' will now show information about the
8177 network sockets a socket unit is listening on.
8179 * 'systemctl status' will also shown information about any
8180 drop-in configuration file for units. (Drop-In configuration
8181 files in this context are files such as
8182 /etc/systemd/systemd/foobar.service.d/*.conf)
8184 * systemd-cgtop now optionally shows summed up CPU times of
8185 cgroups. Press '%' while running cgtop to switch between
8186 percentage and absolute mode. This is useful to determine
8187 which cgroups use up the most CPU time over the entire
8188 runtime of the system. systemd-cgtop has also been updated
8189 to be 'pipeable' for processing with further shell tools.
8191 * 'hostnamectl set-hostname' will now allow setting of FQDN
8194 * The formatting and parsing of time span values has been
8195 changed. The parser now understands fractional expressions
8196 such as "5.5h". The formatter will now output fractional
8197 expressions for all time spans under 1min, i.e. "5.123456s"
8198 rather than "5s 123ms 456us". For time spans under 1s
8199 millisecond values are shown, for those under 1ms
8200 microsecond values are shown. This should greatly improve
8201 all time-related output of systemd.
8203 * libsystemd-login and libsystemd-journal gained new
8204 functions for querying the poll() events mask and poll()
8205 timeout value for integration into arbitrary event
8208 * localectl gained the ability to list available X11 keymaps
8209 (models, layouts, variants, options).
8211 * 'systemd-analyze dot' gained the ability to filter for
8212 specific units via shell-style globs, to create smaller,
8213 more useful graphs. I.e. it is now possible to create simple
8214 graphs of all the dependencies between only target units, or
8215 of all units that Avahi has dependencies with.
8217 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
8218 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
8219 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
8220 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
8221 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
8222 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
8223 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
8227 * The boot-time readahead implementation for rotating media
8228 will now read the read-ahead data in multiple passes which
8229 consist of all read requests made in equidistant time
8230 intervals. This means instead of strictly reading read-ahead
8231 data in its physical order on disk we now try to find a
8232 middle ground between physical and access time order.
8234 * /etc/os-release files gained a new BUILD_ID= field for usage
8235 on operating systems that provide continuous builds of OS
8238 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
8239 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
8240 William Douglas, Zbigniew Jędrzejewski-Szmek
8244 * systemd-python gained an API exposing libsystemd-daemon.
8246 * The SMACK setup logic gained support for uploading CIPSO
8249 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
8250 ReadOnlyDirectories= and InaccessibleDirectories= has
8251 changed. The private /tmp and /var/tmp directories are now
8252 shared by all processes of a service (which means
8253 ExecStartPre= may now leave data in /tmp that ExecStart= of
8254 the same service can still access). When a service is
8255 stopped its temporary directories are immediately deleted
8256 (normal clean-up with tmpfiles is still done in addition to
8259 * By default, systemd will now set a couple of sysctl
8260 variables in the kernel: the safe sysrq options are turned
8261 on, IP route verification is turned on, and source routing
8262 disabled. The recently added hardlink and softlink
8263 protection of the kernel is turned on. These settings should
8264 be reasonably safe, and good defaults for all new systems.
8266 * The predictable network naming logic may now be turned off
8267 with a new kernel command line switch: net.ifnames=0.
8269 * A new libsystemd-bus module has been added that implements a
8270 pretty complete D-Bus client library. For details see:
8272 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
8274 * journald will now explicitly flush the journal files to disk
8275 at the latest 5min after each write. The file will then also
8276 be marked offline until the next write. This should increase
8277 reliability in case of a crash. The synchronization delay
8278 can be configured via SyncIntervalSec= in journald.conf.
8280 * There's a new remote-fs-setup.target unit that can be used
8281 to pull in specific services when at least one remote file
8282 system is to be mounted.
8284 * There are new targets timers.target and paths.target as
8285 canonical targets to pull user timer and path units in
8286 from. This complements sockets.target with a similar
8287 purpose for socket units.
8289 * libudev gained a new call udev_device_set_attribute_value()
8290 to set sysfs attributes of a device.
8292 * The udev daemon now sets the default number of worker
8293 processes executed in parallel based on the number of available
8294 CPUs instead of the amount of available RAM. This is supposed
8295 to provide a more reliable default and limit a too aggressive
8296 parallelism for setups with 1000s of devices connected.
8298 Contributions from: Auke Kok, Colin Walters, Cristian
8299 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
8300 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
8301 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
8302 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
8303 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
8304 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
8305 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
8306 Zbigniew Jędrzejewski-Szmek
8310 * Configuration of unit files may now be extended via drop-in
8311 files without having to edit/override the unit files
8312 themselves. More specifically, if the administrator wants to
8313 change one value for a service file foobar.service he can
8314 now do so by dropping in a configuration snippet into
8315 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
8316 will load all these snippets and apply them on top of the
8317 main unit configuration file, possibly extending or
8318 overriding its settings. Using these drop-in snippets is
8319 generally nicer than the two earlier options for changing
8320 unit files locally: copying the files from
8321 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
8322 them there; or creating a new file in /etc/systemd/system/
8323 that incorporates the original one via ".include". Drop-in
8324 snippets into these .d/ directories can be placed in any
8325 directory systemd looks for units in, and the usual
8326 overriding semantics between /usr/lib, /etc and /run apply
8329 * Most unit file settings which take lists of items can now be
8330 reset by assigning the empty string to them. For example,
8331 normally, settings such as Environment=FOO=BAR append a new
8332 environment variable assignment to the environment block,
8333 each time they are used. By assigning Environment= the empty
8334 string the environment block can be reset to empty. This is
8335 particularly useful with the .d/*.conf drop-in snippets
8336 mentioned above, since this adds the ability to reset list
8337 settings from vendor unit files via these drop-ins.
8339 * systemctl gained a new "list-dependencies" command for
8340 listing the dependencies of a unit recursively.
8342 * Inhibitors are now honored and listed by "systemctl
8343 suspend", "systemctl poweroff" (and similar) too, not only
8344 GNOME. These commands will also list active sessions by
8347 * Resource limits (as exposed by the various control group
8348 controllers) can now be controlled dynamically at runtime
8349 for all units. More specifically, you can now use a command
8350 like "systemctl set-cgroup-attr foobar.service cpu.shares
8351 2000" to alter the CPU shares a specific service gets. These
8352 settings are stored persistently on disk, and thus allow the
8353 administrator to easily adjust the resource usage of
8354 services with a few simple commands. This dynamic resource
8355 management logic is also available to other programs via the
8356 bus. Almost any kernel cgroup attribute and controller is
8359 * systemd-vconsole-setup will now copy all font settings to
8360 all allocated VTs, where it previously applied them only to
8363 * libsystemd-login gained the new sd_session_get_tty() API
8366 * This release drops support for a few legacy or
8367 distribution-specific LSB facility names when parsing init
8368 scripts: $x-display-manager, $mail-transfer-agent,
8369 $mail-transport-agent, $mail-transfer-agent, $smtp,
8370 $null. Also, the mail-transfer-agent.target unit backing
8371 this has been removed. Distributions which want to retain
8372 compatibility with this should carry the burden for
8373 supporting this themselves and patch support for these back
8374 in, if they really need to. Also, the facilities $syslog and
8375 $local_fs are now ignored, since systemd does not support
8376 early-boot LSB init scripts anymore, and these facilities
8377 are implied anyway for normal services. syslog.target has
8380 * There are new bus calls on PID1's Manager object for
8381 cancelling jobs, and removing snapshot units. Previously,
8382 both calls were only available on the Job and Snapshot
8385 * systemd-journal-gatewayd gained SSL support.
