1 systemd System and Service Manager
3 CHANGES WITH 243 in spe:
5 * Previously, filters defined with SystemCallFilter= would have the
6 effect that an calling an offending system call would terminate the
7 calling thread. This behaviour never made much sense, since killing
8 individual threads of unexpecting processes is likely to create more
9 problems than it solves. With this release the default action changed
10 from killing the thread to killing the whole process. For this to
11 work correctly both a kernel version (>= 4.14) and a libseccomp
12 version (>= 2.4.0) supporting this new seccomp action is required. If
13 an older kernel or libseccomp is used the old behaviour continues to
14 be used. This change does not affect any services that have no system
15 call filters defined, or that use SystemCallErrorNumber= (and thus
16 see EPERM or another error instead of being killed when calling an
17 offending system call). Note that systemd documentation always
18 claimed that the whole process is killed. With this change behaviour
19 is thus adjusted to match the documentation.
21 * The "kernel.pid_max" sysctl is now bumped to 4194304 by default,
22 i.e. the full 22bit range the kernel allows, up from the old 16bit
23 range. This should improve security and robustness a bit, as PID
24 collisions are made less likely (though certainly still
25 possible). There are rumours this might create compatibility
26 problems, though at this moment no practical ones are known to
27 us. Downstream distributions are hence advised to undo this change in
28 their builds if they are concerned about maximum compatibility, but
29 for everybody else we recommend leaving the value bumped. Besides
30 improving security and robustness this should also simplify things as
31 the maximum number of allowed concurrent tasks was previously bounded
32 by both "kernel.pid_max" and "kernel.threads-max" and now only a
33 single knob is left ("kernel.threads-max"). There have been concerns
34 that usability is affected by this change because larger PID numbers
35 are harder to type, but we believe the change from 5 digit PIDs to 7
36 digit PIDs is not too hampering for usability.
38 * MemoryLow= and MemoryMin= gained hierarchy-aware counterparts,
39 DefaultMemoryLow= and DefaultMemoryMin=, which can be used to
40 hierarchically set default memory protection values for a particular
41 subtree of the unit hierarchy.
43 * Memory protection directives can now take a value of zero, allowing
44 explicit opting out of a default value propagated by an ancestor.
46 * A new setting DisableControllers= has been added that may be used to
47 explicitly disable one or more cgroups controllers for a unit and all
50 * systemd now defaults to the "unified" cgroup hierarchy setup during
51 build-time, i.e. -Ddefault-hierarchy=unified is now the build-time
52 default. Previously, -Ddefault-hierarchy=hybrid was the default. This
53 change reflects the fact that cgroupsv2 support has matured
54 substantially in both systemd and in the kernel, and is clearly the
55 way forward. Downstream production distributions might want to
56 continue to use -Ddefault-hierarchy=hybrid (or even =legacy) for
57 their builds as unfortunately the popular container managers have not
58 caught up with the kernel API changes.
60 * Man pages are not built by default anymore (html pages were already
61 disabled by default), to make development builds quicker. When
62 building systemd for a full installation with documentation, meson
63 should be called -Dman=true and/or -Dhtml=true as appropriate. The
64 default was changed based on the assumption that quick one-off or
65 repeated development builds are much more common than full optimized
66 builds for installation, and people need to pass various other
67 options to when doing "proper" builds anyway, so the gain from making
68 development builds quicker is bigger than the one time disruption for
71 Two scripts are created in the *build* directory to generate and
72 preview man and html pages on demand, e.g.:
74 build/man/man systemctl
75 build/man/html systemd.index
77 * libidn2 is used by default if both libidn2 and libidn are installed.
78 Please use -Dlibidn=true when libidn is favorable.
80 * The D-Bus "wire format" of the CPUAffinity= attribute is changed on
81 big-endian machines. Before, bytes were written and read in native
82 machine order as exposed by the native libc __cpu_mask interface.
83 Now, little-endian order is always used (CPUs 0–7 are described by
84 bits 0–7 in byte 0, CPUs 8–15 are described by byte 1, and so on).
85 This change fixes D-Bus calls that cross endianness boundary.
87 The presentation format used for CPUAffinity= by "systemctl show" and
88 "systemd-analyze dump" is changed to present CPU indices instead of
89 the raw __cpu_mask bitmask. For example, CPUAffinity=0-1 would be
90 shown as CPUAffinity=03000000000000000000000000000… (on
91 little-endian) or CPUAffinity=00000000000000300000000000000… (on
92 64-bit big-endian), and is now shown as CPUAffinity=0-1, matching the
93 input format. The maximum integer that will be printed in the new
94 format is 8191 (four digits), while the old format always used a very
95 long number (with the length varying by architecture), so they can be
96 unambiguously distinguished.
98 * /usr/sbin/halt.local is no longer supported. Implementation in
99 distributions was inconsistent and it seems this functionality was
102 To replace this functionality, users should:
103 - either define a new unit and make it a dependency of final.target
104 (systemctl add-wants final.target my-halt-local.service)
105 - or move the shutdown script to /usr/lib/systemd/system-shutdown/
106 and ensure that it accepts "halt", "poweroff", "reboot", and
107 "kexec" as an argument, see the description in systemd-shutdown(8).
109 * When a [Match] section in .link or .network file is empty (contains
110 no match patterns), a warning will be emitted. Please add any "match
111 all" pattern instead, e.g. OriginalName=* or Name=* in case all
112 interfaces should really be matched.
114 * A new setting NUMAPolicy= may be used to set process memory
115 allocation policy. Setting can be specified in system.conf and
116 hence will set the default policy for PID1. Default policy can be
117 overridden on per-service basis. Related setting NUMAMask= is used to
118 specify NUMA node mask that should be associated with the selected
121 * PID 1 will now listen to Out-Of-Memory (OOM) events the kernel
122 generates when processes it manages are reaching their memory limits,
123 and will place their units in a special state, and optionally kill or
126 * The service manager will now expose bus properties for the IO
127 resources used by units. This information is also shown in "systemctl
128 status" now (for services that have IOAccounting=yes set). Moreover,
129 the IO accounting data is included in the resource log message
130 generated whenever a unit stops.
132 * units may now configure an explicit time-out to apply to when killed
133 with SIGABRT, for example when a service watchdog is hit. Previously,
134 the regular TimeoutStopSec= time-out was applied in this case too —
135 now a separate time-out may be set using TimeoutAbortSec=.
137 * Services may now send a special WATCHDOG=trigger message with
138 sd_notify() to trigger an immediate "watchdog missed" event, and thus
139 request service take down. This is useful both for testing watchdog
140 handling, but also for defining error paths in services, that shall
141 be handled the same way as watchdog events.
143 * There are two new per-unit settings IPIngressFilterPath= and
144 IPEgressFilterPath= which allow configuration of a BPF program
145 (usually by specifying a path to a program uploaded to /sys/fs/bpf/)
146 to apply to the IP packet ingress/egress path of all processes of a
147 unit. This is useful to allow running systemd services with BPF
148 programs set up externally.
150 * systemctl gained a new "clean" verb for removing the state, cache,
151 runtime or logs directories of a service while it is terminated. The
152 new verb may also be used to remove the state maintained on disk for
153 timer units that have Persistent= configured.
155 * During the last phase of shutdown systemd will now automatically
156 increase the log level configured in the "kernel.printk" sysctl so
157 that any relevant loggable events happening during late shutdown are
158 made visible. Previously, loggable events happening so late during
159 shutdown were generally lost if the "kernel.printk" sysctl was set to
160 high thresholds, as regular logging daemons are terminated at that
161 time and thus nothing is written to disk.
163 * If processes terminated during the last phase of shutdown do not exit
164 quickly systemd will now show their names after a short time, to make
165 debugging easier. After a longer time-out they are forcibly killed,
168 * journalctl (and the other tools that display logs) will now highlight
169 warnings in yellow (previously, both LOG_NOTICE and LOG_WARNING where
170 shown in bright bold, now only LOG_NOTICE is). Moreover, audit logs
171 are now shown in blue color, to separate them visually from regular
172 logs. References to configuration files are now turned into clickable
173 links on terminals that support that.
175 * systemd-journald will now stop logging to /var/log/journal during
176 shutdown when /var/ is on a separate mount, so that it can be
177 unmounted safely during shutdown.
179 * systemd-resolved gained support for a new 'strict' DNS-over-TLS mode.
181 * The predictable naming scheme for network devices now supports
182 generating predictable names for "netdevsim" devices.
184 * systemd-networkd now supports MACsec, nlmon, IPVTAP and Xfrm
187 * systemd-networkd's bridge FDB support now allows configuration of a
188 destination address for each entry (Destination=), as well as the
189 VXLAN VNI (VNI=), as well as an option to declare what an entry is
190 associated with (AssociatedWith=).
192 * systemd-networkd's DHCPv4 support now understands a new MaxAttempts=
193 option for configuring the maximum number of attempts to request a
194 DHCP lease. It also learnt a new BlackList= option for blacklisting
195 DHCP servers (a similar setting has also been added to the IPv6 RA
196 client), as well as a SendRelease= option for configuring whether to
197 send a DHCP RELEASE message when terminating.
199 * systemd-networkd's DHCPv4 and DHCPv6 stacks can now be configured
200 seperately in the [DHCPv4] and [DHCPv6] sections.
202 * systemd-networkd's VXLAN support gained a new option
203 GenericProtocolExtension= for enabling VXLAN Generic Protocol
204 Extension support, as well as IPDoNotFragment= for setting the IP
205 "Don't fragment" bit on outgoing packets. A similar option has been
206 added to the GENEVE support.
208 * In systemd-networkd's [Route] section you may now configure
209 FastOpenNoCookie= for configuring per-route TCP fast-open support, as
210 well as TTLPropagate= for configuring Label Switched Path (LSP) TTL
211 propagation. The Type= setting now supports local, broadcast,
212 anycast, multicast, any, xresolve routes, too.
214 * systemd-networkd's [Network] section learnt a new option
215 DefaultRouteOnDevice= for automatically configuring a default route
216 onto the network device.
218 * systemd-networkd's bridging support gained two new options ProxyARP=
219 and ProxyARPWifi= for configuring proxy ARP behaviour as well as
220 MulticastRouter= for configureing multicast routing behaviour.
222 * systemd-networkd's FooOverUDP support gained the ability to configure
223 local and peer IP addresses via Local= and Peer=. A new option
224 PeerPort= may be used to configure the peer's IP port.
226 * systemd-networkd's TUN support gained a new setting VnetHeader= for
227 tweaking Generic Segment Offload support.
229 * networkctl gained a new "delete" command for removing virtual network
230 devices, as well as a new "--stats" switch for showing device
233 * systemd-networkd's .network and .link files gained a new Property=
234 setting in the [Match] section, to match against devices with
235 specific udev properties.
237 * systemd-networkd's tunnel support gained a new option
238 AssignToLoopback= for selecting whether to use the loopback device
239 "lo" as underlying device.
241 * systemd-networkd's MACAddress= setting in the [Neighbor] section has
242 been renamed to LinkLayerAddress=, and it now allows configuration of
245 * The CriticalConnection= setting in .network files is now deprecated,
246 and replaced by a new KeepConfiguration= setting which allows more
247 detailed configuration of the IP configuration to keep in place.
249 * systemd-analyze gained a new "timestamp" verb for parsing and
250 converting timestamps. It's similar to the existing "systemd-analyze
251 calendar" command which does the same for recurring calendar
252 events. It also gained a new "condition" verb for parsing and testing
253 ConditionXYZ= expressions.
255 * systemd-logind now exposes a per-session SetBrightness() bus call,
256 which may be used to securely change the brightness of a kernel
257 brightness device, if it belongs to the session's seat. By using this
258 call unprivileged clients can make changes to "backlight" and "leds"
259 devices securely with strict requirements on session
260 membership. Desktop environments may use this to generically make
261 brightness changes to such devices without shipping private SUID
262 binaries or specific udev rules for that purpose.
264 * "udevadm info" gained a --wait-for-initialization switch to wait for
265 a device to be initialized.
267 * systemd-hibernate-resume-generator will now look for resumeflags= on
268 the kernel command line, which is similar to rootflags= and may be
269 used to configure device timeouts for waiting for the hibernation
272 * sd-event learnt a new API call sd_event_source_disable_unref() for
273 disabling and unref'ing an event source in a single function. A
274 related call sd_event_source_disable_unrefp() has been added for use
275 with GCC's cleanup extension.
277 * The sd-id128.h public API gained a new definition
278 SD_ID128_UUID_FORMAT_STR for formatting a 128bit ID in UUID format
281 * "busctl introspect" gained a new switch --xml-interface for dumping
282 XML introspection data unmodified.
284 * PID 1 may now show the unit name instead of the unit description
285 string in its status output during boot. This may be configured in
286 the StatusUnitFormat= setting in /etc/systemd/system.conf or the
287 kernel command line option systemd.status_unit_format=.
289 * The systemd.debug_shell kernel command line option now optionally
290 takes a tty name to spawn the debug shell on, which allows selecting
291 a different tty than the built-in default.
297 * In .link files, MACAddressPolicy=persistent (the default) is changed
298 to cover more devices. For devices like bridges, tun, tap, bond, and
299 similar interfaces that do not have other identifying information,
300 the interface name is used as the basis for persistent seed for MAC
301 and IPv4LL addresses. The way that devices that were handled
302 previously is not changed, and this change is about covering more
303 devices then previously by the "persistent" policy.
305 MACAddressPolicy=random may be used to force randomized MACs and
306 IPv4LL addresses for a device if desired.
308 Hint: the log output from udev (at debug level) was enhanced to
309 clarify what policy is followed and which attributes are used.
310 `SYSTEMD_LOG_LEVEL=debug udevadm test-builtin net_setup_link /sys/class/net/<name>`
311 may be used to view this.
313 Hint: if a bridge interface is created without any slaves, and gains
314 a slave later, then now the bridge does not inherit slave's MAC.
315 To inherit slave's MAC, for example, create the following file:
317 # /etc/systemd/network/98-bridge-inherit-mac.link
322 MACAddressPolicy=none
325 * The .device units generated by systemd-fstab-generator and other
326 generators do not automatically pull in the corresponding .mount unit
327 as a Wants= dependency. This means that simply plugging in the device
328 will not cause the mount unit to be started automatically. But please
329 note that the mount unit may be started for other reasons, in
330 particular if it is part of local-fs.target, and any unit which
331 (transitively) depends on local-fs.target is started.
333 * networkctl list/status/lldp now accept globbing wildcards for network
334 interface names to match against all existing interfaces.
336 * The $PIDFILE environment variable is set to point the absolute path
337 configured with PIDFile= for processes of that service.
339 * The fallback DNS server list was augmented with Cloudflare public DNS
340 servers. Use `-Ddns-servers=` to set a different fallback.
342 * A new special target usb-gadget.target will be started automatically
343 when a USB Device Controller is detected (which means that the system
344 is a USB peripheral).
346 * A new unit setting CPUQuotaPeriodSec= assigns the time period
347 relatively to which the CPU time quota specified by CPUQuota= is
350 * A new unit setting ProtectHostname= may be used to prevent services
351 from modifying hostname information (even if they otherwise would
352 have privileges to do so).
354 * A new unit setting NetworkNamespacePath= may be used to specify a
355 namespace for service or socket units through a path referring to a
356 Linux network namespace pseudo-file.
358 * The PrivateNetwork= setting and JoinsNamespaceOf= dependencies now
359 have an effect on .socket units: when used the listening socket is
360 created within the configured network namespace instead of the host
363 * ExecStart= command lines in unit files may now be prefixed with ':'
364 in which case environment variable substitution is
365 disabled. (Supported for the other ExecXYZ= settings, too.)
367 * .timer units gained two new boolean settings OnClockChange= and
368 OnTimezoneChange= which may be used to also trigger a unit when the
369 system clock is changed or the local timezone is
370 modified. systemd-run has been updated to make these options easily
371 accessible from the command line for transient timers.
373 * Two new conditions for units have been added: ConditionMemory= may be
374 used to conditionalize a unit based on installed system
375 RAM. ConditionCPUs= may be used to conditionalize a unit based on
378 * The @default system call filter group understood by SystemCallFilter=
379 has been updated to include the new rseq() system call introduced in
382 * A new time-set.target has been added that indicates that the system
383 time has been set from a local source (possibly imprecise). The
384 existing time-sync.target is stronger and indicates that the time has
385 been synchronized with a precise external source. Services where
386 approximate time is sufficient should use the new target.
388 * "systemctl start" (and related commands) learnt a new
389 --show-transaction option. If specified brief information about all
390 jobs queued because of the requested operation is shown.
392 * systemd-networkd recognizes a new operation state 'enslaved', used
393 (instead of 'degraded' or 'carrier') for interfaces which form a
394 bridge, bond, or similar, and an new 'degraded-carrier' operational
395 state used for the bond or bridge master interface when one of the
396 enslaved devices is not operational.
398 * .network files learnt the new IgnoreCarrierLoss= option for leaving
399 networks configured even if the carrier is lost.
401 * The RequiredForOnline= setting in .network files may now specify a
402 minimum operational state required for the interface to be considered
403 "online" by systemd-networkd-wait-online. Related to this
404 systemd-networkd-wait-online gained a new option --operational-state=
405 to configure the same, and its --interface= option was updated to
406 optionally also take an operational state specific for an interface.
408 * systemd-networkd-wait-online gained a new setting --any for waiting
409 for only one of the requested interfaces instead of all of them.
411 * systemd-networkd now implements L2TP tunnels.
413 * Two new .network settings UseAutonomousPrefix= and UseOnLinkPrefix=
414 may be used to cause autonomous and onlink prefixes received in IPv6
415 Router Advertisements to be ignored.
417 * New MulticastFlood=, NeighborSuppression=, and Learning= .network
418 file settings may be used to tweak bridge behaviour.
420 * The new TripleSampling= option in .network files may be used to
421 configure CAN triple sampling.
423 * A new .netdev settings PrivateKeyFile= and PresharedKeyFile= may be
424 used to point to private or preshared key for a WireGuard interface.
426 * /etc/crypttab now supports the same-cpu-crypt and
427 submit-from-crypt-cpus options to tweak encryption work scheduling
430 * systemd-tmpfiles will now take a BSD file lock before operating on a
431 contents of directory. This may be used to temporarily exclude
432 directories from aging by taking the same lock (useful for example
433 when extracting a tarball into /tmp or /var/tmp as a privileged user,
434 which might create files with really old timestamps, which
435 nevertheless should not be deleted). For further details, see:
437 https://systemd.io/TEMPORARY_DIRECTORIES
439 * systemd-tmpfiles' h line type gained support for the
440 FS_PROJINHERIT_FL ('P') file attribute (introduced in kernel 4.5),
441 controlling project quota inheritance.
443 * sd-boot and bootctl now implement support for an Extended Boot Loader
444 (XBOOTLDR) partition, that is intended to be mounted to /boot, in
445 addition to the ESP partition mounted to /efi or /boot/efi.
446 Configuration file fragments, kernels, initrds and other EFI images
447 to boot will be loaded from both the ESP and XBOOTLDR partitions.
448 The XBOOTLDR partition was previously described by the Boot Loader
449 Specification, but implementation was missing in sd-boot. Support for
450 this concept allows using the sd-boot boot loader in more
451 conservative scenarios where the boot loader itself is placed in the
452 ESP but the kernels to boot (and their metadata) in a separate
455 * A system may now be booted with systemd.volatile=overlay on the
456 kernel command line, which causes the root file system to be set up
457 an overlayfs mount combining the root-only root directory with a
458 writable tmpfs. In this setup, the underlying root device is not
459 modified, and any changes are lost at reboot.
461 * Similar, systemd-nspawn can now boot containers with a volatile
462 overlayfs root with the new --volatile=overlay switch.
464 * systemd-nspawn can now consume OCI runtime bundles using a new
465 --oci-bundle= option. This implementation is fully usable, with most
466 features in the specification implemented, but since this a lot of
467 new code and functionality, this feature should most likely not
468 be used in production yet.
470 * systemd-nspawn now supports various options described by the OCI
471 runtime specification on the command-line and in .nspawn files:
472 --inaccessible=/Inaccessible= may be used to mask parts of the file
473 system tree, --console=/--pipe may be used to configure how standard
474 input, output, and error are set up.
476 * busctl learned the `emit` verb to generate D-Bus signals.
478 * systemd-analyze cat-config may be used to gather and display
479 configuration spread over multiple files, for example system and user
480 presets, tmpfiles.d, sysusers.d, udev rules, etc.
482 * systemd-analyze calendar now takes an optional new parameter
483 --iterations= which may be used to show a maximum number of iterations
484 the specified expression will elapse next.
486 * The sd-bus C API gained support for naming method parameters in the
489 * systemd-logind gained D-Bus APIs to specify the "reboot parameter"
490 the reboot() system call expects.
492 * journalctl learnt a new --cursor-file= option that points to a file
493 from which a cursor should be loaded in the beginning and to which
494 the updated cursor should be stored at the end.
496 * ACRN hypervisor and Windows Subsystem for Linux (WSL) are now
497 detected by systemd-detect-virt (and may also be used in
498 ConditionVirtualization=).
500 * The behaviour of systemd-logind may now be modified with environment
501 variables $SYSTEMD_REBOOT_TO_FIRMWARE_SETUP,
502 $SYSTEMD_REBOOT_TO_BOOT_LOADER_MENU, and
503 $SYSTEMD_REBOOT_TO_BOOT_LOADER_ENTRY. They cause logind to either
504 skip the relevant operation completely (when set to false), or to
505 create a flag file in /run/systemd (when set to true), instead of
506 actually commencing the real operation when requested. The presence
507 of /run/systemd/reboot-to-firmware-setup,
508 /run/systemd/reboot-to-boot-loader-menu, and
509 /run/systemd/reboot-to-boot-loader-entry, may be used by alternative
510 boot loader implementations to replace some steps logind performs
511 during reboot with their own operations.
513 * systemctl can be used to request a reboot into the boot loader menu
514 or a specific boot loader entry with the new --boot-load-menu= and
515 --boot-loader-entry= options to a reboot command. (This requires a
516 boot loader that supports this, for example sd-boot.)