8387 * The various "environment" files, such as /etc/locale.conf
8388 now support continuation lines with a backslash ("\") as
8389 last character in the line, similarly in style (but different)
8390 to how this is supported in shells.
8392 * For normal user processes the _SYSTEMD_USER_UNIT= field is
8393 now implicitly appended to every log entry logged. systemctl
8394 has been updated to filter by this field when operating on a
8395 user systemd instance.
8397 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
8398 CAP_AUDIT_CONTROL capabilities to the capabilities set for
8399 the container. This makes it easier to boot unmodified
8400 Fedora systems in a container, which however still requires
8401 audit=0 to be passed on the kernel command line. Auditing in
8402 kernel and userspace is unfortunately still too broken in
8403 context of containers, hence we recommend compiling it out
8404 of the kernel or using audit=0. Hopefully this will be fixed
8405 one day for good in the kernel.
8407 * nspawn gained the new --bind= and --bind-ro= parameters to
8408 bind mount specific directories from the host into the
8411 * nspawn will now mount its own devpts file system instance
8412 into the container, in order not to leak pty devices from
8413 the host into the container.
8415 * systemd will now read the firmware boot time performance
8416 information from the EFI variables, if the used boot loader
8417 supports this, and takes it into account for boot performance
8418 analysis via "systemd-analyze". This is currently supported
8419 only in conjunction with Gummiboot, but could be supported
8420 by other boot loaders too. For details see:
8422 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
8424 * A new generator has been added that automatically mounts the
8425 EFI System Partition (ESP) to /boot, if that directory
8426 exists, is empty, and no other file system has been
8427 configured to be mounted there.
8429 * logind will now send out PrepareForSleep(false) out
8430 unconditionally, after coming back from suspend. This may be
8431 used by applications as asynchronous notification for
8432 system resume events.
8434 * "systemctl unlock-sessions" has been added, that allows
8435 unlocking the screens of all user sessions at once, similar
8436 to how "systemctl lock-sessions" already locked all users
8437 sessions. This is backed by a new D-Bus call UnlockSessions().
8439 * "loginctl seat-status" will now show the master device of a
8440 seat. (i.e. the device of a seat that needs to be around for
8441 the seat to be considered available, usually the graphics
8444 * tmpfiles gained a new "X" line type, that allows
8445 configuration of files and directories (with wildcards) that
8446 shall be excluded from automatic cleanup ("aging").
8448 * udev default rules set the device node permissions now only
8449 at "add" events, and do not change them any longer with a
8450 later "change" event.
8452 * The log messages for lid events and power/sleep keypresses
8453 now carry a message ID.
8455 * We now have a substantially larger unit test suite, but this
8456 continues to be work in progress.
8458 * udevadm hwdb gained a new --root= parameter to change the
8459 root directory to operate relative to.
8461 * logind will now issue a background sync() request to the kernel
8462 early at shutdown, so that dirty buffers are flushed to disk early
8463 instead of at the last moment, in order to optimize shutdown
8466 * A new bootctl tool has been added that is an interface for
8467 certain boot loader operations. This is currently a preview
8468 and is likely to be extended into a small mechanism daemon
8469 like timedated, localed, hostnamed, and can be used by
8470 graphical UIs to enumerate available boot options, and
8471 request boot into firmware operations.
8473 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
8474 the rest of the package. It also has been updated to work
8475 correctly in initrds.
8477 * polkit previously has been runtime optional, and is now also
8478 compile time optional via a configure switch.
8480 * systemd-analyze has been reimplemented in C. Also "systemctl
8481 dot" has moved into systemd-analyze.
8483 * "systemctl status" with no further parameters will now print
8484 the status of all active or failed units.
8486 * Operations such as "systemctl start" can now be executed
8487 with a new mode "--irreversible" which may be used to queue
8488 operations that cannot accidentally be reversed by a later
8489 job queuing. This is by default used to make shutdown
8490 requests more robust.
8492 * The Python API of systemd now gained a new module for
8493 reading journal files.
8495 * A new tool kernel-install has been added that can install
8496 kernel images according to the Boot Loader Specification:
8498 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
8500 * Boot time console output has been improved to provide
8501 animated boot time output for hanging jobs.
8503 * A new tool systemd-activate has been added which can be used
8504 to test socket activation with, directly from the command
8505 line. This should make it much easier to test and debug
8506 socket activation in daemons.
8508 * journalctl gained a new "--reverse" (or -r) option to show
8509 journal output in reverse order (i.e. newest line first).
8511 * journalctl gained a new "--pager-end" (or -e) option to jump
8512 to immediately jump to the end of the journal in the
8513 pager. This is only supported in conjunction with "less".
8515 * journalctl gained a new "--user-unit=" option, that works
8516 similarly to "--unit=" but filters for user units rather than
8519 * A number of unit files to ease adoption of systemd in
8520 initrds has been added. This moves some minimal logic from
8521 the various initrd implementations into systemd proper.
8523 * The journal files are now owned by a new group
8524 "systemd-journal", which exists specifically to allow access
8525 to the journal, and nothing else. Previously, we used the
8526 "adm" group for that, which however possibly covers more
8527 than just journal/log file access. This new group is now
8528 already used by systemd-journal-gatewayd to ensure this
8529 daemon gets access to the journal files and as little else
8530 as possible. Note that "make install" will also set FS ACLs
8531 up for /var/log/journal to give "adm" and "wheel" read
8532 access to it, in addition to "systemd-journal" which owns
8533 the journal files. We recommend that packaging scripts also
8534 add read access to "adm" + "wheel" to /var/log/journal, and
8535 all existing/future journal files. To normal users and
8536 administrators little changes, however packagers need to
8537 ensure to create the "systemd-journal" system group at
8538 package installation time.
8540 * The systemd-journal-gatewayd now runs as unprivileged user
8541 systemd-journal-gateway:systemd-journal-gateway. Packaging
8542 scripts need to create these system user/group at
8545 * timedated now exposes a new boolean property CanNTP that
8546 indicates whether a local NTP service is available or not.
8548 * systemd-detect-virt will now also detect xen PVs
8550 * The pstore file system is now mounted by default, if it is
8553 * In addition to the SELinux and IMA policies we will now also
8554 load SMACK policies at early boot.
8556 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
8557 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
8558 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
8559 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
8560 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
8561 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
8562 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
8563 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
8564 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
8565 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
8566 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
8567 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
8568 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
8569 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
8573 * Timer units now support calendar time events in addition to
8574 monotonic time events. That means you can now trigger a unit
8575 based on a calendar time specification such as "Thu,Fri
8576 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
8577 or fifth day of any month of the year 2013, given that it is
8578 a thursday or friday. This brings timer event support
8579 considerably closer to cron's capabilities. For details on
8580 the supported calendar time specification language see
8583 * udev now supports a number of different naming policies for
8584 network interfaces for predictable names, and a combination
8585 of these policies is now the default. Please see this wiki
8586 document for details:
8588 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
8590 * Auke Kok's bootchart implementation has been added to the
8591 systemd tree. It is an optional component that can graph the
8592 boot in quite some detail. It is one of the best bootchart
8593 implementations around and minimal in its code and
8596 * nss-myhostname has been integrated into the systemd source
8597 tree. nss-myhostname guarantees that the local hostname
8598 always stays resolvable via NSS. It has been a weak
8599 requirement of systemd-hostnamed since a long time, and
8600 since its code is actually trivial we decided to just
8601 include it in systemd's source tree. It can be turned off
8602 with a configure switch.
8604 * The read-ahead logic is now capable of properly detecting
8605 whether a btrfs file system is on SSD or rotating media, in
8606 order to optimize the read-ahead scheme. Previously, it was
8607 only capable of detecting this on traditional file systems
8610 * In udev, additional device properties are now read from the
8611 IAB in addition to the OUI database. Also, Bluetooth company
8612 identities are attached to the devices as well.
8614 * In service files %U may be used as specifier that is
8615 replaced by the configured user name of the service.
8617 * nspawn may now be invoked without a controlling TTY. This
8618 makes it suitable for invocation as its own service. This
8619 may be used to set up a simple containerized server system
8620 using only core OS tools.
8622 * systemd and nspawn can now accept socket file descriptors
8623 when they are started for socket activation. This enables
8624 implementation of socket activated nspawn
8625 containers. i.e. think about autospawning an entire OS image
8626 when the first SSH or HTTP connection is received. We expect
8627 that similar functionality will also be added to libvirt-lxc
8630 * journalctl will now suppress ANSI color codes when
8631 presenting log data.
8633 * systemctl will no longer show control group information for
8634 a unit if the control group is empty anyway.
8636 * logind can now automatically suspend/hibernate/shutdown the
8639 * /etc/machine-info and hostnamed now also expose the chassis
8640 type of the system. This can be used to determine whether
8641 the local system is a laptop, desktop, handset or
8642 tablet. This information may either be configured by the
8643 user/vendor or is automatically determined from ACPI and DMI
8644 information if possible.
8646 * A number of polkit actions are now bound together with "imply"
8647 rules. This should simplify creating UIs because many actions
8648 will now authenticate similar ones as well.
8650 * Unit files learnt a new condition ConditionACPower= which
8651 may be used to conditionalize a unit depending on whether an
8652 AC power source is connected or not, of whether the system
8653 is running on battery power.
8655 * systemctl gained a new "is-failed" verb that may be used in
8656 shell scripts and suchlike to check whether a specific unit
8657 is in the "failed" state.
8659 * The EnvironmentFile= setting in unit files now supports file
8660 globbing, and can hence be used to easily read a number of
8661 environment files at once.
8663 * systemd will no longer detect and recognize specific
8664 distributions. All distribution-specific #ifdeffery has been
8665 removed, systemd is now fully generic and
8666 distribution-agnostic. Effectively, not too much is lost as
8667 a lot of the code is still accessible via explicit configure
8668 switches. However, support for some distribution specific
8669 legacy configuration file formats has been dropped. We
8670 recommend distributions to simply adopt the configuration
8671 files everybody else uses now and convert the old
8672 configuration from packaging scripts. Most distributions
8673 already did that. If that's not possible or desirable,
8674 distributions are welcome to forward port the specific
8675 pieces of code locally from the git history.
8677 * When logging a message about a unit systemd will now always
8678 log the unit name in the message meta data.
8680 * localectl will now also discover system locale data that is
8681 not stored in locale archives, but directly unpacked.
8683 * logind will no longer unconditionally use framebuffer
8684 devices as seat masters, i.e. as devices that are required
8685 to be existing before a seat is considered preset. Instead,
8686 it will now look for all devices that are tagged as
8687 "seat-master" in udev. By default, framebuffer devices will
8688 be marked as such, but depending on local systems, other
8689 devices might be marked as well. This may be used to
8690 integrate graphics cards using closed source drivers (such
8691 as NVidia ones) more nicely into logind. Note however, that
8692 we recommend using the open source NVidia drivers instead,
8693 and no udev rules for the closed-source drivers will be
8694 shipped from us upstream.
8696 Contributions from: Adam Williamson, Alessandro Crismani, Auke
8697 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
8698 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
8699 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
8700 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
8701 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
8702 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
8703 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
8704 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
8705 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
8706 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
8707 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
8712 * udev gained support for loading additional device properties
8713 from an indexed database that is keyed by vendor/product IDs
8714 and similar device identifiers. For the beginning this
8715 "hwdb" is populated with data from the well-known PCI and
8716 USB database, but also includes PNP, ACPI and OID data. In
8717 the longer run this indexed database shall grow into
8718 becoming the one central database for non-essential
8719 userspace device metadata. Previously, data from the PCI/USB
8720 database was only attached to select devices, since the
8721 lookup was a relatively expensive operation due to O(n) time
8722 complexity (with n being the number of entries in the
8723 database). Since this is now O(1), we decided to add in this
8724 data for all devices where this is available, by
8725 default. Note that the indexed database needs to be rebuilt
8726 when new data files are installed. To achieve this you need
8727 to update your packaging scripts to invoke "udevadm hwdb
8728 --update" after installation of hwdb data files. For
8729 RPM-based distributions we introduced the new
8730 %udev_hwdb_update macro for this purpose.
8732 * The Journal gained support for the "Message Catalog", an
8733 indexed database to link up additional information with
8734 journal entries. For further details please check:
8736 https://www.freedesktop.org/wiki/Software/systemd/catalog
8738 The indexed message catalog database also needs to be
8739 rebuilt after installation of message catalog files. Use
8740 "journalctl --update-catalog" for this. For RPM-based
8741 distributions we introduced the %journal_catalog_update
8742 macro for this purpose.
8744 * The Python Journal bindings gained support for the standard
8745 Python logging framework.
8747 * The Journal API gained new functions for checking whether
8748 the underlying file system of a journal file is capable of
8749 properly reporting file change notifications, or whether
8750 applications that want to reflect journal changes "live"
8751 need to recheck journal files continuously in appropriate
8754 * It is now possible to set the "age" field for tmpfiles
8755 entries to 0, indicating that files matching this entry
8756 shall always be removed when the directories are cleaned up.
8758 * coredumpctl gained a new "gdb" verb which invokes gdb
8759 right-away on the selected coredump.
8761 * There's now support for "hybrid sleep" on kernels that
8762 support this, in addition to "suspend" and "hibernate". Use
8763 "systemctl hybrid-sleep" to make use of this.
8765 * logind's HandleSuspendKey= setting (and related settings)
8766 now gained support for a new "lock" setting to simply
8767 request the screen lock on all local sessions, instead of
8768 actually executing a suspend or hibernation.
8770 * systemd will now mount the EFI variables file system by
8773 * Socket units now gained support for configuration of the
8774 SMACK security label.
8776 * timedatectl will now output the time of the last and next
8777 daylight saving change.
8779 * We dropped support for various legacy and distro-specific
8780 concepts, such as insserv, early-boot SysV services
8781 (i.e. those for non-standard runlevels such as 'b' or 'S')
8782 or ArchLinux /etc/rc.conf support. We recommend the
8783 distributions who still need support this to either continue
8784 to maintain the necessary patches downstream, or find a
8785 different solution. (Talk to us if you have questions!)