518 * kernel-install will no longer unconditionally create the output
519 directory (e.g. /efi/<machine-id>/<kernel-version>) for boot loader
520 snippets, but will do only if the machine-specific parent directory
521 (i.e. /efi/<machine-id>/) already exists. bootctl has been modified
522 to create this parent directory during sd-boot installation.
524 This makes it easier to use kernel-install with plugins which support
525 a different layout of the bootloader partitions (for example grub2).
527 * During package installation (with `ninja install`), we would create
528 symlinks for getty@tty1.service, systemd-networkd.service,
529 systemd-networkd.socket, systemd-resolved.service,
530 remote-cryptsetup.target, remote-fs.target,
531 systemd-networkd-wait-online.service, and systemd-timesyncd.service
532 in /etc, as if `systemctl enable` was called for those units, to make
533 the system usable immediately after installation. Now this is not
534 done anymore, and instead calling `systemctl preset-all` is
535 recommended after the first installation of systemd.
537 * A new boolean sandboxing option RestrictSUIDSGID= has been added that
538 is built on seccomp. When turned on creation of SUID/SGID files is
541 * The NoNewPrivileges= and the new RestrictSUIDSGID= options are now
542 implied if DynamicUser= is turned on for a service. This hardens
543 these services, so that they neither can benefit from nor create
544 SUID/SGID executables. This is a minor compatibility breakage, given
545 that when DynamicUser= was first introduced SUID/SGID behaviour was
546 unaffected. However, the security benefit of these two options is
547 substantial, and the setting is still relatively new, hence we opted
548 to make it mandatory for services with dynamic users.
550 Contributions from: Adam Jackson, Alexander Tsoy, Andrey Yashkin,
551 Andrzej Pietrasiewicz, Anita Zhang, Balint Reczey, Beniamino Galvani,
552 Ben Iofel, Benjamin Berg, Benjamin Dahlhoff, Chris, Chris Morin,
553 Christopher Wong, Claudius Ellsel, Clemens Gruber, dana, Daniel Black,
554 Davide Cavalca, David Michael, David Rheinsberg, emersion, Evgeny
555 Vereshchagin, Filipe Brandenburger, Franck Bui, Frantisek Sumsal,
556 Giacinto Cifelli, Hans de Goede, Hugo Kindel, Ignat Korchagin, Insun
557 Pyo, Jan Engelhardt, Jonas Dorel, Jonathan Lebon, Jonathon Kowalski,
558 Jörg Sommer, Jörg Thalheim, Jussi Pakkanen, Kai-Heng Feng, Lennart
559 Poettering, Lubomir Rintel, Luís Ferreira, Martin Pitt, Matthias
560 Klumpp, Michael Biebl, Michael Niewöhner, Michael Olbrich, Michal
561 Sekletar, Mike Lothian, Paul Menzel, Piotr Drąg, Riccardo Schirone,
562 Robin Elvedi, Roman Kulikov, Ronald Tschalär, Ross Burton, Ryan
563 Gonzalez, Sebastian Krzyszkowiak, Stephane Chazelas, StKob, Susant
564 Sahani, Sylvain Plantefève, Szabolcs Fruhwald, Taro Yamada, Theo
565 Ouzhinski, Thomas Haller, Tobias Jungel, Tom Yan, Tony Asleson, Topi
566 Miettinen, unixsysadmin, Van Laser, Vesa Jääskeläinen, Yu, Li-Yu,
567 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
573 * The default locale can now be configured at compile time. Otherwise,
574 a suitable default will be selected automatically (one of C.UTF-8,
577 * The version string shown by systemd and other tools now includes the
578 git commit hash when built from git. An override may be specified
579 during compilation, which is intended to be used by distributions to
580 include the package release information.
582 * systemd-cat can now filter standard input and standard error streams
583 for different syslog priorities using the new --stderr-priority=
586 * systemd-journald and systemd-journal-remote reject entries which
587 contain too many fields (CVE-2018-16865) and set limits on the
588 process' command line length (CVE-2018-16864).
590 * $DBUS_SESSION_BUS_ADDRESS environment variable is set by pam_systemd
593 * A new network device NamePolicy "keep" is implemented for link files,
594 and used by default in 99-default.link (the fallback configuration
595 provided by systemd). With this policy, if the network device name
596 was already set by userspace, the device will not be renamed again.
597 This matches the naming scheme that was implemented before
598 systemd-240. If naming-scheme < 240 is specified, the "keep" policy
599 is also enabled by default, even if not specified. Effectively, this
600 means that if naming-scheme >= 240 is specified, network devices will
601 be renamed according to the configuration, even if they have been
602 renamed already, if "keep" is not specified as the naming policy in
603 the .link file. The 99-default.link file provided by systemd includes
604 "keep" for backwards compatibility, but it is recommended for user
605 installed .link files to *not* include it.
607 The "kernel" policy, which keeps kernel names declared to be
608 "persistent", now works again as documented.
610 * kernel-install script now optionally takes the paths to one or more
611 initrd files, and passes them to all plugins.
613 * The mincore() system call has been dropped from the @system-service
614 system call filter group, as it is pretty exotic and may potentially
615 used for side-channel attacks.
617 * -fPIE is dropped from compiler and linker options. Please specify
618 -Db_pie=true option to meson to build position-independent
619 executables. Note that the meson option is supported since meson-0.49.
621 * The fs.protected_regular and fs.protected_fifos sysctls, which were
622 added in Linux 4.19 to make some data spoofing attacks harder, are
623 now enabled by default. While this will hopefully improve the
624 security of most installations, it is technically a backwards
625 incompatible change; to disable these sysctls again, place the
626 following lines in /etc/sysctl.d/60-protected.conf or a similar file:
628 fs.protected_regular = 0
629 fs.protected_fifos = 0
631 Note that the similar hardlink and symlink protection has been
632 enabled since v199, and may be disabled likewise.
634 * The files read from the EnvironmentFile= setting in unit files now
635 parse backslashes inside quotes literally, matching the behaviour of
638 * udevadm trigger, udevadm control, udevadm settle and udevadm monitor
639 now automatically become NOPs when run in a chroot() environment.
641 * The tmpfiles.d/ "C" line type will now copy directory trees not only
642 when the destination is so far missing, but also if it already exists
643 as a directory and is empty. This is useful to cater for systems
644 where directory trees are put together from multiple separate mount
645 points but otherwise empty.
647 * A new function sd_bus_close_unref() (and the associated
648 sd_bus_close_unrefp()) has been added to libsystemd, that combines
649 sd_bus_close() and sd_bus_unref() in one.
651 * udevadm control learnt a new option for --ping for testing whether a
652 systemd-udevd instance is running and reacting.
654 * udevadm trigger learnt a new option for --wait-daemon for waiting
655 systemd-udevd daemon to be initialized.
657 Contributions from: Aaron Plattner, Alberts Muktupāvels, Alex Mayer,
658 Ayman Bagabas, Beniamino Galvani, Burt P, Chris Down, Chris Lamb, Chris
659 Morin, Christian Hesse, Claudius Ellsel, dana, Daniel Axtens, Daniele
660 Medri, Dave Reisner, David Santamaría Rogado, Diego Canuhe, Dimitri
661 John Ledkov, Evgeny Vereshchagin, Fabrice Fontaine, Filipe
662 Brandenburger, Franck Bui, Frantisek Sumsal, govwin, Hans de Goede,
663 James Hilliard, Jan Engelhardt, Jani Uusitalo, Jan Janssen, Jan
664 Synacek, Jonathan McDowell, Jonathan Roemer, Jonathon Kowalski, Joost
665 Heitbrink, Jörg Thalheim, Lance, Lennart Poettering, Louis Taylor,
666 Lucas Werkmeister, Mantas Mikulėnas, Marc-Antoine Perennou,
667 marvelousblack, Michael Biebl, Michael Sloan, Michal Sekletar, Mike
668 Auty, Mike Gilbert, Mikhail Kasimov, Neil Brown, Niklas Hambüchen,
669 Patrick Williams, Paul Seyfert, Peter Hutterer, Philip Withnall, Roger
670 James, Ronnie P. Thomas, Ryan Gonzalez, Sam Morris, Stephan Edel,
671 Stephan Gerhold, Susant Sahani, Taro Yamada, Thomas Haller, Topi
672 Miettinen, YiFei Zhu, YmrDtnJu, YunQiang Su, Yu Watanabe, Zbigniew
673 Jędrzejewski-Szmek, zsergeant77, Дамјан Георгиевски
679 * NoNewPrivileges=yes has been set for all long-running services
680 implemented by systemd. Previously, this was problematic due to
681 SELinux (as this would also prohibit the transition from PID1's label
682 to the service's label). This restriction has since been lifted, but
683 an SELinux policy update is required.
684 (See e.g. https://github.com/fedora-selinux/selinux-policy/pull/234.)
686 * DynamicUser=yes is dropped from systemd-networkd.service,
687 systemd-resolved.service and systemd-timesyncd.service, which was
688 enabled in v239 for systemd-networkd.service and systemd-resolved.service,
689 and since v236 for systemd-timesyncd.service. The users and groups
690 systemd-network, systemd-resolve and systemd-timesync are created
691 by systemd-sysusers again. Distributors or system administrators
692 may need to create these users and groups if they not exist (or need
693 to re-enable DynamicUser= for those units) while upgrading systemd.
694 Also, the clock file for systemd-timesyncd may need to move from
695 /var/lib/private/systemd/timesync/clock to /var/lib/systemd/timesync/clock.
697 * When unit files are loaded from disk, previously systemd would
698 sometimes (depending on the unit loading order) load units from the
699 target path of symlinks in .wants/ or .requires/ directories of other
700 units. This meant that unit could be loaded from different paths
701 depending on whether the unit was requested explicitly or as a
702 dependency of another unit, not honouring the priority of directories
703 in search path. It also meant that it was possible to successfully
704 load and start units which are not found in the unit search path, as
705 long as they were requested as a dependency and linked to from
706 .wants/ or .requires/. The target paths of those symlinks are not
707 used for loading units anymore and the unit file must be found in
710 * A new service type has been added: Type=exec. It's very similar to
711 Type=simple but ensures the service manager will wait for both fork()
712 and execve() of the main service binary to complete before proceeding
713 with follow-up units. This is primarily useful so that the manager
714 propagates any errors in the preparation phase of service execution
715 back to the job that requested the unit to be started. For example,
716 consider a service that has ExecStart= set to a file system binary
717 that doesn't exist. With Type=simple starting the unit would be
718 considered instantly successful, as only fork() has to complete
719 successfully and the manager does not wait for execve(), and hence
720 its failure is seen "too late". With the new Type=exec service type
721 starting the unit will fail, as the manager will wait for the
722 execve() and notice its failure, which is then propagated back to the
725 NOTE: with the next release 241 of systemd we intend to change the
726 systemd-run tool to default to Type=exec for transient services
727 started by it. This should be mostly safe, but in specific corner
728 cases might result in problems, as the systemd-run tool will then
729 block on NSS calls (such as user name look-ups due to User=) done
730 between the fork() and execve(), which under specific circumstances
731 might cause problems. It is recommended to specify "-p Type=simple"
732 explicitly in the few cases where this applies. For regular,
733 non-transient services (i.e. those defined with unit files on disk)
734 we will continue to default to Type=simple.
736 * The Linux kernel's current default RLIMIT_NOFILE resource limit for
737 userspace processes is set to 1024 (soft) and 4096
738 (hard). Previously, systemd passed this on unmodified to all
739 processes it forked off. With this systemd release the hard limit
740 systemd passes on is increased to 512K, overriding the kernel's
741 defaults and substantially increasing the number of simultaneous file
742 descriptors unprivileged userspace processes can allocate. Note that
743 the soft limit remains at 1024 for compatibility reasons: the
744 traditional UNIX select() call cannot deal with file descriptors >=
745 1024 and increasing the soft limit globally might thus result in
746 programs unexpectedly allocating a high file descriptor and thus
747 failing abnormally when attempting to use it with select() (of
748 course, programs shouldn't use select() anymore, and prefer
749 poll()/epoll, but the call unfortunately remains undeservedly popular
750 at this time). This change reflects the fact that file descriptor
751 handling in the Linux kernel has been optimized in more recent
752 kernels and allocating large numbers of them should be much cheaper
753 both in memory and in performance than it used to be. Programs that
754 want to take benefit of the increased limit have to "opt-in" into
755 high file descriptors explicitly by raising their soft limit. Of
756 course, when they do that they must acknowledge that they cannot use
757 select() anymore (and neither can any shared library they use — or
758 any shared library used by any shared library they use and so on).
759 Which default hard limit is most appropriate is of course hard to
760 decide. However, given reports that ~300K file descriptors are used
761 in real-life applications we believe 512K is sufficiently high as new
762 default for now. Note that there are also reports that using very
763 high hard limits (e.g. 1G) is problematic: some software allocates
764 large arrays with one element for each potential file descriptor
765 (Java, …) — a high hard limit thus triggers excessively large memory
766 allocations in these applications. Hopefully, the new default of 512K
767 is a good middle ground: higher than what real-life applications
768 currently need, and low enough for avoid triggering excessively large
769 allocations in problematic software. (And yes, somebody should fix
772 * The fs.nr_open and fs.file-max sysctls are now automatically bumped
773 to the highest possible values, as separate accounting of file
774 descriptors is no longer necessary, as memcg tracks them correctly as
775 part of the memory accounting anyway. Thus, from the four limits on
776 file descriptors currently enforced (fs.file-max, fs.nr_open,
777 RLIMIT_NOFILE hard, RLIMIT_NOFILE soft) we turn off the first two,
778 and keep only the latter two. A set of build-time options
779 (-Dbump-proc-sys-fs-file-max=false and -Dbump-proc-sys-fs-nr-open=false)
780 has been added to revert this change in behaviour, which might be
781 an option for systems that turn off memcg in the kernel.
783 * When no /etc/locale.conf file exists (and hence no locale settings
784 are in place), systemd will now use the "C.UTF-8" locale by default,
785 and set LANG= to it. This locale is supported by various
786 distributions including Fedora, with clear indications that upstream
787 glibc is going to make it available too. This locale enables UTF-8
788 mode by default, which appears appropriate for 2018.
790 * The "net.ipv4.conf.all.rp_filter" sysctl will now be set to 2 by
791 default. This effectively switches the RFC3704 Reverse Path filtering
792 from Strict mode to Loose mode. This is more appropriate for hosts
793 that have multiple links with routes to the same networks (e.g.
794 a client with a Wi-Fi and Ethernet both connected to the internet).
796 Consult the kernel documentation for details on this sysctl:
797 https://www.kernel.org/doc/Documentation/networking/ip-sysctl.txt
799 * CPUAccounting=yes no longer enables the CPU controller when using
800 kernel 4.15+ and the unified cgroup hierarchy, as required accounting
801 statistics are now provided independently from the CPU controller.
803 * Support for disabling a particular cgroup controller within a sub-tree
804 has been added through the DisableControllers= directive.
806 * cgroup_no_v1=all on the kernel command line now also implies
807 using the unified cgroup hierarchy, unless one explicitly passes
808 systemd.unified_cgroup_hierarchy=0 on the kernel command line.
810 * The new "MemoryMin=" unit file property may now be used to set the
811 memory usage protection limit of processes invoked by the unit. This
812 controls the cgroup v2 memory.min attribute. Similarly, the new
813 "IODeviceLatencyTargetSec=" property has been added, wrapping the new
814 cgroup v2 io.latency cgroup property for configuring per-service I/O
817 * systemd now supports the cgroup v2 devices BPF logic, as counterpart
818 to the cgroup v1 "devices" cgroup controller.
820 * systemd-escape now is able to combine --unescape with --template. It
821 also learnt a new option --instance for extracting and unescaping the
822 instance part of a unit name.
824 * sd-bus now provides the sd_bus_message_readv() which is similar to
825 sd_bus_message_read() but takes a va_list object. The pair
826 sd_bus_set_method_call_timeout() and sd_bus_get_method_call_timeout()
827 has been added for configuring the default method call timeout to
828 use. sd_bus_error_move() may be used to efficiently move the contents
829 from one sd_bus_error structure to another, invalidating the
830 source. sd_bus_set_close_on_exit() and sd_bus_get_close_on_exit() may
831 be used to control whether a bus connection object is automatically
832 flushed when an sd-event loop is exited.
834 * When processing classic BSD syslog log messages, journald will now
835 save the original time-stamp string supplied in the new
836 SYSLOG_TIMESTAMP= journal field. This permits consumers to
837 reconstruct the original BSD syslog message more correctly.
839 * StandardOutput=/StandardError= in service files gained support for
840 new "append:…" parameters, for connecting STDOUT/STDERR of a service
841 to a file, and appending to it.
843 * The signal to use as last step of killing of unit processes is now
844 configurable. Previously it was hard-coded to SIGKILL, which may now
845 be overridden with the new KillSignal= setting. Note that this is the
846 signal used when regular termination (i.e. SIGTERM) does not suffice.
847 Similarly, the signal used when aborting a program in case of a
848 watchdog timeout may now be configured too (WatchdogSignal=).
850 * The XDG_SESSION_DESKTOP environment variable may now be configured in
851 the pam_systemd argument line, using the new desktop= switch. This is
852 useful to initialize it properly from a display manager without
853 having to touch C code.
855 * Most configuration options that previously accepted percentage values
856 now also accept permille values with the '‰' suffix (instead of '%').
858 * systemd-resolved may now optionally use OpenSSL instead of GnuTLS for
861 * systemd-resolved's configuration file resolved.conf gained a new
862 option ReadEtcHosts= which may be used to turn off processing and
863 honoring /etc/hosts entries.
865 * The "--wait" switch may now be passed to "systemctl
866 is-system-running", in which case the tool will synchronously wait
867 until the system finished start-up.
869 * hostnamed gained a new bus call to determine the DMI product UUID.
871 * On x86-64 systemd will now prefer using the RDRAND processor
872 instruction over /dev/urandom whenever it requires randomness that
873 neither has to be crypto-grade nor should be reproducible. This
874 should substantially reduce the amount of entropy systemd requests
875 from the kernel during initialization on such systems, though not
876 reduce it to zero. (Why not zero? systemd still needs to allocate
877 UUIDs and such uniquely, which require high-quality randomness.)
879 * networkd gained support for Foo-Over-UDP, ERSPAN and ISATAP
880 tunnels. It also gained a new option ForceDHCPv6PDOtherInformation=
881 for forcing the "Other Information" bit in IPv6 RA messages. The
882 bonding logic gained four new options AdActorSystemPriority=,
883 AdUserPortKey=, AdActorSystem= for configuring various 802.3ad
884 aspects, and DynamicTransmitLoadBalancing= for enabling dynamic
885 shuffling of flows. The tunnel logic gained a new
886 IPv6RapidDeploymentPrefix= option for configuring IPv6 Rapid
887 Deployment. The policy rule logic gained four new options IPProtocol=,
888 SourcePort= and DestinationPort=, InvertRule=. The bridge logic gained
889 support for the MulticastToUnicast= option. networkd also gained
890 support for configuring static IPv4 ARP or IPv6 neighbor entries.
892 * .preset files (as read by 'systemctl preset') may now be used to
893 instantiate services.
895 * /etc/crypttab now understands the sector-size= option to configure
896 the sector size for an encrypted partition.
898 * Key material for encrypted disks may now be placed on a formatted
899 medium, and referenced from /etc/crypttab by the UUID of the file
900 system, followed by "=" suffixed by the path to the key file.
902 * The "collect" udev component has been removed without replacement, as
903 it is neither used nor maintained.
905 * When the RuntimeDirectory=, StateDirectory=, CacheDirectory=,
906 LogsDirectory=, ConfigurationDirectory= settings are used in a
907 service the executed processes will now receive a set of environment
908 variables containing the full paths of these directories.
909 Specifically, RUNTIME_DIRECTORY=, STATE_DIRECTORY, CACHE_DIRECTORY,
910 LOGS_DIRECTORY, CONFIGURATION_DIRECTORY are now set if these options
911 are used. Note that these options may be used multiple times per
912 service in which case the resulting paths will be concatenated and
915 * Predictable interface naming has been extended to cover InfiniBand
916 NICs. They will be exposed with an "ib" prefix.
918 * tmpfiles.d/ line types may now be suffixed with a '-' character, in
919 which case the respective line failing is ignored.
921 * .link files may now be used to configure the equivalent to the
922 "ethtool advertise" commands.
924 * The sd-device.h and sd-hwdb.h APIs are now exported, as an
925 alternative to libudev.h. Previously, the latter was just an internal
926 wrapper around the former, but now these two APIs are exposed
929 * sd-id128.h gained a new function sd_id128_get_boot_app_specific()
930 which calculates an app-specific boot ID similar to how
931 sd_id128_get_machine_app_specific() generates an app-specific machine
934 * A new tool systemd-id128 has been added that can be used to determine
935 and generate various 128bit IDs.
937 * /etc/os-release gained two new standardized fields DOCUMENTATION_URL=
940 * systemd-hibernate-resume-generator will now honor the "noresume"
941 kernel command line option, in which case it will bypass resuming
942 from any hibernated image.
944 * The systemd-sleep.conf configuration file gained new options
945 AllowSuspend=, AllowHibernation=, AllowSuspendThenHibernate=,
946 AllowHybridSleep= for prohibiting specific sleep modes even if the
949 * portablectl is now officially supported and has thus moved to
952 * bootctl learnt the two new commands "set-default" and "set-oneshot"
953 for setting the default boot loader item to boot to (either
954 persistently or only for the next boot). This is currently only
955 compatible with sd-boot, but may be implemented on other boot loaders
956 too, that follow the boot loader interface. The updated interface is
959 https://systemd.io/BOOT_LOADER_INTERFACE
961 * A new kernel command line option systemd.early_core_pattern= is now
962 understood which may be used to influence the core_pattern PID 1
963 installs during early boot.
965 * busctl learnt two new options -j and --json= for outputting method
966 call replies, properties and monitoring output in JSON.
968 * journalctl's JSON output now supports simple ANSI coloring as well as
969 a new "json-seq" mode for generating RFC7464 output.
971 * Unit files now support the %g/%G specifiers that resolve to the UNIX
972 group/GID of the service manager runs as, similar to the existing
973 %u/%U specifiers that resolve to the UNIX user/UID.