8787 * Various systemd components will now bypass polkit checks for
8788 root and otherwise handle properly if polkit is not found to
8789 be around. This should fix most issues for polkit-less
8790 systems. Quite frankly this should have been this way since
8791 day one. It is absolutely our intention to make systemd work
8792 fine on polkit-less systems, and we consider it a bug if
8793 something does not work as it should if polkit is not around.
8795 * For embedded systems it is now possible to build udev and
8796 systemd without blkid and/or kmod support.
8798 * "systemctl switch-root" is now capable of switching root
8799 more than once. I.e. in addition to transitions from the
8800 initrd to the host OS it is now possible to transition to
8801 further OS images from the host. This is useful to implement
8802 offline updating tools.
8804 * Various other additions have been made to the RPM macros
8805 shipped with systemd. Use %udev_rules_update() after
8806 installing new udev rules files. %_udevhwdbdir,
8807 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
8808 %_sysctldir are now available which resolve to the right
8809 directories for packages to place various data files in.
8811 * journalctl gained the new --full switch (in addition to
8812 --all, to disable ellipsation for long messages.
8814 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
8815 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
8816 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
8817 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
8818 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
8819 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
8820 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
8821 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
8822 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
8826 * journalctl gained new --since= and --until= switches to
8827 filter by time. It also now supports nice filtering for
8828 units via --unit=/-u.
8830 * Type=oneshot services may use ExecReload= and do the
8833 * The journal daemon now supports time-based rotation and
8834 vacuuming, in addition to the usual disk-space based
8837 * The journal will now index the available field values for
8838 each field name. This enables clients to show pretty drop
8839 downs of available match values when filtering. The bash
8840 completion of journalctl has been updated
8841 accordingly. journalctl gained a new switch -F to list all
8842 values a certain field takes in the journal database.
8844 * More service events are now written as structured messages
8845 to the journal, and made recognizable via message IDs.
8847 * The timedated, localed and hostnamed mini-services which
8848 previously only provided support for changing time, locale
8849 and hostname settings from graphical DEs such as GNOME now
8850 also have a minimal (but very useful) text-based client
8851 utility each. This is probably the nicest way to changing
8852 these settings from the command line now, especially since
8853 it lists available options and is fully integrated with bash
8856 * There's now a new tool "systemd-coredumpctl" to list and
8857 extract coredumps from the journal.
8859 * We now install a README each in /var/log/ and
8860 /etc/rc.d/init.d explaining where the system logs and init
8861 scripts went. This hopefully should help folks who go to
8862 that dirs and look into the otherwise now empty void and
8863 scratch their heads.
8865 * When user-services are invoked (by systemd --user) the
8866 $MANAGERPID env var is set to the PID of systemd.
8868 * SIGRTMIN+24 when sent to a --user instance will now result
8869 in immediate termination of systemd.
8871 * gatewayd received numerous feature additions such as a
8872 "follow" mode, for live syncing and filtering.
8874 * browse.html now allows filtering and showing detailed
8875 information on specific entries. Keyboard navigation and
8876 mouse screen support has been added.
8878 * gatewayd/journalctl now supports HTML5/JSON
8879 Server-Sent-Events as output.
8881 * The SysV init script compatibility logic will now
8882 heuristically determine whether a script supports the
8883 "reload" verb, and only then make this available as
8886 * "systemctl status --follow" has been removed, use "journalctl
8889 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
8890 have been removed since they are hardly useful to be
8893 * And I'd like to take the opportunity to specifically mention
8894 Zbigniew for his great contributions. Zbigniew, you rock!
8896 Contributions from: Andrew Eikum, Christian Hesse, Colin
8897 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
8898 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
8899 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
8900 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
8901 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
8902 Jędrzejewski-Szmek, Сковорода Никита Андреевич
8906 * If /etc/vconsole.conf is non-existent or empty we will no
8907 longer load any console font or key map at boot by
8908 default. Instead the kernel defaults will be left
8909 intact. This is definitely the right thing to do, as no
8910 configuration should mean no configuration, and hard-coding
8911 font names that are different on all archs is probably a bad
8912 idea. Also, the kernel default key map and font should be
8913 good enough for most cases anyway, and mostly identical to
8914 the userspace fonts/key maps we previously overloaded them
8915 with. If distributions want to continue to default to a
8916 non-kernel font or key map they should ship a default
8917 /etc/vconsole.conf with the appropriate contents.
8919 Contributions from: Colin Walters, Daniel J Walsh, Dave
8920 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
8921 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
8925 * journalctl gained a new --cursor= switch to show entries
8926 starting from the specified location in the journal.
8928 * We now enforce a size limit on journal entry fields exported
8929 with "-o json" in journalctl. Fields larger than 4K will be
8930 assigned null. This can be turned off with --all.
8932 * An (optional) journal gateway daemon is now available as
8933 "systemd-journal-gatewayd.service". This service provides
8934 access to the journal via HTTP and JSON. This functionality
8935 will be used to implement live log synchronization in both
8936 pull and push modes, but has various other users too, such
8937 as easy log access for debugging of embedded devices. Right
8938 now it is already useful to retrieve the journal via HTTP:
8940 # systemctl start systemd-journal-gatewayd.service
8941 # wget http://localhost:19531/entries
8943 This will download the journal contents in a
8944 /var/log/messages compatible format. The same as JSON:
8946 # curl -H"Accept: application/json" http://localhost:19531/entries
8948 This service is also accessible via a web browser where a
8949 single static HTML5 app is served that uses the JSON logic
8950 to enable the user to do some basic browsing of the
8951 journal. This will be extended later on. Here's an example
8952 screenshot of this app in its current state:
8954 http://0pointer.de/public/journal-gatewayd
8956 Contributions from: Kay Sievers, Lennart Poettering, Robert
8957 Milasan, Tom Gundersen
8961 * The bash completion logic is now available for journalctl
8964 * We do not mount the "cpuset" controller anymore together with
8965 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
8966 started if no parameters are assigned to it. "cpuset" hence
8967 broke code that assumed it could create "cpu" groups and
8970 * journalctl -f will now subscribe to terminal size changes,
8971 and line break accordingly.
8973 Contributions from: Dave Reisner, Kay Sievers, Lennart
8974 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
8978 * nspawn will now create a symlink /etc/localtime in the
8979 container environment, copying the host's timezone
8980 setting. Previously this has been done via a bind mount, but
8981 since symlinks cannot be bind mounted this has now been
8982 changed to create/update the appropriate symlink.
8984 * journalctl -n's line number argument is now optional, and
8985 will default to 10 if omitted.
8987 * journald will now log the maximum size the journal files may
8988 take up on disk. This is particularly useful if the default
8989 built-in logic of determining this parameter from the file
8990 system size is used. Use "systemctl status
8991 systemd-journald.service" to see this information.
8993 * The multi-seat X wrapper tool has been stripped down. As X
8994 is now capable of enumerating graphics devices via udev in a
8995 seat-aware way the wrapper is not strictly necessary
8996 anymore. A stripped down temporary stop-gap is still shipped
8997 until the upstream display managers have been updated to
8998 fully support the new X logic. Expect this wrapper to be
8999 removed entirely in one of the next releases.
9001 * HandleSleepKey= in logind.conf has been split up into
9002 HandleSuspendKey= and HandleHibernateKey=. The old setting
9003 is not available anymore. X11 and the kernel are
9004 distinguishing between these keys and we should too. This
9005 also means the inhibition lock for these keys has been split
9008 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
9009 Poettering, Lukas Nykryn, Václav Pavlín
9013 * Whenever a unit changes state we will now log this to the
9014 journal and show along the unit's own log output in
9017 * ConditionPathIsMountPoint= can now properly detect bind
9018 mount points too. (Previously, a bind mount of one file
9019 system to another place in the same file system could not be
9020 detected as mount, since they shared struct stat's st_dev
9023 * We will now mount the cgroup controllers cpu, cpuacct,
9024 cpuset and the controllers net_cls, net_prio together by
9027 * nspawn containers will now have a virtualized boot
9028 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
9029 over with a randomized ID at container initialization). This
9030 has the effect of making "journalctl -b" do the right thing
9033 * The JSON output journal serialization has been updated not
9034 to generate "endless" list objects anymore, but rather one
9035 JSON object per line. This is more in line how most JSON
9036 parsers expect JSON objects. The new output mode
9037 "json-pretty" has been added to provide similar output, but
9038 neatly aligned for readability by humans.
9040 * We dropped all explicit sync() invocations in the shutdown
9041 code. The kernel does this implicitly anyway in the kernel
9042 reboot() syscall. halt(8)'s -n option is now a compatibility
9045 * We now support virtualized reboot() in containers, as
9046 supported by newer kernels. We will fall back to exit() if
9047 CAP_SYS_REBOOT is not available to the container. Also,
9048 nspawn makes use of this now and will actually reboot the
9049 container if the containerized OS asks for that.
9051 * journalctl will only show local log output by default
9052 now. Use --merge (-m) to show remote log output, too.
9054 * libsystemd-journal gained the new sd_journal_get_usage()
9055 call to determine the current disk usage of all journal
9056 files. This is exposed in the new "journalctl --disk-usage"
9059 * journald gained a new configuration setting SplitMode= in
9060 journald.conf which may be used to control how user journals
9061 are split off. See journald.conf(5) for details.
9063 * A new condition type ConditionFileNotEmpty= has been added.
9065 * tmpfiles' "w" lines now support file globbing, to write
9066 multiple files at once.
9068 * We added Python bindings for the journal submission
9069 APIs. More Python APIs for a number of selected APIs will
9070 likely follow. Note that we intend to add native bindings
9071 only for the Python language, as we consider it common
9072 enough to deserve bindings shipped within systemd. There are
9073 various projects outside of systemd that provide bindings
9074 for languages such as PHP or Lua.
9076 * Many conditions will now resolve specifiers such as %i. In
9077 addition, PathChanged= and related directives of .path units
9078 now support specifiers as well.
9080 * There's now a new RPM macro definition for the system preset
9083 * journald will now warn if it ca not forward a message to the
9084 syslog daemon because its socket is full.
9086 * timedated will no longer write or process /etc/timezone,
9087 except on Debian. As we do not support late mounted /usr
9088 anymore /etc/localtime always being a symlink is now safe,
9089 and hence the information in /etc/timezone is not necessary
9092 * logind will now always reserve one VT for a text getty (VT6
9093 by default). Previously if more than 6 X sessions where
9094 started they took up all the VTs with auto-spawned gettys,
9095 so that no text gettys were available anymore.
9097 * udev will now automatically inform the btrfs kernel logic
9098 about btrfs RAID components showing up. This should make
9099 simple hotplug based btrfs RAID assembly work.
9101 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
9102 (but not for its children which will stay at the kernel
9103 default). This should allow setups with a lot more listening
9106 * systemd will now always pass the configured timezone to the
9107 kernel at boot. timedated will do the same when the timezone
9110 * logind's inhibition logic has been updated. By default,
9111 logind will now handle the lid switch, the power and sleep
9112 keys all the time, even in graphical sessions. If DEs want
9113 to handle these events on their own they should take the new
9114 handle-power-key, handle-sleep-key and handle-lid-switch
9115 inhibitors during their runtime. A simple way to achieve
9116 that is to invoke the DE wrapped in an invocation of:
9118 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
9120 * Access to unit operations is now checked via SELinux taking
9121 the unit file label and client process label into account.
9123 * systemd will now notify the administrator in the journal
9124 when he over-mounts a non-empty directory.
9126 * There are new specifiers that are resolved in unit files,
9127 for the host name (%H), the machine ID (%m) and the boot ID
9130 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
9131 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
9132 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
9133 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
9134 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
9135 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
9136 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
9140 * Support for reading structured kernel messages from
9141 /dev/kmsg has now been added and is enabled by default.
9143 * Support for reading kernel messages from /proc/kmsg has now
9144 been removed. If you want kernel messages in the journal
9145 make sure to run a recent kernel (>= 3.5) that supports
9146 reading structured messages from /dev/kmsg (see
9147 above). /proc/kmsg is now exclusive property of classic
9148 syslog daemons again.
9150 * The libudev API gained the new
9151 udev_device_new_from_device_id() call.
9153 * The logic for file system namespace (ReadOnlyDirectory=,
9154 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
9155 require pivot_root() anymore. This means fewer temporary
9156 directories are created below /tmp for this feature.
9158 * nspawn containers will now see and receive all submounts
9159 made on the host OS below the root file system of the
9162 * Forward Secure Sealing is now supported for Journal files,
9163 which provide cryptographical sealing of journal files so
9164 that attackers cannot alter log history anymore without this
9165 being detectable. Lennart will soon post a blog story about
9166 this explaining it in more detail.
9168 * There are two new service settings RestartPreventExitStatus=
9169 and SuccessExitStatus= which allow configuration of exit
9170 status (exit code or signal) which will be excepted from the
9171 restart logic, resp. consider successful.
9173 * journalctl gained the new --verify switch that can be used
9174 to check the integrity of the structure of journal files and
9175 (if Forward Secure Sealing is enabled) the contents of
9178 * nspawn containers will now be run with /dev/stdin, /dev/fd/
9179 and similar symlinks pre-created. This makes running shells
9180 as container init process a lot more fun.
9182 * The fstab support can now handle PARTUUID= and PARTLABEL=
9185 * A new ConditionHost= condition has been added to match
9186 against the hostname (with globs) and machine ID. This is
9187 useful for clusters where a single OS image is used to
9188 provision a large number of hosts which shall run slightly
9189 different sets of services.
9191 * Services which hit the restart limit will now be placed in a
9194 Contributions from: Bertram Poettering, Dave Reisner, Huang
9195 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
9196 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
9200 * When running in --user mode systemd will now become a
9201 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
9202 tree a lot more organized.
9204 * A new PartOf= unit dependency type has been introduced that
9205 may be used to group services in a natural way.
9207 * "systemctl enable" may now be used to enable instances of
9210 * journalctl now prints error log levels in red, and
9211 warning/notice log levels in bright white. It also supports
9212 filtering by log level now.
9214 * cgtop gained a new -n switch (similar to top), to configure
9215 the maximum number of iterations to run for. It also gained
9216 -b, to run in batch mode (accepting no input).
9218 * The suffix ".service" may now be omitted on most systemctl
9219 command lines involving service unit names.
9221 * There's a new bus call in logind to lock all sessions, as
9222 well as a loginctl verb for it "lock-sessions".
9224 * libsystemd-logind.so gained a new call sd_journal_perror()
9225 that works similar to libc perror() but logs to the journal
9226 and encodes structured information about the error number.
9228 * /etc/crypttab entries now understand the new keyfile-size=
9231 * shutdown(8) now can send a (configurable) wall message when
9232 a shutdown is cancelled.