975 * systemd-logind learnt a new global configuration option
976 UserStopDelaySec= that may be set in logind.conf. It specifies how
977 long the systemd --user instance shall remain started after a user
978 logs out. This is useful to speed up repetitive re-connections of the
979 same user, as it means the user's service manager doesn't have to be
980 stopped/restarted on each iteration, but can be reused between
981 subsequent options. This setting defaults to 10s. systemd-logind also
982 exports two new properties on its Manager D-Bus objects indicating
983 whether the system's lid is currently closed, and whether the system
986 * systemd gained support for a generic boot counting logic, which
987 generically permits automatic reverting to older boot loader entries
988 if newer updated ones don't work. The boot loader side is implemented
989 in sd-boot, but is kept open for other boot loaders too. For details
992 https://systemd.io/AUTOMATIC_BOOT_ASSESSMENT
994 * The SuccessAction=/FailureAction= unit file settings now learnt two
995 new parameters: "exit" and "exit-force", which result in immediate
996 exiting of the service manager, and are only useful in systemd --user
997 and container environments.
999 * Unit files gained support for a pair of options
1000 FailureActionExitStatus=/SuccessActionExitStatus= for configuring the
1001 exit status to use as service manager exit status when
1002 SuccessAction=/FailureAction= is set to exit or exit-force.
1004 * A pair of LogRateLimitIntervalSec=/LogRateLimitBurst= per-service
1005 options may now be used to configure the log rate limiting applied by
1006 journald per-service.
1008 * systemd-analyze gained a new verb "timespan" for parsing and
1009 normalizing time span values (i.e. strings like "5min 7s 8us").
1011 * systemd-analyze also gained a new verb "security" for analyzing the
1012 security and sand-boxing settings of services in order to determine an
1013 "exposure level" for them, indicating whether a service would benefit
1014 from more sand-boxing options turned on for them.
1016 * "systemd-analyze syscall-filter" will now also show system calls
1017 supported by the local kernel but not included in any of the defined
1020 * .nspawn files now understand the Ephemeral= setting, matching the
1021 --ephemeral command line switch.
1023 * sd-event gained the new APIs sd_event_source_get_floating() and
1024 sd_event_source_set_floating() for controlling whether a specific
1025 event source is "floating", i.e. destroyed along with the even loop
1028 * Unit objects on D-Bus gained a new "Refs" property that lists all
1029 clients that currently have a reference on the unit (to ensure it is
1032 * The JoinControllers= option in system.conf is no longer supported, as
1033 it didn't work correctly, is hard to support properly, is legacy (as
1034 the concept only exists on cgroup v1) and apparently wasn't used.
1036 * Journal messages that are generated whenever a unit enters the failed
1037 state are now tagged with a unique MESSAGE_ID. Similarly, messages
1038 generated whenever a service process exits are now made recognizable,
1039 too. A tagged message is also emitted whenever a unit enters the
1040 "dead" state on success.
1042 * systemd-run gained a new switch --working-directory= for configuring
1043 the working directory of the service to start. A shortcut -d is
1044 equivalent, setting the working directory of the service to the
1045 current working directory of the invoking program. The new --shell
1046 (or just -S) option has been added for invoking the $SHELL of the
1047 caller as a service, and implies --pty --same-dir --wait --collect
1048 --service-type=exec. Or in other words, "systemd-run -S" is now the
1049 quickest way to quickly get an interactive in a fully clean and
1050 well-defined system service context.
1052 * machinectl gained a new verb "import-fs" for importing an OS tree
1053 from a directory. Moreover, when a directory or tarball is imported
1054 and single top-level directory found with the OS itself below the OS
1055 tree is automatically mangled and moved one level up.
1057 * systemd-importd will no longer set up an implicit btrfs loop-back
1058 file system on /var/lib/machines. If one is already set up, it will
1059 continue to be used.
1061 * A new generator "systemd-run-generator" has been added. It will
1062 synthesize a unit from one or more program command lines included in
1063 the kernel command line. This is very useful in container managers
1066 # systemd-nspawn -i someimage.raw -b systemd.run='"some command line"'
1068 This will run "systemd-nspawn" on an image, invoke the specified
1069 command line and immediately shut down the container again, returning
1070 the command line's exit code.
1072 * The block device locking logic is now documented:
1074 https://systemd.io/BLOCK_DEVICE_LOCKING
1076 * loginctl and machinectl now optionally output the various tables in
1077 JSON using the --output= switch. It is our intention to add similar
1078 support to systemctl and all other commands.
1080 * udevadm's query and trigger verb now optionally take a .device unit
1083 * systemd-udevd's network naming logic now understands a new
1084 net.naming-scheme= kernel command line switch, which may be used to
1085 pick a specific version of the naming scheme. This helps stabilizing
1086 interface names even as systemd/udev are updated and the naming logic
1089 * sd-id128.h learnt two new auxiliary helpers: sd_id128_is_allf() and
1090 SD_ID128_ALLF to test if a 128bit ID is set to all 0xFF bytes, and to
1091 initialize one to all 0xFF.
1093 * After loading the SELinux policy systemd will now recursively relabel
1094 all files and directories listed in
1095 /run/systemd/relabel-extra.d/*.relabel (which should be simple
1096 newline separated lists of paths) in addition to the ones it already
1097 implicitly relabels in /run, /dev and /sys. After the relabelling is
1098 completed the *.relabel files (and /run/systemd/relabel-extra.d/) are
1099 removed. This is useful to permit initrds (i.e. code running before
1100 the SELinux policy is in effect) to generate files in the host
1101 filesystem safely and ensure that the correct label is applied during
1102 the transition to the host OS.
1104 * KERNEL API BREAKAGE: Linux kernel 4.18 changed behaviour regarding
1105 mknod() handling in user namespaces. Previously mknod() would always
1106 fail with EPERM in user namespaces. Since 4.18 mknod() will succeed
1107 but device nodes generated that way cannot be opened, and attempts to
1108 open them result in EPERM. This breaks the "graceful fallback" logic
1109 in systemd's PrivateDevices= sand-boxing option. This option is
1110 implemented defensively, so that when systemd detects it runs in a
1111 restricted environment (such as a user namespace, or an environment
1112 where mknod() is blocked through seccomp or absence of CAP_SYS_MKNOD)
1113 where device nodes cannot be created the effect of PrivateDevices= is
1114 bypassed (following the logic that 2nd-level sand-boxing is not
1115 essential if the system systemd runs in is itself already sand-boxed
1116 as a whole). This logic breaks with 4.18 in container managers where
1117 user namespacing is used: suddenly PrivateDevices= succeeds setting
1118 up a private /dev/ file system containing devices nodes — but when
1119 these are opened they don't work.
1121 At this point it is recommended that container managers utilizing
1122 user namespaces that intend to run systemd in the payload explicitly
1123 block mknod() with seccomp or similar, so that the graceful fallback
1126 We are very sorry for the breakage and the requirement to change
1127 container configurations for newer kernels. It's purely caused by an
1128 incompatible kernel change. The relevant kernel developers have been
1129 notified about this userspace breakage quickly, but they chose to
1132 * PermissionsStartOnly= setting is deprecated (but is still supported
1133 for backwards compatibility). The same functionality is provided by
1134 the more flexible "+", "!", and "!!" prefixes to ExecStart= and other
1137 * $DBUS_SESSION_BUS_ADDRESS environment variable is not set by
1138 pam_systemd anymore.
1140 * The naming scheme for network devices was changed to always rename
1141 devices, even if they were already renamed by userspace. The "kernel"
1142 policy was changed to only apply as a fallback, if no other naming
1145 * The requirements to build systemd is bumped to meson-0.46 and
1148 Contributions from: afg, Alan Jenkins, Aleksei Timofeyev, Alexander
1149 Filippov, Alexander Kurtz, Alexey Bogdanenko, Andreas Henriksson,
1150 Andrew Jorgensen, Anita Zhang, apnix-uk, Arkan49, Arseny Maslennikov,
1151 asavah, Asbjørn Apeland, aszlig, Bastien Nocera, Ben Boeckel, Benedikt
1152 Morbach, Benjamin Berg, Bruce Zhang, Carlo Caione, Cedric Viou, Chen
1153 Qi, Chris Chiu, Chris Down, Chris Morin, Christian Rebischke, Claudius
1154 Ellsel, Colin Guthrie, dana, Daniel, Daniele Medri, Daniel Kahn
1155 Gillmor, Daniel Rusek, Daniel van Vugt, Dariusz Gadomski, Dave Reisner,
1156 David Anderson, Davide Cavalca, David Leeds, David Malcolm, David
1157 Strauss, David Tardon, Dimitri John Ledkov, Dmitry Torokhov, dj-kaktus,
1158 Dongsu Park, Elias Probst, Emil Soleyman, Erik Kooistra, Ervin Peters,
1159 Evgeni Golov, Evgeny Vereshchagin, Fabrice Fontaine, Faheel Ahmad,
1160 Faizal Luthfi, Felix Yan, Filipe Brandenburger, Franck Bui, Frank
1161 Schaefer, Frantisek Sumsal, Gautier Husson, Gianluca Boiano, Giuseppe
1162 Scrivano, glitsj16, Hans de Goede, Harald Hoyer, Harry Mallon, Harshit
1163 Jain, Helmut Grohne, Henry Tung, Hui Yiqun, imayoda, Insun Pyo, Iwan
1164 Timmer, Jan Janssen, Jan Pokorný, Jan Synacek, Jason A. Donenfeld,
1165 javitoom, Jérémy Nouhaud, Jeremy Su, Jiuyang Liu, João Paulo Rechi
1166 Vita, Joe Hershberger, Joe Rayhawk, Joerg Behrmann, Joerg Steffens,
1167 Jonas Dorel, Jon Ringle, Josh Soref, Julian Andres Klode, Jun Bo Bi,
1168 Jürg Billeter, Keith Busch, Khem Raj, Kirill Marinushkin, Larry
1169 Bernstone, Lennart Poettering, Lion Yang, Li Song, Lorenz
1170 Hübschle-Schneider, Lubomir Rintel, Lucas Werkmeister, Ludwin Janvier,
1171 Lukáš Nykrýn, Luke Shumaker, mal, Marc-Antoine Perennou, Marcin
1172 Skarbek, Marco Trevisan (Treviño), Marian Cepok, Mario Hros, Marko
1173 Myllynen, Markus Grimm, Martin Pitt, Martin Sobotka, Martin Wilck,
1174 Mathieu Trudel-Lapierre, Matthew Leeds, Michael Biebl, Michael Olbrich,
1175 Michael 'pbone' Pobega, Michael Scherer, Michal Koutný, Michal
1176 Sekletar, Michal Soltys, Mike Gilbert, Mike Palmer, Muhammet Kara, Neal
1177 Gompa, Neil Brown, Network Silence, Niklas Tibbling, Nikolas Nyby,
1178 Nogisaka Sadata, Oliver Smith, Patrik Flykt, Pavel Hrdina, Paweł
1179 Szewczyk, Peter Hutterer, Piotr Drąg, Ray Strode, Reinhold Mueller,
1180 Renaud Métrich, Roman Gushchin, Ronny Chevalier, Rubén Suárez Alvarez,
1181 Ruixin Bao, RussianNeuroMancer, Ryutaroh Matsumoto, Saleem Rashid, Sam
1182 Morris, Samuel Morris, Sandy Carter, scootergrisen, Sébastien Bacher,
1183 Sergey Ptashnick, Shawn Landden, Shengyao Xue, Shih-Yuan Lee
1184 (FourDollars), Silvio Knizek, Sjoerd Simons, Stasiek Michalski, Stephen
1185 Gallagher, Steven Allen, Steve Ramage, Susant Sahani, Sven Joachim,
1186 Sylvain Plantefève, Tanu Kaskinen, Tejun Heo, Thiago Macieira, Thomas
1187 Blume, Thomas Haller, Thomas H. P. Andersen, Tim Ruffing, TJ, Tobias
1188 Jungel, Todd Walton, Tommi Rantala, Tomsod M, Tony Novak, Tore
1189 Anderson, Trevonn, Victor Laskurain, Victor Tapia, Violet Halo, Vojtech
1190 Trefny, welaq, William A. Kennington III, William Douglas, Wyatt Ward,
1191 Xiang Fan, Xi Ruoyao, Xuanwo, Yann E. Morin, YmrDtnJu, Yu Watanabe,
1192 Zbigniew Jędrzejewski-Szmek, Zhang Xianwei, Zsolt Dollenstein
1194 — Warsaw, 2018-12-21
1198 * NETWORK INTERFACE DEVICE NAMING CHANGES: systemd-udevd's "net_id"
1199 builtin will name network interfaces differently than in previous
1200 versions for virtual network interfaces created with SR-IOV and NPAR
1201 and for devices where the PCI network controller device does not have
1202 a slot number associated.
1204 SR-IOV virtual devices are now named based on the name of the parent
1205 interface, with a suffix of "v<N>", where <N> is the virtual device
1206 number. Previously those virtual devices were named as if completely
1209 The ninth and later NPAR virtual devices will be named following the
1210 scheme used for the first eight NPAR partitions. Previously those
1211 devices were not renamed and the kernel default (eth<n>) was used.
1213 "net_id" will also generate names for PCI devices where the PCI
1214 network controller device does not have an associated slot number
1215 itself, but one of its parents does. Previously those devices were
1216 not renamed and the kernel default (eth<n>) was used.
1218 * AF_INET and AF_INET6 are dropped from RestrictAddressFamilies= in
1219 systemd-logind.service. Since v235, IPAddressDeny=any has been set to
1220 the unit. So, it is expected that the default behavior of
1221 systemd-logind is not changed. However, if distribution packagers or
1222 administrators disabled or modified IPAddressDeny= setting by a
1223 drop-in config file, then it may be necessary to update the file to
1224 re-enable AF_INET and AF_INET6 to support network user name services,
1227 * When the RestrictNamespaces= unit property is specified multiple
1228 times, then the specified types are merged now. Previously, only the
1229 last assignment was used. So, if distribution packagers or
1230 administrators modified the setting by a drop-in config file, then it
1231 may be necessary to update the file.
1233 * When OnFailure= is used in combination with Restart= on a service
1234 unit, then the specified units will no longer be triggered on
1235 failures that result in restarting. Previously, the specified units
1236 would be activated each time the unit failed, even when the unit was
1237 going to be restarted automatically. This behaviour contradicted the
1238 documentation. With this release the code is adjusted to match the
1241 * systemd-tmpfiles will now print a notice whenever it encounters
1242 tmpfiles.d/ lines referencing the /var/run/ directory. It will
1243 recommend reworking them to use the /run/ directory instead (for
1244 which /var/run/ is simply a symlinked compatibility alias). This way
1245 systemd-tmpfiles can properly detect line conflicts and merge lines
1246 referencing the same file by two paths, without having to access
1249 * systemctl disable/unmask/preset/preset-all cannot be used with
1250 --runtime. Previously this was allowed, but resulted in unintuitive
1251 behaviour that wasn't useful. systemctl disable/unmask will now undo
1252 both runtime and persistent enablement/masking, i.e. it will remove
1253 any relevant symlinks both in /run and /etc.
1255 * Note that all long-running system services shipped with systemd will
1256 now default to a system call whitelist (rather than a blacklist, as
1257 before). In particular, systemd-udevd will now enforce one too. For
1258 most cases this should be safe, however downstream distributions
1259 which disabled sandboxing of systemd-udevd (specifically the
1260 MountFlags= setting), might want to disable this security feature
1261 too, as the default whitelisting will prohibit all mount, swap,
1262 reboot and clock changing operations from udev rules.
1264 * sd-boot acquired new loader configuration settings to optionally turn
1265 off Windows and MacOS boot partition discovery as well as
1266 reboot-into-firmware menu items. It is also able to pick a better
1267 screen resolution for HiDPI systems, and now provides loader
1268 configuration settings to change the resolution explicitly.
1270 * systemd-resolved now supports DNS-over-TLS. It's still
1271 turned off by default, use DNSOverTLS=opportunistic to turn it on in
1272 resolved.conf. We intend to make this the default as soon as couple
1273 of additional techniques for optimizing the initial latency caused by
1274 establishing a TLS/TCP connection are implemented.
1276 * systemd-resolved.service and systemd-networkd.service now set
1277 DynamicUser=yes. The users systemd-resolve and systemd-network are
1278 not created by systemd-sysusers anymore.
1280 NOTE: This has a chance of breaking nss-ldap and similar NSS modules
1281 that embed a network facing module into any process using getpwuid()
1282 or related call: the dynamic allocation of the user ID for
1283 systemd-resolved.service means the service manager has to check NSS
1284 if the user name is already taken when forking off the service. Since
1285 the user in the common case won't be defined in /etc/passwd the
1286 lookup is likely to trigger nss-ldap which in turn might use NSS to
1287 ask systemd-resolved for hostname lookups. This will hence result in
1288 a deadlock: a user name lookup in order to start
1289 systemd-resolved.service will result in a host name lookup for which
1290 systemd-resolved.service needs to be started already. There are
1291 multiple ways to work around this problem: pre-allocate the
1292 "systemd-resolve" user on such systems, so that nss-ldap won't be
1293 triggered; or use a different NSS package that doesn't do networking
1294 in-process but provides a local asynchronous name cache; or configure
1295 the NSS package to avoid lookups for UIDs in the range `pkg-config
1296 systemd --variable=dynamicuidmin` … `pkg-config systemd
1297 --variable=dynamicuidmax`, so that it does not consider itself
1298 authoritative for the same UID range systemd allocates dynamic users
1301 * The systemd-resolve tool has been renamed to resolvectl (it also
1302 remains available under the old name, for compatibility), and its
1303 interface is now verb-based, similar in style to the other <xyz>ctl
1304 tools, such as systemctl or loginctl.
1306 * The resolvectl/systemd-resolve tool also provides 'resolvconf'
1307 compatibility. It may be symlinked under the 'resolvconf' name, in
1308 which case it will take arguments and input compatible with the
1309 Debian and FreeBSD resolvconf tool.
1311 * Support for suspend-then-hibernate has been added, i.e. a sleep mode
1312 where the system initially suspends, and after a timeout resumes and
1315 * networkd's ClientIdentifier= now accepts a new option "duid-only". If
1316 set the client will only send a DUID as client identifier.
1318 * The nss-systemd glibc NSS module will now enumerate dynamic users and
1319 groups in effect. Previously, it could resolve UIDs/GIDs to user
1320 names/groups and vice versa, but did not support enumeration.
1322 * journald's Compress= configuration setting now optionally accepts a
1323 byte threshold value. All journal objects larger than this threshold
1324 will be compressed, smaller ones will not. Previously this threshold
1325 was not configurable and set to 512.
1327 * A new system.conf setting NoNewPrivileges= is now available which may
1328 be used to turn off acquisition of new privileges system-wide
1329 (i.e. set Linux' PR_SET_NO_NEW_PRIVS for PID 1 itself, and thus also
1330 for all its children). Note that turning this option on means setuid
1331 binaries and file system capabilities lose their special powers.
1332 While turning on this option is a big step towards a more secure
1333 system, doing so is likely to break numerous pre-existing UNIX tools,
1334 in particular su and sudo.
1336 * A new service systemd-time-sync-wait.service has been added. If
1337 enabled it will delay the time-sync.target unit at boot until time
1338 synchronization has been received from the network. This
1339 functionality is useful on systems lacking a local RTC or where it is
1340 acceptable that the boot process shall be delayed by external network
1343 * When hibernating, systemd will now inform the kernel of the image
1344 write offset, on kernels new enough to support this. This means swap
1345 files should work for hibernation now.
1347 * When loading unit files, systemd will now look for drop-in unit files
1348 extensions in additional places. Previously, for a unit file name
1349 "foo-bar-baz.service" it would look for dropin files in
1350 "foo-bar-baz.service.d/*.conf". Now, it will also look in
1351 "foo-bar-.service.d/*.conf" and "foo-.service.d/", i.e. at the
1352 service name truncated after all inner dashes. This scheme allows
1353 writing drop-ins easily that apply to a whole set of unit files at
1354 once. It's particularly useful for mount and slice units (as their
1355 naming is prefix based), but is also useful for service and other
1356 units, for packages that install multiple unit files at once,
1357 following a strict naming regime of beginning the unit file name with
1358 the package's name. Two new specifiers are now supported in unit
1359 files to match this: %j and %J are replaced by the part of the unit
1360 name following the last dash.
1362 * Unit files and other configuration files that support specifier
1363 expansion now understand another three new specifiers: %T and %V will
1364 resolve to /tmp and /var/tmp respectively, or whatever temporary
1365 directory has been set for the calling user. %E will expand to either
1366 /etc (for system units) or $XDG_CONFIG_HOME (for user units).
1368 * The ExecStart= lines of unit files are no longer required to
1369 reference absolute paths. If non-absolute paths are specified the
1370 specified binary name is searched within the service manager's
1371 built-in $PATH, which may be queried with 'systemd-path
1372 search-binaries-default'. It's generally recommended to continue to
1373 use absolute paths for all binaries specified in unit files.
1375 * Units gained a new load state "bad-setting", which is used when a
1376 unit file was loaded, but contained fatal errors which prevent it
1377 from being started (for example, a service unit has been defined
1378 lacking both ExecStart= and ExecStop= lines).
1380 * coredumpctl's "gdb" verb has been renamed to "debug", in order to
1381 support alternative debuggers, for example lldb. The old name
1382 continues to be available however, for compatibility reasons. Use the
1383 new --debugger= switch or the $SYSTEMD_DEBUGGER environment variable
1384 to pick an alternative debugger instead of the default gdb.