9234 * The mount propagation mode for the root file system will now
9235 default to "shared", which is useful to make containers work
9236 nicely out-of-the-box so that they receive new mounts from
9237 the host. This can be undone locally by running "mount
9238 --make-rprivate /" if needed.
9240 * The prefdm.service file has been removed. Distributions
9241 should maintain this unit downstream if they intend to keep
9242 it around. However, we recommend writing normal unit files
9243 for display managers instead.
9245 * Since systemd is a crucial part of the OS we will now
9246 default to a number of compiler switches that improve
9247 security (hardening) such as read-only relocations, stack
9248 protection, and suchlike.
9250 * The TimeoutSec= setting for services is now split into
9251 TimeoutStartSec= and TimeoutStopSec= to allow configuration
9252 of individual time outs for the start and the stop phase of
9255 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
9256 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
9257 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
9258 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
9259 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
9260 Gundersen, Zbigniew Jędrzejewski-Szmek
9264 * The journal and id128 C APIs are now fully documented as man
9267 * Extra safety checks have been added when transitioning from
9268 the initial RAM disk to the main system to avoid accidental
9271 * /etc/crypttab entries now understand the new keyfile-offset=
9274 * systemctl -t can now be used to filter by unit load state.
9276 * The journal C API gained the new sd_journal_wait() call to
9277 make writing synchronous journal clients easier.
9279 * journalctl gained the new -D switch to show journals from a
9282 * journalctl now displays a special marker between log
9283 messages of two different boots.
9285 * The journal is now explicitly flushed to /var via a service
9286 systemd-journal-flush.service, rather than implicitly simply
9287 by seeing /var/log/journal to be writable.
9289 * journalctl (and the journal C APIs) can now match for much
9290 more complex expressions, with alternatives and
9293 * When transitioning from the initial RAM disk to the main
9294 system we will now kill all processes in a killing spree to
9295 ensure no processes stay around by accident.
9297 * Three new specifiers may be used in unit files: %u, %h, %s
9298 resolve to the user name, user home directory resp. user
9299 shell. This is useful for running systemd user instances.
9301 * We now automatically rotate journal files if their data
9302 object hash table gets a fill level > 75%. We also size the
9303 hash table based on the configured maximum file size. This
9304 together should lower hash collisions drastically and thus
9305 speed things up a bit.
9307 * journalctl gained the new "--header" switch to introspect
9308 header data of journal files.
9310 * A new setting SystemCallFilters= has been added to services
9311 which may be used to apply blacklists or whitelists to
9312 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
9314 * nspawn gained a new --link-journal= switch (and quicker: -j)
9315 to link the container journal with the host. This makes it
9316 very easy to centralize log viewing on the host for all
9317 guests while still keeping the journal files separated.
9319 * Many bugfixes and optimizations
9321 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
9322 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
9323 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
9328 * Several tools now understand kernel command line arguments,
9329 which are only read when run in an initial RAM disk. They
9330 usually follow closely their normal counterparts, but are
9333 * There's a new tool to analyze the readahead files that are
9334 automatically generated at boot. Use:
9336 /usr/lib/systemd/systemd-readahead analyze /.readahead
9338 * We now provide an early debug shell on tty9 if this enabled. Use:
9340 systemctl enable debug-shell.service
9342 * All plymouth related units have been moved into the Plymouth
9343 package. Please make sure to upgrade your Plymouth version
9346 * systemd-tmpfiles now supports getting passed the basename of
9347 a configuration file only, in which case it will look for it
9348 in all appropriate directories automatically.
9350 * udevadm info now takes a /dev or /sys path as argument, and
9351 does the right thing. Example:
9353 udevadm info /dev/sda
9354 udevadm info /sys/class/block/sda
9356 * systemctl now prints a warning if a unit is stopped but a
9357 unit that might trigger it continues to run. Example: a
9358 service is stopped but the socket that activates it is left
9361 * "systemctl status" will now mention if the log output was
9362 shortened due to rotation since a service has been started.
9364 * The journal API now exposes functions to determine the
9365 "cutoff" times due to rotation.
9367 * journald now understands SIGUSR1 and SIGUSR2 for triggering
9368 immediately flushing of runtime logs to /var if possible,
9369 resp. for triggering immediate rotation of the journal
9372 * It is now considered an error if a service is attempted to
9373 be stopped that is not loaded.
9375 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
9377 * systemd-analyze now supports Python 3
9379 * tmpfiles now supports cleaning up directories via aging
9380 where the first level dirs are always kept around but
9381 directories beneath it automatically aged. This is enabled
9382 by prefixing the age field with '~'.
9384 * Seat objects now expose CanGraphical, CanTTY properties
9385 which is required to deal with very fast bootups where the
9386 display manager might be running before the graphics drivers
9387 completed initialization.
9389 * Seat objects now expose a State property.
9391 * We now include RPM macros for service enabling/disabling
9392 based on the preset logic. We recommend RPM based
9393 distributions to make use of these macros if possible. This
9394 makes it simpler to reuse RPM spec files across
9397 * We now make sure that the collected systemd unit name is
9398 always valid when services log to the journal via
9401 * There's a new man page kernel-command-line(7) detailing all
9402 command line options we understand.
9404 * The fstab generator may now be disabled at boot by passing
9405 fstab=0 on the kernel command line.
9407 * A new kernel command line option modules-load= is now understood
9408 to load a specific kernel module statically, early at boot.
9410 * Unit names specified on the systemctl command line are now
9411 automatically escaped as needed. Also, if file system or
9412 device paths are specified they are automatically turned
9413 into the appropriate mount or device unit names. Example:
9415 systemctl status /home
9416 systemctl status /dev/sda
9418 * The SysVConsole= configuration option has been removed from
9419 system.conf parsing.
9421 * The SysV search path is no longer exported on the D-Bus
9424 * The Names= option has been removed from unit file parsing.
9426 * There's a new man page bootup(7) detailing the boot process.
9428 * Every unit and every generator we ship with systemd now
9429 comes with full documentation. The self-explanatory boot is
9432 * A couple of services gained "systemd-" prefixes in their
9433 name if they wrap systemd code, rather than only external
9434 code. Among them fsck@.service which is now
9435 systemd-fsck@.service.
9437 * The HaveWatchdog property has been removed from the D-Bus
9440 * systemd.confirm_spawn= on the kernel command line should now
9443 * There's a new man page crypttab(5) which details all options
9444 we actually understand.
9446 * systemd-nspawn gained a new --capability= switch to pass
9447 additional capabilities to the container.
9449 * timedated will now read known NTP implementation unit names
9450 from /usr/lib/systemd/ntp-units.d/*.list,
9451 systemd-timedated-ntp.target has been removed.
9453 * journalctl gained a new switch "-b" that lists log data of
9454 the current boot only.
9456 * The notify socket is in the abstract namespace again, in
9457 order to support daemons which chroot() at start-up.
9459 * There is a new Storage= configuration option for journald
9460 which allows configuration of where log data should go. This
9461 also provides a way to disable journal logging entirely, so
9462 that data collected is only forwarded to the console, the
9463 kernel log buffer or another syslog implementation.
9465 * Many bugfixes and optimizations
9467 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
9468 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
9469 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
9470 Shawn Landden, Tom Gundersen
9474 * "systemctl help <unit>" now shows the man page if one is
9477 * Several new man pages have been added.
9479 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
9480 MaxLevelConsole= can now be specified in
9481 journald.conf. These options allow reducing the amount of
9482 data stored on disk or forwarded by the log level.
9484 * TimerSlackNSec= can now be specified in system.conf for
9485 PID1. This allows system-wide power savings.
9487 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
9488 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
9493 * logind is now capable of (optionally) handling power and
9494 sleep keys as well as the lid switch.
9496 * journalctl now understands the syntax "journalctl
9497 /usr/bin/avahi-daemon" to get all log output of a specific
9500 * CapabilityBoundingSet= in system.conf now also influences
9501 the capability bound set of usermode helpers of the kernel.
9503 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
9504 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
9505 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
9506 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
9510 * Note that we skipped 139 releases here in order to set the
9511 new version to something that is greater than both udev's
9512 and systemd's most recent version number.
9514 * udev: all udev sources are merged into the systemd source tree now.
9515 All future udev development will happen in the systemd tree. It
9516 is still fully supported to use the udev daemon and tools without
9517 systemd running, like in initramfs or other init systems. Building
9518 udev though, will require the *build* of the systemd tree, but
9519 udev can be properly *run* without systemd.
9521 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
9522 should be used to create dead device nodes as workarounds for broken
9525 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
9526 no longer supported. udev_monitor_new_from_netlink() needs to be
9527 used to subscribe to events.
9529 * udev: when udevd is started by systemd, processes which are left
9530 behind by forking them off of udev rules, are unconditionally cleaned
9531 up and killed now after the event handling has finished. Services or
9532 daemons must be started as systemd services. Services can be
9533 pulled-in by udev to get started, but they can no longer be directly
9534 forked by udev rules.
9536 * udev: the daemon binary is called systemd-udevd now and installed
9537 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
9538 to adapt to that, create symlink, or rename the binary after building
9541 * libudev no longer provides these symbols:
9542 udev_monitor_from_socket()
9543 udev_queue_get_failed_list_entry()
9544 udev_get_{dev,sys,run}_path()
9545 The versions number was bumped and symbol versioning introduced.
9547 * systemd-loginctl and systemd-journalctl have been renamed
9548 to loginctl and journalctl to match systemctl.
9550 * The config files: /etc/systemd/systemd-logind.conf and
9551 /etc/systemd/systemd-journald.conf have been renamed to
9552 logind.conf and journald.conf. Package updates should rename
9553 the files to the new names on upgrade.
9555 * For almost all files the license is now LGPL2.1+, changed
9556 from the previous GPL2.0+. Exceptions are some minor stuff
9557 of udev (which will be changed to LGPL2.1 eventually, too),
9558 and the MIT licensed sd-daemon.[ch] library that is suitable
9559 to be used as drop-in files.
9561 * systemd and logind now handle system sleep states, in
9562 particular suspending and hibernating.
9564 * logind now implements a sleep/shutdown/idle inhibiting logic
9565 suitable for a variety of uses. Soonishly Lennart will blog
9566 about this in more detail.
9568 * var-run.mount and var-lock.mount are no longer provided
9569 (which previously bind mounted these directories to their new
9570 places). Distributions which have not converted these
9571 directories to symlinks should consider stealing these files
9572 from git history and add them downstream.
9574 * We introduced the Documentation= field for units and added
9575 this to all our shipped units. This is useful to make it
9576 easier to explore the boot and the purpose of the various
9579 * All smaller setup units (such as
9580 systemd-vconsole-setup.service) now detect properly if they
9581 are run in a container and are skipped when
9582 appropriate. This guarantees an entirely noise-free boot in
9583 Linux container environments such as systemd-nspawn.
9585 * A framework for implementing offline system updates is now
9586 integrated, for details see:
9587 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
9589 * A new service type Type=idle is available now which helps us
9590 avoiding ugly interleaving of getty output and boot status
9593 * There's now a system-wide CapabilityBoundingSet= option to
9594 globally reduce the set of capabilities for the
9595 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
9596 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
9597 even CAP_NET_ADMIN system-wide for secure systems.
9599 * There are now system-wide DefaultLimitXXX= options to
9600 globally change the defaults of the various resource limits
9601 for all units started by PID 1.
9603 * Harald Hoyer's systemd test suite has been integrated into
9604 systemd which allows easy testing of systemd builds in qemu
9605 and nspawn. (This is really awesome! Ask us for details!)
9607 * The fstab parser is now implemented as generator, not inside
9610 * systemctl will now warn you if .mount units generated from
9611 /etc/fstab are out of date due to changes in fstab that
9612 have not been read by systemd yet.
9614 * systemd is now suitable for usage in initrds. Dracut has
9615 already been updated to make use of this. With this in place
9616 initrds get a slight bit faster but primarily are much
9617 easier to introspect and debug since "systemctl status" in
9618 the host system can be used to introspect initrd services,
9619 and the journal from the initrd is kept around too.
9621 * systemd-delta has been added, a tool to explore differences
9622 between user/admin configuration and vendor defaults.
9624 * PrivateTmp= now affects both /tmp and /var/tmp.
9626 * Boot time status messages are now much prettier and feature
9627 proper english language. Booting up systemd has never been
9630 * Read-ahead pack files now include the inode number of all
9631 files to pre-cache. When the inode changes the pre-caching
9632 is not attempted. This should be nicer to deal with updated
9633 packages which might result in changes of read-ahead
9636 * We now temporaritly lower the kernel's read_ahead_kb variable
9637 when collecting read-ahead data to ensure the kernel's
9638 built-in read-ahead does not add noise to our measurements
9639 of necessary blocks to pre-cache.
9641 * There's now RequiresMountsFor= to add automatic dependencies
9642 for all mounts necessary for a specific file system path.
9644 * MountAuto= and SwapAuto= have been removed from
9645 system.conf. Mounting file systems at boot has to take place
9648 * nspawn now learned a new switch --uuid= to set the machine
9649 ID on the command line.
9651 * nspawn now learned the -b switch to automatically search
9654 * vt102 is now the default TERM for serial TTYs, upgraded from
9657 * systemd-logind now works on VT-less systems.
9659 * The build tree has been reorganized. The individual
9660 components now have directories of their own.
9662 * A new condition type ConditionPathIsReadWrite= is now available.
9664 * nspawn learned the new -C switch to create cgroups for the
9665 container in other hierarchies.
9667 * We now have support for hardware watchdogs, configurable in
9670 * The scheduled shutdown logic now has a public API.
9672 * We now mount /tmp as tmpfs by default, but this can be
9673 masked and /etc/fstab can override it.
9675 * Since udisks does not make use of /media anymore we are not
9676 mounting a tmpfs on it anymore.
9678 * journalctl gained a new --local switch to only interleave
9679 locally generated journal files.
9681 * We can now load the IMA policy at boot automatically.
9683 * The GTK tools have been split off into a systemd-ui.
9685 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
9686 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
9687 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
9688 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
9689 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
9690 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
9691 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
9692 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
9693 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
9698 * This is mostly a bugfix release
9700 * Support optional initialization of the machine ID from the
9701 KVM or container configured UUID.
9703 * Support immediate reboots with "systemctl reboot -ff"
9705 * Show /etc/os-release data in systemd-analyze output
9707 * Many bugfixes for the journal, including endianness fixes and
9708 ensuring that disk space enforcement works
9710 * sd-login.h is C++ compatible again
9712 * Extend the /etc/os-release format on request of the Debian
9715 * We now refuse non-UTF8 strings used in various configuration
9716 and unit files. This is done to ensure we do not pass invalid
9717 data over D-Bus or expose it elsewhere.