1386 * systemctl and the other tools will now output escape sequences that
1387 generate proper clickable hyperlinks in various terminal emulators
1388 where useful (for example, in the "systemctl status" output you can
1389 now click on the unit file name to quickly open it in the
1390 editor/viewer of your choice). Note that not all terminal emulators
1391 support this functionality yet, but many do. Unfortunately, the
1392 "less" pager doesn't support this yet, hence this functionality is
1393 currently automatically turned off when a pager is started (which
1394 happens quite often due to auto-paging). We hope to remove this
1395 limitation as soon as "less" learns these escape sequences. This new
1396 behaviour may also be turned off explicitly with the $SYSTEMD_URLIFY
1397 environment variable. For details on these escape sequences see:
1398 https://gist.github.com/egmontkob/eb114294efbcd5adb1944c9f3cb5feda
1400 * networkd's .network files now support a new IPv6MTUBytes= option for
1401 setting the MTU used by IPv6 explicitly as well as a new MTUBytes=
1402 option in the [Route] section to configure the MTU to use for
1403 specific routes. It also gained support for configuration of the DHCP
1404 "UserClass" option through the new UserClass= setting. It gained
1405 three new options in the new [CAN] section for configuring CAN
1406 networks. The MULTICAST and ALLMULTI interface flags may now be
1407 controlled explicitly with the new Multicast= and AllMulticast=
1410 * networkd will now automatically make use of the kernel's route
1411 expiration feature, if it is available.
1413 * udevd's .link files now support setting the number of receive and
1414 transmit channels, using the RxChannels=, TxChannels=,
1415 OtherChannels=, CombinedChannels= settings.
1417 * Support for UDPSegmentationOffload= has been removed, given its
1418 limited support in hardware, and waning software support.
1420 * networkd's .netdev files now support creating "netdevsim" interfaces.
1422 * PID 1 learnt a new bus call GetUnitByControlGroup() which may be used
1423 to query the unit belonging to a specific kernel control group.
1425 * systemd-analyze gained a new verb "cat-config", which may be used to
1426 dump the contents of any configuration file, with all its matching
1427 drop-in files added in, and honouring the usual search and masking
1428 logic applied to systemd configuration files. For example use
1429 "systemd-analyze cat-config systemd/system.conf" to get the complete
1430 system configuration file of systemd how it would be loaded by PID 1
1431 itself. Similar to this, various tools such as systemd-tmpfiles or
1432 systemd-sysusers, gained a new option "--cat-config", which does the
1433 corresponding operation for their own configuration settings. For
1434 example, "systemd-tmpfiles --cat-config" will now output the full
1435 list of tmpfiles.d/ lines in place.
1437 * timedatectl gained three new verbs: "show" shows bus properties of
1438 systemd-timedated, "timesync-status" shows the current NTP
1439 synchronization state of systemd-timesyncd, and "show-timesync"
1440 shows bus properties of systemd-timesyncd.
1442 * systemd-timesyncd gained a bus interface on which it exposes details
1445 * A new environment variable $SYSTEMD_TIMEDATED_NTP_SERVICES is now
1446 understood by systemd-timedated. It takes a colon-separated list of
1447 unit names of NTP client services. The list is used by
1448 "timedatectl set-ntp".
1450 * systemd-nspawn gained a new --rlimit= switch for setting initial
1451 resource limits for the container payload. There's a new switch
1452 --hostname= to explicitly override the container's hostname. A new
1453 --no-new-privileges= switch may be used to control the
1454 PR_SET_NO_NEW_PRIVS flag for the container payload. A new
1455 --oom-score-adjust= switch controls the OOM scoring adjustment value
1456 for the payload. The new --cpu-affinity= switch controls the CPU
1457 affinity of the container payload. The new --resolv-conf= switch
1458 allows more detailed control of /etc/resolv.conf handling of the
1459 container. Similarly, the new --timezone= switch allows more detailed
1460 control of /etc/localtime handling of the container.
1462 * systemd-detect-virt gained a new --list switch, which will print a
1463 list of all currently known VM and container environments.
1465 * Support for "Portable Services" has been added, see
1466 doc/PORTABLE_SERVICES.md for details. Currently, the support is still
1467 experimental, but this is expected to change soon. Reflecting this
1468 experimental state, the "portablectl" binary is not installed into
1469 /usr/bin yet. The binary has to be called with the full path
1470 /usr/lib/systemd/portablectl instead.
1472 * journalctl's and systemctl's -o switch now knows a new log output
1473 mode "with-unit". The output it generates is very similar to the
1474 regular "short" mode, but displays the unit name instead of the
1475 syslog tag for each log line. Also, the date is shown with timezone
1476 information. This mode is probably more useful than the classic
1477 "short" output mode for most purposes, except where pixel-perfect
1478 compatibility with classic /var/log/messages formatting is required.
1480 * A new --dump-bus-properties switch has been added to the systemd
1481 binary, which may be used to dump all supported D-Bus properties.
1482 (Options which are still supported, but are deprecated, are *not*
1485 * sd-bus gained a set of new calls:
1486 sd_bus_slot_set_floating()/sd_bus_slot_get_floating() may be used to
1487 enable/disable the "floating" state of a bus slot object,
1488 i.e. whether the slot object pins the bus it is allocated for into
1489 memory or if the bus slot object gets disconnected when the bus goes
1490 away. sd_bus_open_with_description(),
1491 sd_bus_open_user_with_description(),
1492 sd_bus_open_system_with_description() may be used to allocate bus
1493 objects and set their description string already during allocation.
1495 * sd-event gained support for watching inotify events from the event
1496 loop, in an efficient way, sharing inotify handles between multiple
1497 users. For this a new function sd_event_add_inotify() has been added.
1499 * sd-event and sd-bus gained support for calling special user-supplied
1500 destructor functions for userdata pointers associated with
1501 sd_event_source, sd_bus_slot, and sd_bus_track objects. For this new
1502 functions sd_bus_slot_set_destroy_callback,
1503 sd_bus_slot_get_destroy_callback, sd_bus_track_set_destroy_callback,
1504 sd_bus_track_get_destroy_callback,
1505 sd_event_source_set_destroy_callback,
1506 sd_event_source_get_destroy_callback have been added.
1508 * The "net.ipv4.tcp_ecn" sysctl will now be turned on by default.
1510 * PID 1 will now automatically reschedule .timer units whenever the
1511 local timezone changes. (They previously got rescheduled
1512 automatically when the system clock changed.)
1514 * New documentation has been added to document cgroups delegation,
1515 portable services and the various code quality tools we have set up:
1517 https://github.com/systemd/systemd/blob/master/docs/CGROUP_DELEGATION.md
1518 https://github.com/systemd/systemd/blob/master/docs/PORTABLE_SERVICES.md
1519 https://github.com/systemd/systemd/blob/master/docs/CODE_QUALITY.md
1521 * The Boot Loader Specification has been added to the source tree.
1523 https://github.com/systemd/systemd/blob/master/docs/BOOT_LOADER_SPECIFICATION.md
1525 While moving it into our source tree we have updated it and further
1526 changes are now accepted through the usual github PR workflow.
1528 * pam_systemd will now look for PAM userdata fields systemd.memory_max,
1529 systemd.tasks_max, systemd.cpu_weight, systemd.io_weight set by
1530 earlier PAM modules. The data in these fields is used to initialize
1531 the session scope's resource properties. Thus external PAM modules
1532 may now configure per-session limits, for example sourced from
1533 external user databases.
1535 * socket units with Accept=yes will now maintain a "refused" counter in
1536 addition to the existing "accepted" counter, counting connections
1537 refused due to the enforced limits.
1539 * The "systemd-path search-binaries-default" command may now be use to
1540 query the default, built-in $PATH PID 1 will pass to the services it
1543 * A new unit file setting PrivateMounts= has been added. It's a boolean
1544 option. If enabled the unit's processes are invoked in their own file
1545 system namespace. Note that this behaviour is also implied if any
1546 other file system namespacing options (such as PrivateTmp=,
1547 PrivateDevices=, ProtectSystem=, …) are used. This option is hence
1548 primarily useful for services that do not use any of the other file
1549 system namespacing options. One such service is systemd-udevd.service
1550 where this is now used by default.
1552 * ConditionSecurity= gained a new value "uefi-secureboot" that is true
1553 when the system is booted in UEFI "secure mode".
1555 * A new unit "system-update-pre.target" is added, which defines an
1556 optional synchronization point for offline system updates, as
1557 implemented by the pre-existing "system-update.target" unit. It
1558 allows ordering services before the service that executes the actual
1559 update process in a generic way.
1561 * Systemd now emits warnings whenever .include syntax is used.
1563 Contributions from: Adam Duskett, Alan Jenkins, Alessandro Casale,
1564 Alexander Kurtz, Alex Gartrell, Anssi Hannula, Arnaud Rebillout, Brian
1565 J. Murrell, Bruno Vernay, Chris Lamb, Chris Lesiak, Christian Brauner,
1566 Christian Hesse, Christian Rebischke, Colin Guthrie, Daniel Dao, Daniel
1567 Lin, Danylo Korostil, Davide Cavalca, David Tardon, Dimitri John
1568 Ledkov, Dmitriy Geels, Douglas Christman, Elia Geretto, emelenas, Emil
1569 Velikov, Evgeny Vereshchagin, Felipe Sateler, Feng Sun, Filipe
1570 Brandenburger, Franck Bui, futpib, Giuseppe Scrivano, Guillem Jover,
1571 guixxx, Hannes Reinecke, Hans de Goede, Harald Hoyer, Henrique Dante de
1572 Almeida, Hiram van Paassen, Ian Miell, Igor Gnatenko, Ivan Shapovalov,
1573 Iwan Timmer, James Cowgill, Jan Janssen, Jan Synacek, Jared Kazimir,
1574 Jérémy Rosen, João Paulo Rechi Vita, Joost Heitbrink, Jui-Chi Ricky
1575 Liang, Jürg Billeter, Kai-Heng Feng, Karol Augustin, Kay Sievers,
1576 Krzysztof Nowicki, Lauri Tirkkonen, Lennart Poettering, Leonard König,
1577 Long Li, Luca Boccassi, Lucas Werkmeister, Marcel Hoppe, Marc
1578 Kleine-Budde, Mario Limonciello, Martin Jansa, Martin Wilck, Mathieu
1579 Malaterre, Matteo F. Vescovi, Matthew McGinn, Matthias-Christian Ott,
1580 Michael Biebl, Michael Olbrich, Michael Prokop, Michal Koutný, Michal
1581 Sekletar, Mike Gilbert, Mikhail Kasimov, Milan Broz, Milan Pässler,
1582 Mladen Pejaković, Muhammet Kara, Nicolas Boichat, Omer Katz, Paride
1583 Legovini, Paul Menzel, Paul Milliken, Pavel Hrdina, Peter A. Bigot,
1584 Peter D'Hoye, Peter Hutterer, Peter Jones, Philip Sequeira, Philip
1585 Withnall, Piotr Drąg, Radostin Stoyanov, Ricardo Salveti de Araujo,
1586 Ronny Chevalier, Rosen Penev, Rubén Suárez Alvarez, Ryan Gonzalez,
1587 Salvo Tomaselli, Sebastian Reichel, Sergey Ptashnick, Sergio Lindo
1588 Mansilla, Stefan Schweter, Stephen Hemminger, Stuart Hayes, Susant
1589 Sahani, Sylvain Plantefève, Thomas H. P. Andersen, Tobias Jungel,
1590 Tomasz Torcz, Vito Caputo, Will Dietz, Will Thompson, Wim van Mourik,
1591 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
1593 — Berlin, 2018-06-22
1597 * The MemoryAccounting= unit property now defaults to on. After
1598 discussions with the upstream control group maintainers we learnt
1599 that the negative impact of cgroup memory accounting on current
1600 kernels is finally relatively minimal, so that it should be safe to
1601 enable this by default without affecting system performance. Besides
1602 memory accounting only task accounting is turned on by default, all
1603 other forms of resource accounting (CPU, IO, IP) remain off for now,
1604 because it's not clear yet that their impact is small enough to move
1605 from opt-in to opt-out. We recommend downstreams to leave memory
1606 accounting on by default if kernel 4.14 or higher is primarily
1607 used. On very resource constrained systems or when support for old
1608 kernels is a necessity, -Dmemory-accounting-default=false can be used
1609 to revert this change.
1611 * rpm scriptlets to update the udev hwdb and rules (%udev_hwdb_update,
1612 %udev_rules_update) and the journal catalog (%journal_catalog_update)
1613 from the upgrade scriptlets of individual packages now do nothing.
1614 Transfiletriggers have been added which will perform those updates
1615 once at the end of the transaction.
1617 Similar transfiletriggers have been added to execute any sysctl.d
1618 and binfmt.d rules. Thus, it should be unnecessary to provide any
1619 scriptlets to execute this configuration from package installation
1622 * systemd-sysusers gained a mode where the configuration to execute is
1623 specified on the command line, but this configuration is not executed
1624 directly, but instead it is merged with the configuration on disk,
1625 and the result is executed. This is useful for package installation
1626 scripts which want to create the user before installing any files on
1627 disk (in case some of those files are owned by that user), while
1628 still allowing local admin overrides.
1630 This functionality is exposed to rpm scriptlets through a new
1631 %sysusers_create_package macro. Old %sysusers_create and
1632 %sysusers_create_inline macros are deprecated.
1634 A transfiletrigger for sysusers.d configuration is now installed,
1635 which means that it should be unnecessary to call systemd-sysusers from
1636 package installation scripts, unless the package installs any files
1637 owned by those newly-created users, in which case
1638 %sysusers_create_package should be used.
1640 * Analogous change has been done for systemd-tmpfiles: it gained a mode
1641 where the command-line configuration is merged with the configuration
1642 on disk. This is exposed as the new %tmpfiles_create_package macro,
1643 and %tmpfiles_create is deprecated. A transfiletrigger is installed
1644 for tmpfiles.d, hence it should be unnecessary to call systemd-tmpfiles
1645 from package installation scripts.
1647 * sysusers.d configuration for a user may now also specify the group
1648 number, in addition to the user number ("u username 123:456"), or
1649 without the user number ("u username -:456").
1651 * Configution items for systemd-sysusers can now be specified as
1652 positional arguments when the new --inline switch is used.
1654 * The login shell of users created through sysusers.d may now be
1655 specified (previously, it was always /bin/sh for root and
1656 /sbin/nologin for other users).
1658 * systemd-analyze gained a new --global switch to look at global user
1659 configuration. It also gained a unit-paths verb to list the unit load
1660 paths that are compiled into systemd (which can be used with
1661 --systemd, --user, or --global).
1663 * udevadm trigger gained a new --settle/-w option to wait for any
1664 triggered events to finish (but just those, and not any other events
1665 which are triggered meanwhile).
1667 * The action that systemd-logind takes when the lid is closed and the
1668 machine is connected to external power can now be configured using
1669 HandleLidSwitchExternalPower= in logind.conf. Previously, this action
1670 was determined by HandleLidSwitch=, and, for backwards compatibility,
1671 is still is, if HandleLidSwitchExternalPower= is not explicitly set.
1673 * journalctl will periodically call sd_journal_process() to make it
1674 resilient against inotify queue overruns when journal files are
1675 rotated very quickly.
1677 * Two new functions in libsystemd — sd_bus_get_n_queued_read and
1678 sd_bus_get_n_queued_write — may be used to check the number of
1679 pending bus messages.
1681 * systemd gained a new
1682 org.freedesktop.systemd1.Manager.AttachProcessesToUnit dbus call
1683 which can be used to migrate foreign processes to scope and service
1684 units. The primary user for this new API is systemd itself: the
1685 systemd --user instance uses this call of the systemd --system
1686 instance to migrate processes if it itself gets the request to
1687 migrate processes and the kernel refuses this due to access
1688 restrictions. Thanks to this "systemd-run --scope --user …" works
1689 again in pure cgroup v2 environments when invoked from the user
1692 * A new TemporaryFileSystem= setting can be used to mask out part of
1693 the real file system tree with tmpfs mounts. This may be combined
1694 with BindPaths= and BindReadOnlyPaths= to hide files or directories
1695 not relevant to the unit, while still allowing some paths lower in
1696 the tree to be accessed.
1698 ProtectHome=tmpfs may now be used to hide user home and runtime
1699 directories from units, in a way that is mostly equivalent to
1700 "TemporaryFileSystem=/home /run/user /root".
1702 * Non-service units are now started with KeyringMode=shared by default.
1703 This means that mount and swapon and other mount tools have access
1704 to keys in the main keyring.
1706 * /sys/fs/bpf is now mounted automatically.
1708 * QNX virtualization is now detected by systemd-detect-virt and may
1709 be used in ConditionVirtualization=.
1711 * IPAccounting= may now be enabled also for slice units.
1713 * A new -Dsplit-bin= build configuration switch may be used to specify
1714 whether bin and sbin directories are merged, or if they should be
1715 included separately in $PATH and various listings of executable
1716 directories. The build configuration scripts will try to autodetect
1717 the proper values of -Dsplit-usr= and -Dsplit-bin= based on build
1718 system, but distributions are encouraged to configure this
1721 * A new -Dok-color= build configuration switch may be used to change
1722 the colour of "OK" status messages.
1724 * UPGRADE ISSUE: serialization of units using JoinsNamespaceOf= with
1725 PrivateNetwork=yes was buggy in previous versions of systemd. This
1726 means that after the upgrade and daemon-reexec, any such units must
1729 * INCOMPATIBILITY: as announced in the NEWS for 237, systemd-tmpfiles
1730 will not exclude read-only files owned by root from cleanup.
1732 Contributions from: Alan Jenkins, Alexander F Rødseth, Alexis Jeandet,
1733 Andika Triwidada, Andrei Gherzan, Ansgar Burchardt, antizealot1337,
1734 Batuhan Osman Taşkaya, Beniamino Galvani, Bill Yodlowsky, Caio Marcelo
1735 de Oliveira Filho, CuBiC, Daniele Medri, Daniel Mouritzen, Daniel
1736 Rusek, Davide Cavalca, Dimitri John Ledkov, Douglas Christman, Evgeny
1737 Vereshchagin, Faalagorn, Filipe Brandenburger, Franck Bui, futpib,
1738 Giacomo Longo, Gunnar Hjalmarsson, Hans de Goede, Hermann Gausterer,
1739 Iago López Galeiras, Jakub Filak, Jan Synacek, Jason A. Donenfeld,
1740 Javier Martinez Canillas, Jérémy Rosen, Lennart Poettering, Lucas
1741 Werkmeister, Mao Huang, Marco Gulino, Michael Biebl, Michael Vogt,
1742 MilhouseVH, Neal Gompa (ニール・ゴンパ), Oleander Reis, Olof Mogren,
1743 Patrick Uiterwijk, Peter Hutterer, Peter Portante, Piotr Drąg, Robert
1744 Antoni Buj Gelonch, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
1745 Fowler, SjonHortensius, snorreflorre, Susant Sahani, Sylvain
1746 Plantefève, Thomas Blume, Thomas Haller, Vito Caputo, Yu Watanabe,
1747 Zbigniew Jędrzejewski-Szmek, Марко М. Костић (Marko M. Kostić)
1749 — Warsaw, 2018-03-05
1753 * Some keyboards come with a zoom see-saw or rocker which until now got
1754 mapped to the Linux "zoomin/out" keys in hwdb. However, these
1755 keycodes are not recognized by any major desktop. They now produce
1756 Up/Down key events so that they can be used for scrolling.
1758 * INCOMPATIBILITY: systemd-tmpfiles' "f" lines changed behaviour
1759 slightly: previously, if an argument was specified for lines of this
1760 type (i.e. the right-most column was set) this string was appended to
1761 existing files each time systemd-tmpfiles was run. This behaviour was
1762 different from what the documentation said, and not particularly
1763 useful, as repeated systemd-tmpfiles invocations would not be
1764 idempotent and grow such files without bounds. With this release
1765 behaviour has been altered to match what the documentation says:
1766 lines of this type only have an effect if the indicated files don't
1767 exist yet, and only then the argument string is written to the file.
1769 * FUTURE INCOMPATIBILITY: In systemd v238 we intend to slightly change
1770 systemd-tmpfiles behaviour: previously, read-only files owned by root
1771 were always excluded from the file "aging" algorithm (i.e. the
1772 automatic clean-up of directories like /tmp based on
1773 atime/mtime/ctime). We intend to drop this restriction, and age files
1774 by default even when owned by root and read-only. This behaviour was
1775 inherited from older tools, but there have been requests to remove
1776 it, and it's not obvious why this restriction was made in the first
1777 place. Please speak up now, if you are aware of software that reqires
1778 this behaviour, otherwise we'll remove the restriction in v238.
1780 * A new environment variable $SYSTEMD_OFFLINE is now understood by
1781 systemctl. It takes a boolean argument. If on, systemctl assumes it
1782 operates on an "offline" OS tree, and will not attempt to talk to the
1783 service manager. Previously, this mode was implicitly enabled if a
1784 chroot() environment was detected, and this new environment variable
1785 now provides explicit control.
1787 * .path and .socket units may now be created transiently, too.
1788 Previously only service, mount, automount and timer units were
1789 supported as transient units. The systemd-run tool has been updated
1790 to expose this new functionality, you may hence use it now to bind
1791 arbitrary commands to path or socket activation on-the-fly from the
1792 command line. Moreover, almost all properties are now exposed for the
1793 unit types that already supported transient operation.
1795 * The systemd-mount command gained support for a new --owner= parameter
1796 which takes a user name, which is then resolved and included in uid=
1797 and gid= mount options string of the file system to mount.
1799 * A new unit condition ConditionControlGroupController= has been added
1800 that checks whether a specific cgroup controller is available.
1802 * Unit files, udev's .link files, and systemd-networkd's .netdev and
1803 .network files all gained support for a new condition
1804 ConditionKernelVersion= for checking against specific kernel
1807 * In systemd-networkd, the [IPVLAN] section in .netdev files gained
1808 support for configuring device flags in the Flags= setting. In the
1809 same files, the [Tunnel] section gained support for configuring
1810 AllowLocalRemote=. The [Route] section in .network files gained
1811 support for configuring InitialCongestionWindow=,
1812 InitialAdvertisedReceiveWindow= and QuickAck=. The [DHCP] section now
1813 understands RapidCommit=.
1815 * systemd-networkd's DHCPv6 support gained support for Prefix
1818 * sd-bus gained support for a new "watch-bind" feature. When this
1819 feature is enabled, an sd_bus connection may be set up to connect to
1820 an AF_UNIX socket in the file system as soon as it is created. This
1821 functionality is useful for writing early-boot services that
1822 automatically connect to the system bus as soon as it is started,
1823 without ugly time-based polling. systemd-networkd and
1824 systemd-resolved have been updated to make use of this
1825 functionality. busctl exposes this functionality in a new
1826 --watch-bind= command line switch.
1828 * sd-bus will now optionally synthesize a local "Connected" signal as
1829 soon as a D-Bus connection is set up fully. This message mirrors the
1830 already existing "Disconnected" signal which is synthesized when the
1831 connection is terminated. This signal is generally useful but
1832 particularly handy in combination with the "watch-bind" feature
1833 described above. Synthesizing of this message has to be requested
1834 explicitly through the new API call sd_bus_set_connected_signal(). In
1835 addition a new call sd_bus_is_ready() has been added that checks
1836 whether a connection is fully set up (i.e. between the "Connected" and
1837 "Disconnected" signals).
1839 * sd-bus gained two new calls sd_bus_request_name_async() and
1840 sd_bus_release_name_async() for asynchronously registering bus
1841 names. Similar, there is now sd_bus_add_match_async() for installing
1842 a signal match asynchronously. All of systemd's own services have
1843 been updated to make use of these calls. Doing these operations
1844 asynchronously has two benefits: it reduces the risk of deadlocks in
1845 case of cyclic dependencies between bus services, and it speeds up
1846 service initialization since synchronization points for bus
1847 round-trips are removed.
1849 * sd-bus gained two new calls sd_bus_match_signal() and
1850 sd_bus_match_signal_async(), which are similar to sd_bus_add_match()
1851 and sd_bus_add_match_async() but instead of taking a D-Bus match
1852 string take match fields as normal function parameters.
1854 * sd-bus gained two new calls sd_bus_set_sender() and
1855 sd_bus_message_set_sender() for setting the sender name of outgoing
1856 messages (either for all outgoing messages or for just one specific
1857 one). These calls are only useful in direct connections as on
1858 brokered connections the broker fills in the sender anyway,
1859 overwriting whatever the client filled in.
1861 * sd-event gained a new pseudo-handle that may be specified on all API
1862 calls where an "sd_event*" object is expected: SD_EVENT_DEFAULT. When
1863 used this refers to the default event loop object of the calling
1864 thread. Note however that this does not implicitly allocate one —
1865 which has to be done prior by using sd_event_default(). Similarly
1866 sd-bus gained three new pseudo-handles SD_BUS_DEFAULT,
1867 SD_BUS_DEFAULT_USER, SD_BUS_DEFAULT_SYSTEM that may be used to refer
1868 to the default bus of the specified type of the calling thread. Here
1869 too this does not implicitly allocate bus connection objects, this
1870 has to be done prior with sd_bus_default() and friends.
1872 * sd-event gained a new call pair
1873 sd_event_source_{get|set}_io_fd_own(). This may be used to request
1874 automatic closure of the file descriptor an IO event source watches
1875 when the event source is destroyed.
1877 * systemd-networkd gained support for natively configuring WireGuard
1880 * In previous versions systemd synthesized user records both for the
1881 "nobody" (UID 65534) and "root" (UID 0) users in nss-systemd and
1882 internally. In order to simplify distribution-wide renames of the
1883 "nobody" user (like it is planned in Fedora: nfsnobody → nobody), a
1884 new transitional flag file has been added: if
1885 /etc/systemd/dont-synthesize-nobody exists synthesizing of the 65534
1886 user and group record within the systemd codebase is disabled.
1888 * systemd-notify gained a new --uid= option for selecting the source
1889 user/UID to use for notification messages sent to the service
1892 * journalctl gained a new --grep= option to list only entries in which
1893 the message matches a certain pattern. By default matching is case
1894 insensitive if the pattern is lowercase, and case sensitive
1895 otherwise. Option --case-sensitive=yes|no can be used to override
1896 this an specify case sensitivity or case insensitivity.
1898 * There's now a "systemd-analyze service-watchdogs" command for printing
1899 the current state of the service runtime watchdog, and optionally
1900 enabling or disabling the per-service watchdogs system-wide if given a
1901 boolean argument (i.e. the concept you configure in WatchdogSec=), for
1902 debugging purposes. There's also a kernel command line option
1903 systemd.service_watchdogs= for controlling the same.
1905 * Two new "log-level" and "log-target" options for systemd-analyze were
1906 added that merge the now deprecated get-log-level, set-log-level and
1907 get-log-target, set-log-target pairs. The deprecated options are still
1908 understood for backwards compatibility. The two new options print the
1909 current value when no arguments are given, and set them when a
1910 level/target is given as an argument.
1912 * sysusers.d's "u" lines now optionally accept both a UID and a GID
1913 specification, separated by a ":" character, in order to create users
1914 where UID and GID do not match.
1916 Contributions from: Adam Duskett, Alan Jenkins, Alexander Kuleshov,
1917 Alexis Deruelle, Andrew Jeddeloh, Armin Widegreen, Batuhan Osman
1918 Taşkaya, Björn Esser, bleep_blop, Bruce A. Johnson, Chris Down, Clinton
1919 Roy, Colin Walters, Daniel Rusek, Dimitri John Ledkov, Dmitry Rozhkov,
1920 Evgeny Vereshchagin, Ewout van Mansom, Felipe Sateler, Franck Bui,
1921 Frantisek Sumsal, George Gaydarov, Gianluca Boiano, Hans-Christian
1922 Noren Egtvedt, Hans de Goede, Henrik Grindal Bakken, Jan Alexander
1923 Steffens, Jan Klötzke, Jason A. Donenfeld, jdkbx, Jérémy Rosen,
1924 Jerónimo Borque, John Lin, John Paul Herold, Jonathan Rudenberg, Jörg
1925 Thalheim, Ken (Bitsko) MacLeod, Larry Bernstone, Lennart Poettering,
1926 Lucas Werkmeister, Maciej S. Szmigiero, Marek Čermák, Martin Pitt,
1927 Mathieu Malaterre, Matthew Thode, Matthias-Christian Ott, Max Harmathy,
1928 Michael Biebl, Michael Vogt, Michal Koutný, Michal Sekletar, Michał
1929 Szczepański, Mike Gilbert, Nathaniel McCallum, Nicolas Chauvet, Olaf
1930 Hering, Olivier Schwander, Patrik Flykt, Paul Cercueil, Peter Hutterer,
1931 Piotr Drąg, Raphael Vogelgsang, Reverend Homer, Robert Kolchmeyer,
1932 Samuel Dionne-Riel, Sergey Ptashnick, Shawn Landden, Susant Sahani,
1933 Sylvain Plantefève, Thomas H. P. Andersen, Thomas Huth, Tomasz
1934 Bachorski, Vladislav Vishnyakov, Wieland Hoffmann, Yu Watanabe, Zachary
1935 Winnerman, Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски, Дилян
1942 * The modprobe.d/ drop-in for the bonding.ko kernel module introduced
1943 in v235 has been extended to also set the dummy.ko module option
1944 numdummies=0, preventing the kernel from automatically creating
1945 dummy0. All dummy interfaces must now be explicitly created.
1947 * Unknown '%' specifiers in configuration files are now rejected. This
1948 applies to units and tmpfiles.d configuration. Any percent characters
1949 that are followed by a letter or digit that are not supposed to be
1950 interpreted as the beginning of a specifier should be escaped by
1951 doubling ("%%"). (So "size=5%" is still accepted, as well as
1952 "size=5%,foo=bar", but not "LABEL=x%y%z" since %y and %z are not
1953 valid specifiers today.)
1955 * systemd-resolved now maintains a new dynamic
1956 /run/systemd/resolve/stub-resolv.conf compatibility file. It is
1957 recommended to make /etc/resolv.conf a symlink to it. This file
1958 points at the systemd-resolved stub DNS 127.0.0.53 resolver and
1959 includes dynamically acquired search domains, achieving more correct
1960 DNS resolution by software that bypasses local DNS APIs such as NSS.
1962 * The "uaccess" udev tag has been dropped from /dev/kvm and
1963 /dev/dri/renderD*. These devices now have the 0666 permissions by
1964 default (but this may be changed at build-time). /dev/dri/renderD*
1965 will now be owned by the "render" group along with /dev/kfd.
1967 * "DynamicUser=yes" has been enabled for systemd-timesyncd.service,
1968 systemd-journal-gatewayd.service and
1969 systemd-journal-upload.service. This means "nss-systemd" must be
1970 enabled in /etc/nsswitch.conf to ensure the UIDs assigned to these
1971 services are resolved properly.
1973 * In /etc/fstab two new mount options are now understood:
1974 x-systemd.makefs and x-systemd.growfs. The former has the effect that
1975 the configured file system is formatted before it is mounted, the
1976 latter that the file system is resized to the full block device size
1977 after it is mounted (i.e. if the file system is smaller than the
1978 partition it resides on, it's grown). This is similar to the fsck
1979 logic in /etc/fstab, and pulls in systemd-makefs@.service and
1980 systemd-growfs@.service as necessary, similar to
1981 systemd-fsck@.service. Resizing is currently only supported on ext4
1984 * In systemd-networkd, the IPv6 RA logic now optionally may announce
1985 DNS server and domain information.
1987 * Support for the LUKS2 on-disk format for encrypted partitions has
1988 been added. This requires libcryptsetup2 during compilation and
1991 * The systemd --user instance will now signal "readiness" when its
1992 basic.target unit has been reached, instead of when the run queue ran
1993 empty for the first time.
1995 * Tmpfiles.d with user configuration are now also supported.
1996 systemd-tmpfiles gained a new --user switch, and snippets placed in
1997 ~/.config/user-tmpfiles.d/ and corresponding directories will be
1998 executed by systemd-tmpfiles --user running in the new
1999 systemd-tmpfiles-setup.service and systemd-tmpfiles-clean.service
2000 running in the user session.
2002 * Unit files and tmpfiles.d snippets learnt three new % specifiers:
2003 %S resolves to the top-level state directory (/var/lib for the system
2004 instance, $XDG_CONFIG_HOME for the user instance), %C resolves to the
2005 top-level cache directory (/var/cache for the system instance,
2006 $XDG_CACHE_HOME for the user instance), %L resolves to the top-level
2007 logs directory (/var/log for the system instance,
2008 $XDG_CONFIG_HOME/log/ for the user instance). This matches the
2009 existing %t specifier, that resolves to the top-level runtime
2010 directory (/run for the system instance, and $XDG_RUNTIME_DIR for the
2013 * journalctl learnt a new parameter --output-fields= for limiting the
2014 set of journal fields to output in verbose and JSON output modes.
2016 * systemd-timesyncd's configuration file gained a new option
2017 RootDistanceMaxSec= for setting the maximum root distance of servers
2018 it'll use, as well as the new options PollIntervalMinSec= and
2019 PollIntervalMaxSec= to tweak the minimum and maximum poll interval.
2021 * bootctl gained a new command "list" for listing all available boot
2022 menu items on systems that follow the boot loader specification.
2024 * systemctl gained a new --dry-run switch that shows what would be done
2025 instead of doing it, and is currently supported by the shutdown and
2028 * ConditionSecurity= can now detect the TOMOYO security module.
2030 * Unit file [Install] sections are now also respected in unit drop-in
2031 files. This is intended to be used by drop-ins under /usr/lib/.
2033 * systemd-firstboot may now also set the initial keyboard mapping.
2035 * Udev "changed" events for devices which are exposed as systemd
2036 .device units are now propagated to units specified in
2037 ReloadPropagatedFrom= as reload requests.
2039 * If a udev device has a SYSTEMD_WANTS= property containing a systemd
2040 unit template name (i.e. a name in the form of 'foobar@.service',
2041 without the instance component between the '@' and - the '.'), then
2042 the escaped sysfs path of the device is automatically used as the
2045 * SystemCallFilter= in unit files has been extended so that an "errno"
2046 can be specified individually for each system call. Example:
2047 SystemCallFilter=~uname:EILSEQ.
2049 * The cgroup delegation logic has been substantially updated. Delegate=
2050 now optionally takes a list of controllers (instead of a boolean, as
2051 before), which lists the controllers to delegate at least.
2053 * The networkd DHCPv6 client now implements the FQDN option (RFC 4704).
2055 * A new LogLevelMax= setting configures the maximum log level any
2056 process of the service may log at (i.e. anything with a lesser
2057 priority than what is specified is automatically dropped). A new
2058 LogExtraFields= setting allows configuration of additional journal
2059 fields to attach to all log records generated by any of the unit's
2062 * New StandardInputData= and StandardInputText= settings along with the
2063 new option StandardInput=data may be used to configure textual or
2064 binary data that shall be passed to the executed service process via
2065 standard input, encoded in-line in the unit file.
2067 * StandardInput=, StandardOutput= and StandardError= may now be used to
2068 connect stdin/stdout/stderr of executed processes directly with a
2069 file or AF_UNIX socket in the file system, using the new "file:" option.
2071 * A new unit file option CollectMode= has been added, that allows
2072 tweaking the garbage collection logic for units. It may be used to
2073 tell systemd to garbage collect units that have failed automatically
2074 (normally it only GCs units that exited successfully). systemd-run
2075 and systemd-mount expose this new functionality with a new -G option.
2077 * "machinectl bind" may now be used to bind mount non-directories
2078 (i.e. regularfiles, devices, fifos, sockets).
2080 * systemd-analyze gained a new verb "calendar" for validating and
2081 testing calendar time specifications to use for OnCalendar= in timer
2082 units. Besides validating the expression it will calculate the next
2083 time the specified expression would elapse.
2085 * In addition to the pre-existing FailureAction= unit file setting
2086 there's now SuccessAction=, for configuring a shutdown action to
2087 execute when a unit completes successfully. This is useful in
2088 particular inside containers that shall terminate after some workload
2089 has been completed. Also, both options are now supported for all unit
2090 types, not just services.
2092 * networkds's IP rule support gained two new options
2093 IncomingInterface= and OutgoingInterface= for configuring the incoming
2094 and outgoing interfaces of configured rules. systemd-networkd also
2095 gained support for "vxcan" network devices.
2097 * networkd gained a new setting RequiredForOnline=, taking a
2098 boolean. If set, systemd-wait-online will take it into consideration
2099 when determining that the system is up, otherwise it will ignore the
2100 interface for this purpose.
2102 * The sd_notify() protocol gained support for a new operation: with
2103 FDSTOREREMOVE=1 file descriptors may be removed from the per-service
2104 store again, ahead of POLLHUP or POLLERR when they are removed
2107 * A new document doc/UIDS-GIDS.md has been added to the source tree,
2108 that documents the UID/GID range and assignment assumptions and
2109 requirements of systemd.
2111 * The watchdog device PID 1 will ping may now be configured through the
2112 WatchdogDevice= configuration file setting, or by setting the
2113 systemd.watchdog_service= kernel commandline option.
2115 * systemd-resolved's gained support for registering DNS-SD services on
2116 the local network using MulticastDNS. Services may either be
2117 registered by dropping in a .dnssd file in /etc/systemd/dnssd/ (or
2118 the same dir below /run, /usr/lib), or through its D-Bus API.
2120 * The sd_notify() protocol can now with EXTEND_TIMEOUT_USEC=microsecond
2121 extend the effective start, runtime, and stop time. The service must
2122 continue to send EXTEND_TIMEOUT_USEC within the period specified to
2123 prevent the service manager from making the service as timedout.
2125 * systemd-resolved's DNSSEC support gained support for RFC 8080
2126 (Ed25519 keys and signatures).
2128 * The systemd-resolve command line tool gained a new set of options
2129 --set-dns=, --set-domain=, --set-llmnr=, --set-mdns=, --set-dnssec=,
2130 --set-nta= and --revert to configure per-interface DNS configuration
2131 dynamically during runtime. It's useful for pushing DNS information
2132 into systemd-resolved from DNS hook scripts that various interface
2133 managing software supports (such as pppd).
2135 * systemd-nspawn gained a new --network-namespace-path= command line
2136 option, which may be used to make a container join an existing
2137 network namespace, by specifying a path to a "netns" file.
2139 Contributions from: Alan Jenkins, Alan Robertson, Alessandro Ghedini,
2140 Andrew Jeddeloh, Antonio Rojas, Ari, asavah, bleep_blop, Carsten
2141 Strotmann, Christian Brauner, Christian Hesse, Clinton Roy, Collin
2142 Eggert, Cong Wang, Daniel Black, Daniel Lockyer, Daniel Rusek, Dimitri
2143 John Ledkov, Dmitry Rozhkov, Dongsu Park, Edward A. James, Evgeny
2144 Vereshchagin, Florian Klink, Franck Bui, Gwendal Grignou, Hans de
2145 Goede, Harald Hoyer, Hristo Venev, Iago López Galeiras, Ikey Doherty,
2146 Jakub Wilk, Jérémy Rosen, Jiahui Xie, John Lin, José Bollo, Josef
2147 Andersson, juga0, Krzysztof Nowicki, Kyle Walker, Lars Karlitski, Lars
2148 Kellogg-Stedman, Lauri Tirkkonen, Lennart Poettering, Lubomir Rintel,
2149 Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn, Lukáš Říha, Lukasz
2150 Rubaszewski, Maciej S. Szmigiero, Mantas Mikulėnas, Marcus Folkesson,
2151 Martin Steuer, Mathieu Trudel-Lapierre, Matija Skala,
2152 Matthias-Christian Ott, Max Resch, Michael Biebl, Michael Vogt, Michal
2153 Koutný, Michal Sekletar, Mike Gilbert, Muhammet Kara, Neil Brown, Olaf
2154 Hering, Ondrej Kozina, Patrik Flykt, Patryk Kocielnik, Peter Hutterer,
2155 Piotr Drąg, Razvan Cojocaru, Robin McCorkell, Roland Hieber, Saran
2156 Tunyasuvunakool, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
2157 Arlott, Simon Peeters, Stanislav Angelovič, Stefan Agner, Susant
2158 Sahani, Sylvain Plantefève, Thomas Blume, Thomas Haller, Tiago Salem
2159 Herrmann, Tinu Weber, Tom Stellard, Topi Miettinen, Torsten Hilbrich,
2160 Vito Caputo, Vladislav Vishnyakov, WaLyong Cho, Yu Watanabe, Zbigniew
2161 Jędrzejewski-Szmek, Zeal Jagannatha
2163 — Berlin, 2017-12-14
2167 * INCOMPATIBILITY: systemd-logind.service and other long-running
2168 services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP
2169 communication with the outside. This generally improves security of
2170 the system, and is in almost all cases a safe and good choice, as
2171 these services do not and should not provide any network-facing
2172 functionality. However, systemd-logind uses the glibc NSS API to
2173 query the user database. This creates problems on systems where NSS
2174 is set up to directly consult network services for user database
2175 lookups. In particular, this creates incompatibilities with the
2176 "nss-nis" module, which attempts to directly contact the NIS/YP
2177 network servers it is configured for, and will now consistently
2178 fail. In such cases, it is possible to turn off IP sandboxing for
2179 systemd-logind.service (set IPAddressDeny= in its [Service] section
2180 to the empty string, via a .d/ unit file drop-in). Downstream
2181 distributions might want to update their nss-nis packaging to include
2182 such a drop-in snippet, accordingly, to hide this incompatibility
2183 from the user. Another option is to make use of glibc's nscd service
2184 to proxy such network requests through a privilege-separated, minimal
2185 local caching daemon, or to switch to more modern technologies such
2186 sssd, whose NSS hook-ups generally do not involve direct network
2187 access. In general, we think it's definitely time to question the
2188 implementation choices of nss-nis, i.e. whether it's a good idea
2189 today to embed a network-facing loadable module into all local
2190 processes that need to query the user database, including the most
2191 trivial and benign ones, such as "ls". For more details about
2192 IPAddressDeny= see below.
2194 * A new modprobe.d drop-in is now shipped by default that sets the
2195 bonding module option max_bonds=0. This overrides the kernel default,
2196 to avoid conflicts and ambiguity as to whether or not bond0 should be
2197 managed by systemd-networkd or not. This resolves multiple issues
2198 with bond0 properties not being applied, when bond0 is configured
2199 with systemd-networkd. Distributors may choose to not package this,
2200 however in that case users will be prevented from correctly managing
2201 bond0 interface using systemd-networkd.
2203 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
2204 which print the logging level and target of the system manager. They
2205 complement the existing "set-log-level" and "set-log-target" verbs
2206 used to change those values.
2208 * journald.conf gained a new boolean setting ReadKMsg= which defaults
2209 to on. If turned off kernel log messages will not be read by
2210 systemd-journald or included in the logs. It also gained a new
2211 setting LineMax= for configuring the maximum line length in
2212 STDOUT/STDERR log streams. The new default for this value is 48K, up
2213 from the previous hardcoded 2048.
2215 * A new unit setting RuntimeDirectoryPreserve= has been added, which
2216 allows more detailed control of what to do with a runtime directory
2217 configured with RuntimeDirectory= (i.e. a directory below /run or
2218 $XDG_RUNTIME_DIR) after a unit is stopped.
2220 * The RuntimeDirectory= setting for units gained support for creating
2221 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
2222 one top-level directory.
2224 * Units gained new options StateDirectory=, CacheDirectory=,
2225 LogsDirectory= and ConfigurationDirectory= which are closely related
2226 to RuntimeDirectory= but manage per-service directories below
2227 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
2228 possible to write unit files which when activated automatically gain
2229 properly owned service specific directories in these locations, thus
2230 making unit files self-contained and increasing compatibility with
2231 stateless systems and factory reset where /etc or /var are
2232 unpopulated at boot. Matching these new settings there's also
2233 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
2234 ConfigurationDirectoryMode= for configuring the access mode of these
2235 directories. These settings are particularly useful in combination
2236 with DynamicUser=yes as they provide secure, properly-owned,
2237 writable, and stateful locations for storage, excluded from the
2238 sandbox that such services live in otherwise.
2240 * Automake support has been removed from this release. systemd is now
2243 * systemd-journald will now aggressively cache client metadata during
2244 runtime, speeding up log write performance under pressure. This comes
2245 at a small price though: as much of the metadata is read
2246 asynchronously from /proc/ (and isn't implicitly attached to log
2247 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
2248 metadata stored alongside a log entry might be slightly
2249 out-of-date. Previously it could only be slightly newer than the log
2250 message. The time window is small however, and given that the kernel
2251 is unlikely to be improved anytime soon in this regard, this appears
2254 * nss-myhostname/systemd-resolved will now by default synthesize an
2255 A/AAAA resource record for the "_gateway" hostname, pointing to the
2256 current default IP gateway. Previously it did that for the "gateway"
2257 name, hampering adoption, as some distributions wanted to leave that
2258 host name open for local use. The old behaviour may still be
2259 requested at build time.