9719 * Register Mimo USB Screens as suitable for automatic seat
9722 * Read SELinux client context from journal clients in a race
9725 * Reorder configuration file lookup order. /etc now always
9726 overrides /run in order to allow the administrator to always
9727 and unconditionally override vendor-supplied or
9728 automatically generated data.
9730 * The various user visible bits of the journal now have man
9731 pages. We still lack man pages for the journal API calls
9734 * We now ship all man pages in HTML format again in the
9737 Contributions from: Dave Reisner, Dirk Eibach, Frederic
9738 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
9739 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
9744 * This is mostly a bugfix release
9746 * systems lacking /etc/os-release are no longer supported.
9748 * Various functionality updates to libsystemd-login.so
9750 * Track class of PAM logins to distinguish greeters from
9753 Contributions from: Kay Sievers, Lennart Poettering, Michael
9758 * This is an important bugfix release for v41.
9760 * Building man pages is now optional which should be useful
9761 for those building systemd from git but unwilling to install
9764 * Watchdog support for supervising services is now usable. In
9765 a future release support for hardware watchdogs
9766 (i.e. /dev/watchdog) will be added building on this.
9768 * Service start rate limiting is now configurable and can be
9769 turned off per service. When a start rate limit is hit a
9770 reboot can automatically be triggered.
9772 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
9774 Contributions from: Benjamin Franzke, Bill Nottingham,
9775 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
9776 Schmidt, Michał Górny, Piotr Drąg
9780 * The systemd binary is installed /usr/lib/systemd/systemd now;
9781 An existing /sbin/init symlink needs to be adapted with the
9784 * The code that loads kernel modules has been ported to invoke
9785 libkmod directly, instead of modprobe. This means we do not
9786 support systems with module-init-tools anymore.
9788 * Watchdog support is now already useful, but still not
9791 * A new kernel command line option systemd.setenv= is
9792 understood to set system wide environment variables
9793 dynamically at boot.
9795 * We now limit the set of capabilities of systemd-journald.
9797 * We now set SIGPIPE to ignore by default, since it only is
9798 useful in shell pipelines, and has little use in general
9799 code. This can be disabled with IgnoreSIPIPE=no in unit
9802 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
9803 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
9808 * This is mostly a bugfix release
9810 * We now expose the reason why a service failed in the
9811 "Result" D-Bus property.
9813 * Rudimentary service watchdog support (will be completed over
9814 the next few releases.)
9816 * When systemd forks off in order execute some service we will
9817 now immediately changes its argv[0] to reflect which process
9818 it will execute. This is useful to minimize the time window
9819 with a generic argv[0], which makes bootcharts more useful
9821 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
9822 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
9823 Mike Kazantsev, Ray Strode
9827 * This is mostly a test release, but incorporates many
9830 * New systemd-cgtop tool to show control groups by their
9833 * Linking against libacl for ACLs is optional again. If
9834 disabled, support tracking device access for active logins
9835 goes becomes unavailable, and so does access to the user
9836 journals by the respective users.
9838 * If a group "adm" exists, journal files are automatically
9839 owned by them, thus allow members of this group full access
9840 to the system journal as well as all user journals.
9842 * The journal now stores the SELinux context of the logging
9843 client for all entries.
9845 * Add C++ inclusion guards to all public headers
9847 * New output mode "cat" in the journal to print only text
9848 messages, without any meta data like date or time.
9850 * Include tiny X server wrapper as a temporary stop-gap to
9851 teach XOrg udev display enumeration. This is used by display
9852 managers such as gdm, and will go away as soon as XOrg
9853 learned native udev hotplugging for display devices.
9855 * Add new systemd-cat tool for executing arbitrary programs
9856 with STDERR/STDOUT connected to the journal. Can also act as
9857 BSD logger replacement, and does so by default.
9859 * Optionally store all locally generated coredumps in the
9860 journal along with meta data.
9862 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
9863 writing short strings to files (for usage for /sys), and for
9864 creating symlinks, character and block device nodes.
9866 * New unit file option ControlGroupPersistent= to make cgroups
9867 persistent, following the mechanisms outlined in
9868 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
9870 * Support multiple local RTCs in a sane way
9872 * No longer monopolize IO when replaying readahead data on
9873 rotating disks, since we might starve non-file-system IO to
9874 death, since fanotify() will not see accesses done by blkid,
9877 * Do not show kernel threads in systemd-cgls anymore, unless
9878 requested with new -k switch.
9880 Contributions from: Dan Horák, Kay Sievers, Lennart
9881 Poettering, Michal Schmidt
9885 * This is mostly a test release, but incorporates many
9888 * The git repository moved to:
9889 git://anongit.freedesktop.org/systemd/systemd
9890 ssh://git.freedesktop.org/git/systemd/systemd
9892 * First release with the journal
9893 http://0pointer.de/blog/projects/the-journal.html
9895 * The journal replaces both systemd-kmsg-syslogd and
9896 systemd-stdout-bridge.
9898 * New sd_pid_get_unit() API call in libsystemd-logind
9900 * Many systemadm clean-ups
9902 * Introduce remote-fs-pre.target which is ordered before all
9903 remote mounts and may be used to start services before all
9906 * Added Mageia support
9908 * Add bash completion for systemd-loginctl
9910 * Actively monitor PID file creation for daemons which exit in
9911 the parent process before having finished writing the PID
9912 file in the daemon process. Daemons which do this need to be
9913 fixed (i.e. PID file creation must have finished before the
9914 parent exits), but we now react a bit more gracefully to them.
9916 * Add colourful boot output, mimicking the well-known output
9917 of existing distributions.
9919 * New option PassCredentials= for socket units, for
9920 compatibility with a recent kernel ABI breakage.
9922 * /etc/rc.local is now hooked in via a generator binary, and
9923 thus will no longer act as synchronization point during
9926 * systemctl list-unit-files now supports --root=.
9928 * systemd-tmpfiles now understands two new commands: z, Z for
9929 relabelling files according to the SELinux database. This is
9930 useful to apply SELinux labels to specific files in /sys,
9933 * Output of SysV services is now forwarded to both the console
9934 and the journal by default, not only just the console.
9936 * New man pages for all APIs from libsystemd-login.
9938 * The build tree got reorganized and the build system is a
9939 lot more modular allowing embedded setups to specifically
9940 select the components of systemd they are interested in.
9942 * Support for Linux systems lacking the kernel VT subsystem is
9945 * configure's --with-rootdir= got renamed to
9946 --with-rootprefix= to follow the naming used by udev and
9949 * Unless specified otherwise we will now install to /usr instead
9950 of /usr/local by default.
9952 * Processes with '@' in argv[0][0] are now excluded from the
9953 final shut-down killing spree, following the logic explained
9955 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
9957 * All processes remaining in a service cgroup when we enter
9958 the START or START_PRE states are now killed with
9959 SIGKILL. That means it is no longer possible to spawn
9960 background processes from ExecStart= lines (which was never
9961 supported anyway, and bad style).
9963 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
9964 reloading of units together.
9966 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
9967 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
9968 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
9969 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
9970 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
projects / platform / upstream / systemd.git / blob