2261 * systemd-networkd's [Address] section in .network files gained a new
2262 Scope= setting for configuring the IP address scope. The [Network]
2263 section gained a new boolean setting ConfigureWithoutCarrier= that
2264 tells systemd-networkd to ignore link sensing when configuring the
2265 device. The [DHCP] section gained a new Anonymize= boolean option for
2266 turning on a number of options suggested in RFC 7844. A new
2267 [RoutingPolicyRule] section has been added for configuring the IP
2268 routing policy. The [Route] section has gained support for a new
2269 Type= setting which permits configuring
2270 blackhole/unreachable/prohibit routes.
2272 * The [VRF] section in .netdev files gained a new Table= setting for
2273 configuring the routing table to use. The [Tunnel] section gained a
2274 new Independent= boolean field for configuring tunnels independent of
2275 an underlying network interface. The [Bridge] section gained a new
2276 GroupForwardMask= option for configuration of propagation of link
2277 local frames between bridge ports.
2279 * The WakeOnLan= setting in .link files gained support for a number of
2280 new modes. A new TCP6SegmentationOffload= setting has been added for
2281 configuring TCP/IPv6 hardware segmentation offload.
2283 * The IPv6 RA sender implementation may now optionally send out RDNSS
2284 and RDNSSL records to supply DNS configuration to peers.
2286 * systemd-nspawn gained support for a new --system-call-filter= command
2287 line option for adding and removing entries in the default system
2288 call filter it applies. Moreover systemd-nspawn has been changed to
2289 implement a system call whitelist instead of a blacklist.
2291 * systemd-run gained support for a new --pipe command line option. If
2292 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
2293 are directly passed on to the activated transient service
2294 executable. This allows invoking arbitrary processes as systemd
2295 services (for example to take benefit of dependency management,
2296 accounting management, resource management or log management that is
2297 done automatically for services) — while still allowing them to be
2298 integrated in a classic UNIX shell pipeline.
2300 * When a service sends RELOAD=1 via sd_notify() and reload propagation
2301 using ReloadPropagationTo= is configured, a reload is now propagated
2302 to configured units. (Previously this was only done on explicitly
2303 requested reloads, using "systemctl reload" or an equivalent
2306 * For each service unit a restart counter is now kept: it is increased
2307 each time the service is restarted due to Restart=, and may be
2308 queried using "systemctl show -p NRestarts …".
2310 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
2311 @signal and @timer have been added, for usage with SystemCallFilter=
2312 in unit files and the new --system-call-filter= command line option
2313 of systemd-nspawn (see above).
2315 * ExecStart= lines in unit files gained two new modifiers: when a
2316 command line is prefixed with "!" the command will be executed as
2317 configured, except for the credentials applied by
2318 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
2319 "+", but does still apply namespacing options unlike "+". There's
2320 also "!!" now, which is mostly identical, but becomes a NOP on
2321 systems that support ambient capabilities. This is useful to write
2322 unit files that work with ambient capabilities where possible but
2323 automatically fall back to traditional privilege dropping mechanisms
2324 on systems where this is not supported.
2326 * ListenNetlink= settings in socket units now support RDMA netlink
2329 * A new unit file setting LockPersonality= has been added which permits
2330 locking down the chosen execution domain ("personality") of a service
2333 * A new special target "getty-pre.target" has been added, which is
2334 ordered before all text logins, and may be used to order services
2335 before textual logins acquire access to the console.
2337 * systemd will now attempt to load the virtio-rng.ko kernel module very
2338 early on if a VM environment supporting this is detected. This should
2339 improve entropy during early boot in virtualized environments.
2341 * A _netdev option is now supported in /etc/crypttab that operates in a
2342 similar way as the same option in /etc/fstab: it permits configuring
2343 encrypted devices that need to be ordered after the network is up.
2344 Following this logic, two new special targets
2345 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
2346 added that are to cryptsetup.target what remote-fs.target and
2347 remote-fs-pre.target are to local-fs.target.
2349 * Service units gained a new UnsetEnvironment= setting which permits
2350 unsetting specific environment variables for services that are
2351 normally passed to it (for example in order to mask out locale
2352 settings for specific services that can't deal with it).
2354 * Units acquired a new boolean option IPAccounting=. When turned on, IP
2355 traffic accounting (packet count as well as byte count) is done for
2356 the service, and shown as part of "systemctl status" or "systemd-run
2359 * Service units acquired two new options IPAddressAllow= and
2360 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
2361 for configuring a simple IP access control list for all sockets of
2362 the unit. These options are available also on .slice and .socket
2363 units, permitting flexible access list configuration for individual
2364 services as well as groups of services (as defined by a slice unit),
2365 including system-wide. Note that IP ACLs configured this way are
2366 enforced on every single IPv4 and IPv6 socket created by any process
2367 of the service unit, and apply to ingress as well as egress traffic.
2369 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
2370 structured log message is generated each time the unit is stopped,
2371 containing information about the consumed resources of this
2374 * A new setting KeyringMode= has been added to unit files, which may be
2375 used to control how the kernel keyring is set up for executed
2378 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
2379 "systemctl kexec" and "systemctl exit" are now always asynchronous in
2380 behaviour (that is: these commands return immediately after the
2381 operation was enqueued instead of waiting for the operation to
2382 complete). Previously, "systemctl poweroff" and "systemctl reboot"
2383 were asynchronous on systems using systemd-logind (i.e. almost
2384 always, and like they were on sysvinit), and the other three commands
2385 were unconditionally synchronous. With this release this is cleaned
2386 up, and callers will see the same asynchronous behaviour on all
2387 systems for all five operations.
2389 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
2392 * .timer units now accept calendar specifications in other timezones
2393 than UTC or the local timezone.
2395 * The tmpfiles snippet var.conf has been changed to create
2396 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
2397 the "utmp" group already, and it appears to be generally understood
2398 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
2399 databases. Previously this was implemented correctly for all these
2400 databases excepts btmp, which has been opened up like this now
2401 too. Note that while the other databases are world-readable
2402 (i.e. 0644), btmp is not and remains more restrictive.
2404 * The systemd-resolve tool gained a new --reset-server-features
2405 switch. When invoked like this systemd-resolved will forget
2406 everything it learnt about the features supported by the configured
2407 upstream DNS servers, and restarts the feature probing logic on the
2408 next resolver look-up for them at the highest feature level
2411 * The status dump systemd-resolved sends to the logs upon receiving
2412 SIGUSR1 now also includes information about all DNS servers it is
2413 configured to use, and the features levels it probed for them.
2415 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
2416 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
2417 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
2418 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
2419 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
2420 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
2421 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
2422 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
2423 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
2424 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
2425 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
2426 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
2427 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
2428 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
2429 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
2430 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
2431 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
2432 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
2433 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
2434 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
2436 — Berlin, 2017-10-06
2440 * Meson is now supported as build system in addition to Automake. It is
2441 our plan to remove Automake in one of our next releases, so that
2442 Meson becomes our exclusive build system. Hence, please start using
2443 the Meson build system in your downstream packaging. There's plenty
2444 of documentation around how to use Meson, the extremely brief
2447 ./autogen.sh && ./configure && make && sudo make install
2451 meson build && ninja -C build && sudo ninja -C build install
2453 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
2454 which permits configuring a timeout on the time a job is
2455 running. This is particularly useful for setting timeouts on jobs for
2458 * Unit files gained two new options ConditionUser= and ConditionGroup=
2459 for conditionalizing units based on the identity of the user/group
2460 running a systemd user instance.
2462 * systemd-networkd now understands a new FlowLabel= setting in the
2463 [VXLAN] section of .network files, as well as a Priority= in
2464 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
2465 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
2466 gained support for configuration of GENEVE links, and IPv6 address
2467 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
2469 * .link files now understand a new Port= setting.
2471 * systemd-networkd's DHCP support gained support for DHCP option 119
2472 (domain search list).
2474 * systemd-networkd gained support for serving IPv6 address ranges using
2475 the Router Advertisement protocol. The new .network configuration
2476 section [IPv6Prefix] may be used to configure the ranges to
2477 serve. This is implemented based on a new, minimal, native server
2478 implementation of RA.
2480 * journalctl's --output= switch gained support for a new parameter
2481 "short-iso-precise" for a mode where timestamps are shown as precise
2484 * systemd-udevd's "net_id" builtin may now generate stable network
2485 interface names from IBM PowerVM VIO devices as well as ACPI platform
2488 * MulticastDNS support in systemd-resolved may now be explicitly
2489 enabled/disabled using the new MulticastDNS= configuration file
2492 * systemd-resolved may now optionally use libidn2 instead of the libidn
2493 for processing internationalized domain names. Support for libidn2
2494 should be considered experimental and should not be enabled by
2497 * "machinectl pull-tar" and related call may now do verification of
2498 downloaded images using SUSE-style .sha256 checksum files in addition
2499 to the already existing support for validating using Ubuntu-style
2502 * sd-bus gained support for a new sd_bus_message_appendv() call which
2503 is va_list equivalent of sd_bus_message_append().
2505 * sd-boot gained support for validating images using SHIM/MOK.
2507 * The SMACK code learnt support for "onlycap".
2509 * systemd-mount --umount is now much smarter in figuring out how to
2510 properly unmount a device given its mount or device path.
2512 * The code to call libnss_dns as a fallback from libnss_resolve when
2513 the communication with systemd-resolved fails was removed. This
2514 fallback was redundant and interfered with the [!UNAVAIL=return]
2515 suffix. See nss-resolve(8) for the recommended configuration.
2517 * systemd-logind may now be restarted without losing state. It stores
2518 the file descriptors for devices it manages in the system manager
2519 using the FDSTORE= mechanism. Please note that further changes in
2520 other components may be required to make use of this (for example
2521 Xorg has code to listen for stops of systemd-logind and terminate
2522 itself when logind is stopped or restarted, in order to avoid using
2523 stale file descriptors for graphical devices, which is now
2524 counterproductive and must be reverted in order for restarts of
2525 systemd-logind to be safe. See
2526 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
2528 * All kernel-install plugins are called with the environment variable
2529 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
2530 /etc/machine-id. If the machine ID could not be determined,
2531 $KERNEL_INSTALL_MACHINE_ID will be empty. Plugins should not put
2532 anything in the entry directory (passed as the second argument) if
2533 $KERNEL_INSTALL_MACHINE_ID is empty. For backwards compatibility, a
2534 temporary directory is passed as the entry directory and removed
2535 after all the plugins exit.
2537 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
2538 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
2539 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
2540 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
2541 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
2542 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
2543 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
2544 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
2545 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
2546 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
2547 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
2548 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
2549 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
2550 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
2551 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
2552 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
2553 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
2554 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
2555 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
2556 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
2557 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
2558 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
2559 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
2560 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
2561 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
2562 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
2563 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
2564 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
2565 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
2568 — Berlin, 2017-07-12
2572 * The "hybrid" control group mode has been modified to improve
2573 compatibility with "legacy" cgroups-v1 setups. Specifically, the
2574 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
2575 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
2576 cgroups-v1 hierarchy), the only externally visible change being that
2577 the cgroups-v2 hierarchy is also mounted, to
2578 /sys/fs/cgroup/unified. This should provide a large degree of
2579 compatibility with "legacy" cgroups-v1, while taking benefit of the
2580 better management capabilities of cgroups-v2.
2582 * The default control group setup mode may be selected both a boot-time
2583 via a set of kernel command line parameters (specifically:
2584 systemd.unified_cgroup_hierarchy= and
2585 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
2586 default selected on the configure command line
2587 (--with-default-hierarchy=). The upstream default is "hybrid"
2588 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
2589 this will change in a future systemd version to be "unified" (pure
2590 cgroups-v2 mode). The third option for the compile time option is
2591 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
2592 distributions to default to "hybrid" mode for release distributions,
2593 starting with v233. We recommend "unified" for development
2594 distributions (specifically: distributions such as Fedora's rawhide)
2595 as that's where things are headed in the long run. Use "legacy" for
2596 greatest stability and compatibility only.
2598 * Note one current limitation of "unified" and "hybrid" control group
2599 setup modes: the kernel currently does not permit the systemd --user
2600 instance (i.e. unprivileged code) to migrate processes between two
2601 disconnected cgroup subtrees, even if both are managed and owned by
2602 the user. This effectively means "systemd-run --user --scope" doesn't
2603 work when invoked from outside of any "systemd --user" service or
2604 scope. Specifically, it is not supported from session scopes. We are
2605 working on fixing this in a future systemd version. (See #3388 for
2606 further details about this.)
2608 * DBus policy files are now installed into /usr rather than /etc. Make
2609 sure your system has dbus >= 1.9.18 running before upgrading to this
2610 version, or override the install path with --with-dbuspolicydir= .
2612 * All python scripts shipped with systemd (specifically: the various
2613 tests written in Python) now require Python 3.
2615 * systemd unit tests can now run standalone (without the source or
2616 build directories), and can be installed into /usr/lib/systemd/tests/
2617 with 'make install-tests'.
2619 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
2620 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
2623 * Support for the %c, %r, %R specifiers in unit files has been
2624 removed. Specifiers are not supposed to be dependent on configuration
2625 in the unit file itself (so that they resolve the same regardless
2626 where used in the unit files), but these specifiers were influenced
2627 by the Slice= option.
2629 * The shell invoked by debug-shell.service now defaults to /bin/sh in
2630 all cases. If distributions want to use a different shell for this
2631 purpose (for example Fedora's /sbin/sushell) they need to specify
2632 this explicitly at configure time using --with-debug-shell=.
2634 * The confirmation spawn prompt has been reworked to offer the
2637 (c)ontinue, proceed without asking anymore
2638 (D)ump, show the state of the unit
2639 (f)ail, don't execute the command and pretend it failed
2641 (i)nfo, show a short summary of the unit
2642 (j)obs, show jobs that are in progress
2643 (s)kip, don't execute the command and pretend it succeeded
2644 (y)es, execute the command
2646 The 'n' choice for the confirmation spawn prompt has been removed,
2647 because its meaning was confusing.
2649 The prompt may now also be redirected to an alternative console by
2650 specifying the console as parameter to systemd.confirm_spawn=.
2652 * Services of Type=notify require a READY=1 notification to be sent
2653 during startup. If no such message is sent, the service now fails,
2654 even if the main process exited with a successful exit code.
2656 * Services that fail to start up correctly now always have their
2657 ExecStopPost= commands executed. Previously, they'd enter "failed"
2658 state directly, without executing these commands.
2660 * The option MulticastDNS= of network configuration files has acquired
2661 an actual implementation. With MulticastDNS=yes a host can resolve
2662 names of remote hosts and reply to mDNS A and AAAA requests.
2664 * When units are about to be started an additional check is now done to
2665 ensure that all dependencies of type BindsTo= (when used in
2666 combination with After=) have been started.
2668 * systemd-analyze gained a new verb "syscall-filter" which shows which
2669 system call groups are defined for the SystemCallFilter= unit file
2670 setting, and which system calls they contain.
2672 * A new system call filter group "@filesystem" has been added,
2673 consisting of various file system related system calls. Group
2674 "@reboot" has been added, covering reboot, kexec and shutdown related
2675 calls. Finally, group "@swap" has been added covering swap
2676 configuration related calls.
2678 * A new unit file option RestrictNamespaces= has been added that may be
2679 used to restrict access to the various process namespace types the
2680 Linux kernel provides. Specifically, it may be used to take away the
2681 right for a service unit to create additional file system, network,
2682 user, and other namespaces. This sandboxing option is particularly
2683 relevant due to the high amount of recently discovered namespacing
2684 related vulnerabilities in the kernel.
2686 * systemd-udev's .link files gained support for a new AutoNegotiation=
2687 setting for configuring Ethernet auto-negotiation.
2689 * systemd-networkd's .network files gained support for a new
2690 ListenPort= setting in the [DHCP] section to explicitly configure the
2691 UDP client port the DHCP client shall listen on.
2693 * .network files gained a new Unmanaged= boolean setting for explicitly
2694 excluding one or more interfaces from management by systemd-networkd.
2696 * The systemd-networkd ProxyARP= option has been renamed to
2697 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
2698 renamed to ReduceARPProxy=. The old names continue to be available
2701 * systemd-networkd gained support for configuring IPv6 Proxy NDP
2702 addresses via the new IPv6ProxyNDPAddress= .network file setting.
2704 * systemd-networkd's bonding device support gained support for two new
2705 configuration options ActiveSlave= and PrimarySlave=.
2707 * The various options in the [Match] section of .network files gained
2708 support for negative matching.
2710 * New systemd-specific mount options are now understood in /etc/fstab:
2712 x-systemd.mount-timeout= may be used to configure the maximum
2713 permitted runtime of the mount command.
2715 x-systemd.device-bound may be set to bind a mount point to its
2716 backing device unit, in order to automatically remove a mount point
2717 if its backing device is unplugged. This option may also be
2718 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
2719 on the block device, which is now automatically set for all CDROM
2720 drives, so that mounted CDs are automatically unmounted when they are
2721 removed from the drive.
2723 x-systemd.after= and x-systemd.before= may be used to explicitly
2724 order a mount after or before another unit or mount point.
2726 * Enqueued start jobs for device units are now automatically garbage
2727 collected if there are no jobs waiting for them anymore.
2729 * systemctl list-jobs gained two new switches: with --after, for every
2730 queued job the jobs it's waiting for are shown; with --before the
2731 jobs which it's blocking are shown.
2733 * systemd-nspawn gained support for ephemeral boots from disk images
2734 (or in other words: --ephemeral and --image= may now be
2735 combined). Moreover, ephemeral boots are now supported for normal
2736 directories, even if the backing file system is not btrfs. Of course,
2737 if the file system does not support file system snapshots or
2738 reflinks, the initial copy operation will be relatively expensive, but
2739 this should still be suitable for many use cases.
2741 * Calendar time specifications in .timer units now support
2742 specifications relative to the end of a month by using "~" instead of
2743 "-" as separator between month and day. For example, "*-02~03" means
2744 "the third last day in February". In addition a new syntax for
2745 repeated events has been added using the "/" character. For example,
2746 "9..17/2:00" means "every two hours from 9am to 5pm".
2748 * systemd-socket-proxyd gained a new parameter --connections-max= for
2749 configuring the maximum number of concurrent connections.
2751 * sd-id128 gained a new API for generating unique IDs for the host in a
2752 way that does not leak the machine ID. Specifically,
2753 sd_id128_get_machine_app_specific() derives an ID based on the
2754 machine ID a in well-defined, non-reversible, stable way. This is
2755 useful whenever an identifier for the host is needed but where the
2756 identifier shall not be useful to identify the system beyond the
2757 scope of the application itself. (Internally this uses HMAC-SHA256 as
2758 keyed hash function using the machine ID as input.)
2760 * NotifyAccess= gained a new supported value "exec". When set
2761 notifications are accepted from all processes systemd itself invoked,
2762 including all control processes.
2764 * .nspawn files gained support for defining overlay mounts using the
2765 Overlay= and OverlayReadOnly= options. Previously this functionality
2766 was only available on the systemd-nspawn command line.
2768 * systemd-nspawn's --bind= and --overlay= options gained support for
2769 bind/overlay mounts whose source lies within the container tree by
2770 prefixing the source path with "+".
2772 * systemd-nspawn's --bind= and --overlay= options gained support for
2773 automatically allocating a temporary source directory in /var/tmp
2774 that is removed when the container dies. Specifically, if the source
2775 directory is specified as empty string this mechanism is selected. An
2776 example usage is --overlay=+/var::/var, which creates an overlay
2777 mount based on the original /var contained in the image, overlayed
2778 with a temporary directory in the host's /var/tmp. This way changes
2779 to /var are automatically flushed when the container shuts down.
2781 * systemd-nspawn --image= option does now permit raw file system block
2782 devices (in addition to images containing partition tables, as
2785 * The disk image dissection logic in systemd-nspawn gained support for
2786 automatically setting up LUKS encrypted as well as Verity protected
2787 partitions. When a container is booted from an encrypted image the
2788 passphrase is queried at start-up time. When a container with Verity
2789 data is started, the root hash is search in a ".roothash" file
2790 accompanying the disk image (alternatively, pass the root hash via
2791 the new --root-hash= command line option).
2793 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
2794 be used to dissect disk images the same way as systemd-nspawn does
2795 it, following the Bootable Partition Specification. It may even be
2796 used to mount disk images with complex partition setups (including
2797 LUKS and Verity partitions) to a local host directory, in order to
2798 inspect them. This tool is not considered public API (yet), and is
2799 thus not installed into /usr/bin. Please do not rely on its
2800 existence, since it might go away or be changed in later systemd
2803 * A new generator "systemd-verity-generator" has been added, similar in
2804 style to "systemd-cryptsetup-generator", permitting automatic setup of
2805 Verity root partitions when systemd boots up. In order to make use of
2806 this your partition setup should follow the Discoverable Partitions
2807 Specification, and the GPT partition ID of the root file system
2808 partition should be identical to the upper 128bit of the Verity root
2809 hash. The GPT partition ID of the Verity partition protecting it
2810 should be the lower 128bit of the Verity root hash. If the partition
2811 image follows this model it is sufficient to specify a single
2812 "roothash=" kernel command line argument to both configure which root
2813 image and verity partition to use as well as the root hash for
2814 it. Note that systemd-nspawn's Verity support follows the same
2815 semantics, meaning that disk images with proper Verity data in place
2816 may be booted in containers with systemd-nspawn as well as on
2817 physical systems via the verity generator. Also note that the "mkosi"
2818 tool available at https://github.com/systemd/mkosi has been updated
2819 to generate Verity protected disk images following this scheme. In
2820 fact, it has been updated to generate disk images that optionally
2821 implement a complete UEFI SecureBoot trust chain, involving a signed
2822 kernel and initrd image that incorporates such a root hash as well as
2823 a Verity-enabled root partition.
2825 * The hardware database (hwdb) udev supports has been updated to carry
2826 accelerometer quirks.
2828 * All system services are now run with a fresh kernel keyring set up
2829 for them. The invocation ID is stored by default in it, thus
2830 providing a safe, non-overridable way to determine the invocation
2833 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
2834 options for bind mounting arbitrary paths in a service-specific
2835 way. When these options are used, arbitrary host or service files and
2836 directories may be mounted to arbitrary locations in the service's
2839 * Documentation has been added that lists all of systemd's low-level
2840 environment variables:
2842 https://github.com/systemd/systemd/blob/master/docs/ENVIRONMENT.md
2844 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
2845 whether a specific socket file descriptor matches a specified socket
2848 * systemd-firstboot has been updated to check for the
2849 systemd.firstboot= kernel command line option. It accepts a boolean
2850 and when set to false the first boot questions are skipped.
2852 * systemd-fstab-generator has been updated to check for the
2853 systemd.volatile= kernel command line option, which either takes an
2854 optional boolean parameter or the special value "state". If used the
2855 system may be booted in a "volatile" boot mode. Specifically,
2856 "systemd.volatile" is used, the root directory will be mounted as
2857 tmpfs, and only /usr is mounted from the actual root file system. If
2858 "systemd.volatile=state" is used, the root directory will be mounted
2859 as usual, but /var is mounted as tmpfs. This concept provides similar
2860 functionality as systemd-nspawn's --volatile= option, but provides it
2861 on physical boots. Use this option for implementing stateless
2862 systems, or testing systems with all state and/or configuration reset
2863 to the defaults. (Note though that many distributions are not
2864 prepared to boot up without a populated /etc or /var, though.)
2866 * systemd-gpt-auto-generator gained support for LUKS encrypted root
2867 partitions. Previously it only supported LUKS encrypted partitions
2868 for all other uses, except for the root partition itself.
2870 * Socket units gained support for listening on AF_VSOCK sockets for
2871 communication in virtualized QEMU environments.
2873 * The "configure" script gained a new option --with-fallback-hostname=
2874 for specifying the fallback hostname to use if none is configured in
2875 /etc/hostname. For example, by specifying
2876 --with-fallback-hostname=fedora it is possible to default to a
2877 hostname of "fedora" on pristine installations.
2879 * systemd-cgls gained support for a new --unit= switch for listing only
2880 the control groups of a specific unit. Similar --user-unit= has been
2881 added for listing only the control groups of a specific user unit.
2883 * systemd-mount gained a new --umount switch for unmounting a mount or
2884 automount point (and all mount/automount points below it).
2886 * systemd will now refuse full configuration reloads (via systemctl
2887 daemon-reload and related calls) unless at least 16MiB of free space
2888 are available in /run. This is a safety precaution in order to ensure
2889 that generators can safely operate after the reload completed.
2891 * A new unit file option RootImage= has been added, which has a similar
2892 effect as RootDirectory= but mounts the service's root directory from
2893 a disk image instead of plain directory. This logic reuses the same
2894 image dissection and mount logic that systemd-nspawn already uses,
2895 and hence supports any disk images systemd-nspawn supports, including
2896 those following the Discoverable Partition Specification, as well as
2897 Verity enabled images. This option enables systemd to run system
2898 services directly off disk images acting as resource bundles,
2899 possibly even including full integrity data.
2901 * A new MountAPIVFS= unit file option has been added, taking a boolean
2902 argument. If enabled /proc, /sys and /dev (collectively called the
2903 "API VFS") will be mounted for the service. This is only relevant if
2904 RootDirectory= or RootImage= is used for the service, as these mounts
2905 are of course in place in the host mount namespace anyway.
2907 * systemd-nspawn gained support for a new --pivot-root= switch. If
2908 specified the root directory within the container image is pivoted to
2909 the specified mount point, while the original root disk is moved to a
2910 different place. This option enables booting of ostree images
2911 directly with systemd-nspawn.
2913 * The systemd build scripts will no longer complain if the NTP server
2914 addresses are not changed from the defaults. Google now supports
2915 these NTP servers officially. We still recommend downstreams to
2916 properly register an NTP pool with the NTP pool project though.
2918 * coredumpctl gained a new "--reverse" option for printing the list
2919 of coredumps in reverse order.
2921 * coredumpctl will now show additional information about truncated and
2922 inaccessible coredumps, as well as coredumps that are still being
2923 processed. It also gained a new --quiet switch for suppressing
2924 additional informational message in its output.
2926 * coredumpctl gained support for only showing coredumps newer and/or
2927 older than specific timestamps, using the new --since= and --until=
2928 options, reminiscent of journalctl's options by the same name.
2930 * The systemd-coredump logic has been improved so that it may be reused
2931 to collect backtraces in non-compiled languages, for example in
2932 scripting languages such as Python.
2934 * machinectl will now show the UID shift of local containers, if user
2935 namespacing is enabled for them.
2937 * systemd will now optionally run "environment generator" binaries at
2938 configuration load time. They may be used to add environment
2939 variables to the environment block passed to services invoked. One
2940 user environment generator is shipped by default that sets up
2941 environment variables based on files dropped into /etc/environment.d
2942 and ~/.config/environment.d/.
2944 * systemd-resolved now includes the new, recently published 2017 DNSSEC
2947 * hostnamed has been updated to report a new chassis type of
2948 "convertible" to cover "foldable" laptops that can both act as a
2949 tablet and as a laptop, such as various Lenovo Yoga devices.
2951 Contributions from: Adrián López, Alexander Galanin, Alexander
2952 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
2953 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
2954 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
2955 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
2956 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
2957 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
2958 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
2959 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
2960 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
2961 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
2962 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
2963 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
2964 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
2965 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
2966 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
2967 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
2968 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
2969 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
2970 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
2971 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
2972 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
2973 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
2974 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
2975 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
2976 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
2977 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
2978 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
2981 — Berlin, 2017-03-01
2985 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
2986 RestrictAddressFamilies= enabled. These sandboxing options should
2987 generally be compatible with the various external udev call-out
2988 binaries we are aware of, however there may be exceptions, in
2989 particular when exotic languages for these call-outs are used. In
2990 this case, consider turning off these settings locally.
2992 * The new RemoveIPC= option can be used to remove IPC objects owned by
2993 the user or group of a service when that service exits.
2995 * The new ProtectKernelModules= option can be used to disable explicit
2996 load and unload operations of kernel modules by a service. In
2997 addition access to /usr/lib/modules is removed if this option is set.
2999 * ProtectSystem= option gained a new value "strict", which causes the
3000 whole file system tree with the exception of /dev, /proc, and /sys,
3001 to be remounted read-only for a service.
3003 * The new ProtectKernelTunables= option can be used to disable
3004 modification of configuration files in /sys and /proc by a service.
3005 Various directories and files are remounted read-only, so access is
3006 restricted even if the file permissions would allow it.
3008 * The new ProtectControlGroups= option can be used to disable write
3009 access by a service to /sys/fs/cgroup.
3011 * Various systemd services have been hardened with
3012 ProtectKernelTunables=yes, ProtectControlGroups=yes,
3013 RestrictAddressFamilies=.
3015 * Support for dynamically creating users for the lifetime of a service
3016 has been added. If DynamicUser=yes is specified, user and group IDs
3017 will be allocated from the range 61184..65519 for the lifetime of the
3018 service. They can be resolved using the new nss-systemd.so NSS
3019 module. The module must be enabled in /etc/nsswitch.conf. Services
3020 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
3021 any resources allocated by the service will be cleaned up when the
3022 service exits. They also have ProtectHome=read-only and
3023 ProtectSystem=strict enabled, so they are not able to make any
3024 permanent modifications to the system.
3026 * The nss-systemd module also always resolves root and nobody, making
3027 it possible to have no /etc/passwd or /etc/group files in minimal
3028 container or chroot environments.
3030 * Services may be started with their own user namespace using the new
3031 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
3032 under which the service is running are mapped. All other users are
3035 * Support for the cgroup namespace has been added to systemd-nspawn. If
3036 supported by kernel, the container system started by systemd-nspawn
3037 will have its own view of the cgroup hierarchy. This new behaviour
3038 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
3040 * The new MemorySwapMax= option can be used to limit the maximum swap
3041 usage under the unified cgroup hierarchy.
3043 * Support for the CPU controller in the unified cgroup hierarchy has
3044 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
3045 options. This controller requires out-of-tree patches for the kernel
3046 and the support is provisional.
3048 * Mount and automount units may now be created transiently
3049 (i.e. dynamically at runtime via the bus API, instead of requiring
3050 unit files in the file system).
3052 * systemd-mount is a new tool which may mount file systems – much like
3053 mount(8), optionally pulling in additional dependencies through
3054 transient .mount and .automount units. For example, this tool
3055 automatically runs fsck on a backing block device before mounting,
3056 and allows the automount logic to be used dynamically from the
3057 command line for establishing mount points. This tool is particularly
3058 useful when dealing with removable media, as it will ensure fsck is
3059 run – if necessary – before the first access and that the file system
3060 is quickly unmounted after each access by utilizing the automount
3061 logic. This maximizes the chance that the file system on the
3062 removable media stays in a clean state, and if it isn't in a clean
3063 state is fixed automatically.
3065 * LazyUnmount=yes option for mount units has been added to expose the
3066 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
3069 * /efi will be used as the mount point of the EFI boot partition, if
3070 the directory is present, and the mount point was not configured
3071 through other means (e.g. fstab). If /efi directory does not exist,
3072 /boot will be used as before. This makes it easier to automatically
3073 mount the EFI partition on systems where /boot is used for something
3076 * When operating on GPT disk images for containers, systemd-nspawn will
3077 now mount the ESP to /boot or /efi according to the same rules as PID
3078 1 running on a host. This allows tools like "bootctl" to operate
3079 correctly within such containers, in order to make container images
3080 bootable on physical systems.
3082 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
3084 * Two new user session targets have been added to support running
3085 graphical sessions under the systemd --user instance:
3086 graphical-session.target and graphical-session-pre.target. See
3087 systemd.special(7) for a description of how those targets should be
3090 * The vconsole initialization code has been significantly reworked to
3091 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
3092 support unicode keymaps. Font and keymap configuration will now be
3093 copied to all allocated virtual consoles.
3095 * FreeBSD's bhyve virtualization is now detected.
3097 * Information recorded in the journal for core dumps now includes the
3098 contents of /proc/mountinfo and the command line of the process at
3099 the top of the process hierarchy (which is usually the init process
3102 * systemd-journal-gatewayd learned the --directory= option to serve
3103 files from the specified location.
3105 * journalctl --root=… can be used to peruse the journal in the
3106 /var/log/ directories inside of a container tree. This is similar to
3107 the existing --machine= option, but does not require the container to
3110 * The hardware database has been extended to support
3111 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
3114 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
3115 specify the click rate for mice which include a horizontal wheel with
3116 a click rate that is different than the one for the vertical wheel.
3118 * systemd-run gained a new --wait option that makes service execution
3119 synchronous. (Specifically, the command will not return until the
3120 specified service binary exited.)
3122 * systemctl gained a new --wait option that causes the start command to
3123 wait until the units being started have terminated again.
3125 * A new journal output mode "short-full" has been added which displays
3126 timestamps with abbreviated English day names and adds a timezone
3127 suffix. Those timestamps include more information than the default
3128 "short" output mode, and can be passed directly to journalctl's
3129 --since= and --until= options.
3131 * /etc/resolv.conf will be bind-mounted into containers started by
3132 systemd-nspawn, if possible, so any changes to resolv.conf contents
3133 are automatically propagated to the container.
3135 * The number of instances for socket-activated services originating
3136 from a single IP address can be limited with
3137 MaxConnectionsPerSource=, extending the existing setting of
3140 * systemd-networkd gained support for vcan ("Virtual CAN") interface
3143 * .netdev and .network configuration can now be extended through
3146 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
3147 Segmentation Offload, Generic Receive Offload, Large Receive Offload
3148 can be enabled and disabled using the new UDPSegmentationOffload=,
3149 TCPSegmentationOffload=, GenericSegmentationOffload=,
3150 GenericReceiveOffload=, LargeReceiveOffload= options in the
3151 [Link] section of .link files.
3153 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
3154 Port VLAN ID can be configured for bridge devices using the new STP=,
3155 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
3156 section of .netdev files.
3158 * The route table to which routes received over DHCP or RA should be
3159 added can be configured with the new RouteTable= option in the [DHCP]
3160 and [IPv6AcceptRA] sections of .network files.
3162 * The Address Resolution Protocol can be disabled on links managed by
3163 systemd-networkd using the ARP=no setting in the [Link] section of
3166 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
3167 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
3168 encode information about the result and exit codes of the current
3169 service runtime cycle.
3171 * systemd-sysctl will now configure kernel parameters in the order
3172 they occur in the configuration files. This matches what sysctl
3173 has been traditionally doing.
3175 * kernel-install "plugins" that are executed to perform various
3176 tasks after a new kernel is added and before an old one is removed
3177 can now return a special value to terminate the procedure and
3178 prevent any later plugins from running.
3180 * Journald's SplitMode=login setting has been deprecated. It has been
3181 removed from documentation, and its use is discouraged. In a future
3182 release it will be completely removed, and made equivalent to current
3183 default of SplitMode=uid.
3185 * Storage=both option setting in /etc/systemd/coredump.conf has been
3186 removed. With fast LZ4 compression storing the core dump twice is not
3189 * The --share-system systemd-nspawn option has been replaced with an
3190 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
3191 this functionality is discouraged. In addition the variables
3192 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
3193 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
3194 individual namespaces.
3196 * "machinectl list" now shows the IP address of running containers in
3197 the output, as well as OS release information.
3199 * "loginctl list" now shows the TTY of each session in the output.
3201 * sd-bus gained new API calls sd_bus_track_set_recursive(),
3202 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
3203 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
3204 tracking objects in a "recursive" mode, where a single client can be
3205 counted multiple times, if it takes multiple references.
3207 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
3208 sd_bus_get_exit_on_disconnect(). They may be used to make a
3209 process using sd-bus automatically exit if the bus connection is
3212 * Bus clients of the service manager may now "pin" loaded units into
3213 memory, by taking an explicit reference on them. This is useful to
3214 ensure the client can retrieve runtime data about the service even
3215 after the service completed execution. Taking such a reference is
3216 available only for privileged clients and should be helpful to watch
3217 running services in a race-free manner, and in particular collect
3218 information about exit statuses and results.
3220 * The nss-resolve module has been changed to strictly return UNAVAIL
3221 when communication via D-Bus with resolved failed, and NOTFOUND when
3222 a lookup completed but was negative. This means it is now possible to
3223 neatly configure fallbacks using nsswitch.conf result checking
3224 expressions. Taking benefit of this, the new recommended
3225 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
3227 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
3229 * A new setting CtrlAltDelBurstAction= has been added to
3230 /etc/systemd/system.conf which may be used to configure the precise
3231 behaviour if the user on the console presses Ctrl-Alt-Del more often
3232 than 7 times in 2s. Previously this would unconditionally result in
3233 an expedited, immediate reboot. With this new setting the precise
3234 operation may be configured in more detail, and also turned off
3237 * In .netdev files two new settings RemoteChecksumTx= and
3238 RemoteChecksumRx= are now understood that permit configuring the
3239 remote checksumming logic for VXLAN networks.
3241 * The service manager learnt a new "invocation ID" concept for invoked
3242 services. Each runtime cycle of a service will get a new invocation
3243 ID (a 128bit random UUID) assigned that identifies the current
3244 run of the service uniquely and globally. A new invocation ID
3245 is generated each time a service starts up. The journal will store
3246 the invocation ID of a service along with any logged messages, thus
3247 making the invocation ID useful for matching the online runtime of a
3248 service with the offline log data it generated in a safe way without
3249 relying on synchronized timestamps. In many ways this new service
3250 invocation ID concept is similar to the kernel's boot ID concept that
3251 uniquely and globally identifies the runtime of each boot. The
3252 invocation ID of a service is passed to the service itself via an
3253 environment variable ($INVOCATION_ID). A new bus call
3254 GetUnitByInvocationID() has been added that is similar to GetUnit()
3255 but instead of retrieving the bus path for a unit by its name
3256 retrieves it by its invocation ID. The returned path is valid only as
3257 long as the passed invocation ID is current.
3259 * systemd-resolved gained a new "DNSStubListener" setting in
3260 resolved.conf. It either takes a boolean value or the special values
3261 "udp" and "tcp", and configures whether to enable the stub DNS
3262 listener on 127.0.0.53:53.
3264 * IP addresses configured via networkd may now carry additional
3265 configuration settings supported by the kernel. New options include:
3266 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
3267 PrefixRoute=, AutoJoin=.
3269 * The PAM configuration fragment file for "user@.service" shipped with
3270 systemd (i.e. the --user instance of systemd) has been stripped to
3271 the minimum necessary to make the system boot. Previously, it
3272 contained Fedora-specific stanzas that did not apply to other
3273 distributions. It is expected that downstream distributions add
3274 additional configuration lines, matching their needs to this file,
3275 using it only as rough template of what systemd itself needs. Note
3276 that this reduced fragment does not even include an invocation of
3277 pam_limits which most distributions probably want to add, even though
3278 systemd itself does not need it. (There's also the new build time
3279 option --with-pamconfdir=no to disable installation of the PAM
3282 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
3283 capability is now also dropped from its set (in addition to
3284 CAP_SYS_MKNOD as before).
3286 * In service unit files it is now possible to connect a specific named
3287 file descriptor with stdin/stdout/stdout of an executed service. The
3288 name may be specified in matching .socket units using the
3289 FileDescriptorName= setting.
3291 * A number of journal settings may now be configured on the kernel
3292 command line. Specifically, the following options are now understood:
3293 systemd.journald.max_level_console=,
3294 systemd.journald.max_level_store=,
3295 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
3296 systemd.journald.max_level_wall=.
3298 * "systemctl is-enabled --full" will now show by which symlinks a unit
3299 file is enabled in the unit dependency tree.
3301 * Support for VeraCrypt encrypted partitions has been added to the
3302 "cryptsetup" logic and /etc/crypttab.
3304 * systemd-detect-virt gained support for a new --private-users switch
3305 that checks whether the invoking processes are running inside a user
3306 namespace. Similar, a new special value "private-users" for the
3307 existing ConditionVirtualization= setting has been added, permitting
3308 skipping of specific units in user namespace environments.
3310 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
3311 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
3312 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
3313 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
3314 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
3315 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
3316 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
3317 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
3318 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
3319 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
3320 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
3321 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
3322 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
3323 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
3324 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
3325 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
3326 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
3327 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
3328 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
3329 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
3330 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
3331 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
3332 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
3333 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
3334 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
3335 Jędrzejewski-Szmek, Zeal Jagannatha
3337 — Santa Fe, 2016-11-03
3341 * In service units the various ExecXYZ= settings have been extended
3342 with an additional special character as first argument of the
3343 assigned value: if the character '+' is used the specified command
3344 line it will be run with full privileges, regardless of User=,
3345 Group=, CapabilityBoundingSet= and similar options. The effect is
3346 similar to the existing PermissionsStartOnly= option, but allows
3347 configuration of this concept for each executed command line
3350 * Services may now alter the service watchdog timeout at runtime by
3351 sending a WATCHDOG_USEC= message via sd_notify().
3353 * MemoryLimit= and related unit settings now optionally take percentage
3354 specifications. The percentage is taken relative to the amount of
3355 physical memory in the system (or in case of containers, the assigned
3356 amount of memory). This allows scaling service resources neatly with
3357 the amount of RAM available on the system. Similarly, systemd-logind's
3358 RuntimeDirectorySize= option now also optionally takes percentage
3361 * In similar fashion TasksMax= takes percentage values now, too. The
3362 value is taken relative to the configured maximum number of processes
3363 on the system. The per-service task maximum has been changed to 15%
3364 using this functionality. (Effectively this is an increase of 512 →
3365 4915 for service units, given the kernel's default pid_max setting.)
3367 * Calendar time specifications in .timer units now understand a ".."
3368 syntax for time ranges. Example: "4..7:10" may now be used for
3369 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
3372 * The InaccessableDirectories=, ReadOnlyDirectories= and
3373 ReadWriteDirectories= unit file settings have been renamed to
3374 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
3375 applied to all kinds of file nodes, and not just directories, with
3376 the exception of symlinks. Specifically these settings may now be
3377 used on block and character device nodes, UNIX sockets and FIFOS as
3378 well as regular files. The old names of these settings remain
3379 available for compatibility.
3381 * systemd will now log about all service processes it kills forcibly
3382 (using SIGKILL) because they remained after the clean shutdown phase
3383 of the service completed. This should help identifying services that
3384 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
3385 systemd-logind's configuration a similar log message is generated for
3386 processes killed at the end of each session due to this setting.
3388 * systemd will now set the $JOURNAL_STREAM environment variable for all
3389 services whose stdout/stderr are connected to the Journal (which
3390 effectively means by default: all services). The variable contains
3391 the device and inode number of the file descriptor used for
3392 stdout/stderr. This may be used by invoked programs to detect whether
3393 their stdout/stderr is connected to the Journal, in which case they
3394 can switch over to direct Journal communication, thus being able to
3395 pass extended, structured metadata along with their log messages. As
3396 one example, this is now used by glib's logging primitives.
3398 * When using systemd's default tmp.mount unit for /tmp, the mount point
3399 will now be established with the "nosuid" and "nodev" options. This
3400 avoids privilege escalation attacks that put traps and exploits into
3401 /tmp. However, this might cause problems if you e. g. put container
3402 images or overlays into /tmp; if you need this, override tmp.mount's
3403 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
3406 * systemd now supports the "memory" cgroup controller also on
3409 * The systemd-cgtop tool now optionally takes a control group path as
3410 command line argument. If specified, the control group list shown is
3411 limited to subgroups of that group.
3413 * The SystemCallFilter= unit file setting gained support for
3414 pre-defined, named system call filter sets. For example
3415 SystemCallFilter=@clock is now an effective way to make all clock
3416 changing-related system calls unavailable to a service. A number of
3417 similar pre-defined groups are defined. Writing system call filters
3418 for system services is simplified substantially with this new
3419 concept. Accordingly, all of systemd's own, long-running services now
3420 enable system call filtering based on this, by default.
3422 * A new service setting MemoryDenyWriteExecute= has been added, taking
3423 a boolean value. If turned on, a service may no longer create memory
3424 mappings that are writable and executable at the same time. This
3425 enhances security for services where this is enabled as it becomes
3426 harder to dynamically write and then execute memory in exploited
3427 service processes. This option has been enabled for all of systemd's
3428 own long-running services.
3430 * A new RestrictRealtime= service setting has been added, taking a
3431 boolean argument. If set the service's processes may no longer
3432 acquire realtime scheduling. This improves security as realtime
3433 scheduling may otherwise be used to easily freeze the system.
3435 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
3436 value. This may be used for requesting that the system manager inside
3437 of the container reports start-up completion to nspawn which then
3438 propagates this notification further to the service manager
3439 supervising nspawn itself. A related option NotifyReady= in .nspawn
3440 files has been added too. This functionality allows ordering of the
3441 start-up of multiple containers using the usual systemd ordering
3444 * machinectl gained a new command "stop" that is an alias for
3447 * systemd-resolved gained support for contacting DNS servers on
3448 link-local IPv6 addresses.
3450 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
3451 its caches. A method call for requesting the same operation has been
3452 added to the bus API too, and is made available via "systemd-resolve
3455 * systemd-resolve gained a new --status switch. If passed a brief
3456 summary of the used DNS configuration with per-interface information
3459 * resolved.conf gained a new Cache= boolean option, defaulting to
3460 on. If turned off local DNS caching is disabled. This comes with a
3461 performance penalty in particular when DNSSEC is enabled. Note that
3462 resolved disables its internal caching implicitly anyway, when the
3463 configured DNS server is on a host-local IP address such as ::1 or
3464 127.0.0.1, thus automatically avoiding double local caching.
3466 * systemd-resolved now listens on the local IP address 127.0.0.53:53
3467 for DNS requests. This improves compatibility with local programs
3468 that do not use the libc NSS or systemd-resolved's bus APIs for name
3469 resolution. This minimal DNS service is only available to local
3470 programs and does not implement the full DNS protocol, but enough to
3471 cover local DNS clients. A new, static resolv.conf file, listing just
3472 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
3473 now recommended to make /etc/resolv.conf a symlink to this file in
3474 order to route all DNS lookups to systemd-resolved, regardless if
3475 done via NSS, the bus API or raw DNS packets. Note that this local
3476 DNS service is not as fully featured as the libc NSS or
3477 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
3478 used to deliver link-local address information (as this implies
3479 sending a local interface index along), LLMNR/mDNS support via this
3480 interface is severely restricted. It is thus strongly recommended for
3481 all applications to use the libc NSS API or native systemd-resolved
3484 * systemd-networkd's bridge support learned a new setting
3485 VLANFiltering= for controlling VLAN filtering. Moreover a new section
3486 in .network files has been added for configuring VLAN bridging in
3487 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
3489 * systemd-networkd's IPv6 Router Advertisement code now makes use of
3490 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
3491 now be acquired without relying on DHCPv6. Two new options
3492 UseDomains= and UseDNS= have been added to configure this behaviour.
3494 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
3495 renamed IPv6AcceptRA=, without altering its behaviour. The old
3496 setting name remains available for compatibility reasons.
3498 * The systemd-networkd VTI/VTI6 tunneling support gained new options
3499 Key=, InputKey= and OutputKey=.
3501 * systemd-networkd gained support for VRF ("Virtual Routing Function")
3502 interface configuration.
3504 * "systemctl edit" may now be used to create new unit files by
3505 specifying the --force switch.
3507 * sd-event gained a new function sd_event_get_iteration() for
3508 requesting the current iteration counter of the event loop. It starts
3509 at zero and is increased by one with each event loop iteration.
3511 * A new rpm macro %systemd_ordering is provided by the macros.systemd
3512 file. It can be used in lieu of %systemd_requires in packages which
3513 don't use any systemd functionality and are intended to be installed
3514 in minimal containers without systemd present. This macro provides
3515 ordering dependencies to ensure that if the package is installed in
3516 the same rpm transaction as systemd, systemd will be installed before
3517 the scriptlets for the package are executed, allowing unit presets
3520 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
3521 been added to simplify packaging of generators.
3523 * The os-release file gained VERSION_CODENAME field for the
3524 distribution nickname (e.g. VERSION_CODENAME=woody).
3526 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
3527 can be set to disable parsing of metadata and the creation
3528 of persistent symlinks for that device.
3530 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
3531 to make them available to logged-in users has been reverted.
3533 * Much of the common code of the various systemd components is now
3534 built into an internal shared library libsystemd-shared-231.so
3535 (incorporating the systemd version number in the name, to be updated
3536 with future releases) that the components link to. This should
3537 decrease systemd footprint both in memory during runtime and on
3538 disk. Note that the shared library is not for public use, and is
3539 neither API nor ABI stable, but is likely to change with every new
3540 released update. Packagers need to make sure that binaries
3541 linking to libsystemd-shared.so are updated in step with the
3544 * Configuration for "mkosi" is now part of the systemd
3545 repository. mkosi is a tool to easily build legacy-free OS images,
3546 and is available on github: https://github.com/systemd/mkosi. If
3547 "mkosi" is invoked in the build tree a new raw OS image is generated
3548 incorporating the systemd sources currently being worked on and a
3549 clean, fresh distribution installation. The generated OS image may be
3550 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
3551 UEFI PC. This functionality is particularly useful to easily test
3552 local changes made to systemd in a pristine, defined environment. See
3553 doc/HACKING for details.
3555 * configure learned the --with-support-url= option to specify the
3556 distribution's bugtracker.
3558 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
3559 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
3560 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
3561 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
3562 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
3563 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
3564 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
3565 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
3566 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
3567 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
3568 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
3569 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
3570 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
3571 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
3572 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
3573 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
3574 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
3575 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
3576 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
3578 — Berlin, 2016-07-25
3582 * DNSSEC is now turned on by default in systemd-resolved (in
3583 "allow-downgrade" mode), but may be turned off during compile time by
3584 passing "--with-default-dnssec=no" to "configure" (and of course,
3585 during runtime with DNSSEC= in resolved.conf). We recommend
3586 downstreams to leave this on at least during development cycles and
3587 report any issues with the DNSSEC logic upstream. We are very
3588 interested in collecting feedback about the DNSSEC validator and its
3589 limitations in the wild. Note however, that DNSSEC support is
3590 probably nothing downstreams should turn on in stable distros just
3591 yet, as it might create incompatibilities with a few DNS servers and
3592 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
3593 automatically whenever we detect such incompatible setups, but there
3594 might be systems we do not cover yet. Hence: please help us testing
3595 the DNSSEC code, leave this on where you can, report back, but then
3596 again don't consider turning this on in your stable, LTS or
3597 production release just yet. (Note that you have to enable
3598 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
3599 and its DNSSEC mode for host name resolution from local
3602 * systemd-resolve conveniently resolves DANE records with the --tlsa
3603 option and OPENPGPKEY records with the --openpgp option. It also
3604 supports dumping raw DNS record data via the new --raw= switch.
3606 * systemd-logind will now by default terminate user processes that are
3607 part of the user session scope unit (session-XX.scope) when the user
3608 logs out. This behavior is controlled by the KillUserProcesses=
3609 setting in logind.conf, and the previous default of "no" is now
3610 changed to "yes". This means that user sessions will be properly
3611 cleaned up after, but additional steps are necessary to allow
3612 intentionally long-running processes to survive logout.
3614 While the user is logged in at least once, user@.service is running,
3615 and any service that should survive the end of any individual login
3616 session can be started at a user service or scope using systemd-run.
3617 systemd-run(1) man page has been extended with an example which shows
3618 how to run screen in a scope unit underneath user@.service. The same
3619 command works for tmux.
3621 After the user logs out of all sessions, user@.service will be
3622 terminated too, by default, unless the user has "lingering" enabled.
3623 To effectively allow users to run long-term tasks even if they are
3624 logged out, lingering must be enabled for them. See loginctl(1) for
3625 details. The default polkit policy was modified to allow users to
3626 set lingering for themselves without authentication.
3628 Previous defaults can be restored at compile time by the
3629 --without-kill-user-processes option to "configure".
3631 * systemd-logind gained new configuration settings SessionsMax= and
3632 InhibitorsMax=, both with a default of 8192. It will not register new
3633 user sessions or inhibitors above this limit.
3635 * systemd-logind will now reload configuration on SIGHUP.
3637 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
3638 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
3639 enable. Also, support for the "io" cgroup controller in the unified
3640 hierarchy has been added, so that the "memory", "pids" and "io" are
3641 now the controllers that are supported on the unified hierarchy.
3643 WARNING: it is not possible to use previous systemd versions with
3644 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
3645 is necessary to also update systemd in the initramfs if using the
3646 unified hierarchy. An updated SELinux policy is also required.
3648 * LLDP support has been extended, and both passive (receive-only) and
3649 active (sender) modes are supported. Passive mode ("routers-only") is
3650 enabled by default in systemd-networkd. Active LLDP mode is enabled
3651 by default for containers on the internal network. The "networkctl
3652 lldp" command may be used to list information gathered. "networkctl
3653 status" will also show basic LLDP information on connected peers now.
3655 * The IAID and DUID unique identifier sent in DHCP requests may now be
3656 configured for the system and each .network file managed by
3657 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
3659 * systemd-networkd gained support for configuring proxy ARP support for
3660 each interface, via the ProxyArp= setting in .network files. It also
3661 gained support for configuring the multicast querier feature of
3662 bridge devices, via the new MulticastQuerier= setting in .netdev
3663 files. Similarly, snooping on the IGMP traffic can be controlled
3664 via the new setting MulticastSnooping=.
3666 A new setting PreferredLifetime= has been added for addresses
3667 configured in .network file to configure the lifetime intended for an
3670 The systemd-networkd DHCP server gained the option EmitRouter=, which
3671 defaults to yes, to configure whether the DHCP Option 3 (Router)
3674 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
3675 systemd-socket-activate and installed into /usr/bin. It is now fully
3678 * systemd-journald now uses separate threads to flush changes to disk
3679 when closing journal files, thus reducing impact of slow disk I/O on
3680 logging performance.
3682 * The sd-journal API gained two new calls
3683 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
3684 can be used to open journal files using file descriptors instead of
3685 file or directory paths. sd_journal_open_container() has been
3686 deprecated, sd_journal_open_directory_fd() should be used instead
3687 with the flag SD_JOURNAL_OS_ROOT.
3689 * journalctl learned a new output mode "-o short-unix" that outputs log
3690 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
3691 UTC). It also gained support for a new --no-hostname setting to
3692 suppress the hostname column in the family of "short" output modes.
3694 * systemd-ask-password now optionally skips printing of the password to
3695 stdout with --no-output which can be useful in scripts.
3697 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
3698 (devices tagged with ID_MAKER_TOOL) are now tagged with
3699 "uaccess" and are available to logged in users.
3701 * The DeviceAllow= unit setting now supports specifiers (with "%").
3703 * "systemctl show" gained a new --value switch, which allows print a
3704 only the contents of a specific unit property, without also printing
3705 the property's name. Similar support was added to "show*" verbs
3706 of loginctl and machinectl that output "key=value" lists.
3708 * A new unit type "generated" was added for files dynamically generated
3709 by generator tools. Similarly, a new unit type "transient" is used
3710 for unit files created using the runtime API. "systemctl enable" will
3711 refuse to operate on such files.
3713 * A new command "systemctl revert" has been added that may be used to
3714 revert to the vendor version of a unit file, in case local changes
3715 have been made by adding drop-ins or overriding the unit file.
3717 * "machinectl clean" gained a new verb to automatically remove all or
3718 just hidden container images.
3720 * systemd-tmpfiles gained support for a new line type "e" for emptying
3721 directories, if they exist, without creating them if they don't.
3723 * systemd-nspawn gained support for automatically patching the UID/GIDs
3724 of the owners and the ACLs of all files and directories in a
3725 container tree to match the UID/GID user namespacing range selected
3726 for the container invocation. This mode is enabled via the new
3727 --private-users-chown switch. It also gained support for
3728 automatically choosing a free, previously unused UID/GID range when
3729 starting a container, via the new --private-users=pick setting (which
3730 implies --private-users-chown). Together, these options for the first
3731 time make user namespacing for nspawn containers fully automatic and
3732 thus deployable. The systemd-nspawn@.service template unit file has
3733 been changed to use this functionality by default.
3735 * systemd-nspawn gained a new --network-zone= switch, that allows
3736 creating ad-hoc virtual Ethernet links between multiple containers,
3737 that only exist as long as at least one container referencing them is
3738 running. This allows easy connecting of multiple containers with a
3739 common link that implements an Ethernet broadcast domain. Each of
3740 these network "zones" may be named relatively freely by the user, and
3741 may be referenced by any number of containers, but each container may
3742 only reference one of these "zones". On the lower level, this is
3743 implemented by an automatically managed bridge network interface for
3744 each zone, that is created when the first container referencing its
3745 zone is created and removed when the last one referencing its zone
3748 * The default start timeout may now be configured on the kernel command
3749 line via systemd.default_timeout_start_sec=. It was already
3750 configurable via the DefaultTimeoutStartSec= option in
3751 /etc/systemd/system.conf.
3753 * Socket units gained a new TriggerLimitIntervalSec= and
3754 TriggerLimitBurst= setting to configure a limit on the activation
3755 rate of the socket unit.
3757 * The LimitNICE= setting now optionally takes normal UNIX nice values
3758 in addition to the raw integer limit value. If the specified
3759 parameter is prefixed with "+" or "-" and is in the range -20..19 the
3760 value is understood as UNIX nice value. If not prefixed like this it
3761 is understood as raw RLIMIT_NICE limit.
3763 * Note that the effect of the PrivateDevices= unit file setting changed
3764 slightly with this release: the per-device /dev file system will be
3765 mounted read-only from this version on, and will have "noexec"
3766 set. This (minor) change of behavior might cause some (exceptional)
3767 legacy software to break, when PrivateDevices=yes is set for its
3768 service. Please leave PrivateDevices= off if you run into problems
3771 * systemd-bootchart has been split out to a separate repository:
3772 https://github.com/systemd/systemd-bootchart
3774 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
3775 merged into the kernel in its current form.
3777 * The compatibility libraries libsystemd-daemon.so,
3778 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
3779 which have been deprecated since systemd-209 have been removed along
3780 with the corresponding pkg-config files. All symbols provided by
3781 those libraries are provided by libsystemd.so.
3783 * The Capabilities= unit file setting has been removed (it is ignored
3784 for backwards compatibility). AmbientCapabilities= and
3785 CapabilityBoundingSet= should be used instead.
3787 * A new special target has been added, initrd-root-device.target,
3788 which creates a synchronization point for dependencies of the root
3789 device in early userspace. Initramfs builders must ensure that this
3790 target is now included in early userspace.
3792 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
3793 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
3794 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
3795 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
3796 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
3797 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
3798 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
3799 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
3800 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
3801 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
3802 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
3803 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
3804 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
3805 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
3806 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
3807 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
3808 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
3809 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
3810 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
3811 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
3812 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
3813 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
3814 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
3815 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
3816 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
3819 — Fairfax, 2016-05-21
3823 * The systemd-resolved DNS resolver service has gained a substantial
3824 set of new features, most prominently it may now act as a DNSSEC
3825 validating stub resolver. DNSSEC mode is currently turned off by
3826 default, but is expected to be turned on by default in one of the
3827 next releases. For now, we invite everybody to test the DNSSEC logic
3828 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
3829 service also gained a full set of D-Bus interfaces, including calls
3830 to configure DNS and DNSSEC settings per link (for use by external
3831 network management software). systemd-resolved and systemd-networkd
3832 now distinguish between "search" and "routing" domains. The former
3833 are used to qualify single-label names, the latter are used purely
3834 for routing lookups within certain domains to specific links.
3835 resolved now also synthesizes RRs for all entries from /etc/hosts.
3837 * The systemd-resolve tool (which is a client utility for
3838 systemd-resolved) has been improved considerably and is now fully
3839 supported and documented. Hence it has moved from /usr/lib/systemd to
3842 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
3845 * The coredump collection logic has been reworked: when a coredump is
3846 collected it is now written to disk, compressed and processed
3847 (including stacktrace extraction) from a new instantiated service
3848 systemd-coredump@.service, instead of directly from the
3849 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
3850 processing large coredumps can take up a substantial amount of
3851 resources and time, and this previously happened entirely outside of
3852 systemd's service supervision. With the new logic the core_pattern
3853 hook only does minimal metadata collection before passing off control
3854 to the new instantiated service, which is configured with a time
3855 limit, a nice level and other settings to minimize negative impact on
3856 the rest of the system. Also note that the new logic will honour the
3857 RLIMIT_CORE setting of the crashed process, which now allows users
3858 and processes to turn off coredumping for their processes by setting
3861 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
3862 and all forked processes by default. Previously, PID 1 would leave
3863 the setting at "0" for all processes, as set by the kernel. Note that
3864 the resource limit traditionally has no effect on the generated
3865 coredumps on the system if the /proc/sys/kernel/core_pattern hook
3866 logic is used. Since the limit is now honoured (see above) its
3867 default has been changed so that the coredumping logic is enabled by
3868 default for all processes, while allowing specific opt-out.
3870 * When the stacktrace is extracted from processes of system users, this
3871 is now done as "systemd-coredump" user, in order to sandbox this
3872 potentially security sensitive parsing operation. (Note that when
3873 processing coredumps of normal users this is done under the user ID
3874 of process that crashed, as before.) Packagers should take notice
3875 that it is now necessary to create the "systemd-coredump" system user
3876 and group at package installation time.
3878 * The systemd-activate socket activation testing tool gained support
3879 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
3880 and --seqpacket switches. It also has been extended to support both
3881 new-style and inetd-style file descriptor passing. Use the new
3882 --inetd switch to request inetd-style file descriptor passing.
3884 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
3885 variable, which takes a boolean value. If set to false, ANSI color
3886 output is disabled in the tools even when run on a terminal that
3889 * The VXLAN support in networkd now supports two new settings
3890 DestinationPort= and PortRange=.
3892 * A new systemd.machine_id= kernel command line switch has been added,
3893 that may be used to set the machine ID in /etc/machine-id if it is
3894 not initialized yet. This command line option has no effect if the
3895 file is already initialized.
3897 * systemd-nspawn gained a new --as-pid2 switch that invokes any
3898 specified command line as PID 2 rather than PID 1 in the
3899 container. In this mode PID 1 is a minimal stub init process that
3900 implements the special POSIX and Linux semantics of PID 1 regarding
3901 signal and child process management. Note that this stub init process
3902 is implemented in nspawn itself and requires no support from the
3903 container image. This new logic is useful to support running
3904 arbitrary commands in the container, as normal processes are
3905 generally not prepared to run as PID 1.
3907 * systemd-nspawn gained a new --chdir= switch for setting the current
3908 working directory for the process started in the container.
3910 * "journalctl /dev/sda" will now output all kernel log messages for
3911 specified device from the current boot, in addition to all devices
3912 that are parents of it. This should make log output about devices
3913 pretty useful, as long as kernel drivers attach enough metadata to
3914 the log messages. (The usual SATA drivers do.)
3916 * The sd-journal API gained two new calls
3917 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
3918 that report whether log data from /run or /var has been found.
3920 * journalctl gained a new switch "--fields" that prints all journal
3921 record field names currently in use in the journal. This is backed
3922 by two new sd-journal API calls sd_journal_enumerate_fields() and
3923 sd_journal_restart_fields().
3925 * Most configurable timeouts in systemd now expect an argument of
3926 "infinity" to turn them off, instead of "0" as before. The semantics
3927 from now on is that a timeout of "0" means "now", and "infinity"
3928 means "never". To maintain backwards compatibility, "0" continues to
3929 turn off previously existing timeout settings.
3931 * "systemctl reload-or-try-restart" has been renamed to "systemctl
3932 try-reload-or-restart" to clarify what it actually does: the "try"
3933 logic applies to both reloading and restarting, not just restarting.
3934 The old name continues to be accepted for compatibility.
3936 * On boot-up, when PID 1 detects that the system clock is behind the
3937 release date of the systemd version in use, the clock is now set
3938 to the latter. Previously, this was already done in timesyncd, in order
3939 to avoid running with clocks set to the various clock epochs such as
3940 1902, 1938 or 1970. With this change the logic is now done in PID 1
3941 in addition to timesyncd during early boot-up, so that it is enforced
3942 before the first process is spawned by systemd. Note that the logic
3943 in timesyncd remains, as it is more comprehensive and ensures
3944 clock monotonicity by maintaining a persistent timestamp file in
3945 /var. Since /var is generally not available in earliest boot or the
3946 initrd, this part of the logic remains in timesyncd, and is not done
3949 * Support for tweaking details in net_cls.class_id through the
3950 NetClass= configuration directive has been removed, as the kernel
3951 people have decided to deprecate that controller in cgroup v2.
3952 Userspace tools such as nftables are moving over to setting rules
3953 that are specific to the full cgroup path of a task, which obsoletes
3954 these controllers anyway. The NetClass= directive is kept around for
3955 legacy compatibility reasons. For a more in-depth description of the
3956 kernel change, please refer to the respective upstream commit:
3958 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
3960 * A new service setting RuntimeMaxSec= has been added that may be used
3961 to specify a maximum runtime for a service. If the timeout is hit, the
3962 service is terminated and put into a failure state.
3964 * A new service setting AmbientCapabilities= has been added. It allows
3965 configuration of additional Linux process capabilities that are
3966 passed to the activated processes. This is only available on very
3969 * The process resource limit settings in service units may now be used
3970 to configure hard and soft limits individually.
3972 * The various libsystemd APIs such as sd-bus or sd-event now publicly
3973 expose support for gcc's __attribute__((cleanup())) C extension.
3974 Specifically, for many object destructor functions alternative
3975 versions have been added that have names suffixed with "p" and take a
3976 pointer to a pointer to the object to destroy, instead of just a
3977 pointer to the object itself. This is useful because these destructor
3978 functions may be used directly as parameters to the cleanup
3979 construct. Internally, systemd has been a heavy user of this GCC
3980 extension for a long time, and with this change similar support is
3981 now available to consumers of the library outside of systemd. Note
3982 that by using this extension in your sources compatibility with old
3983 and strictly ANSI compatible C compilers is lost. However, all gcc or
3984 LLVM versions of recent years support this extension.
3986 * Timer units gained support for a new setting RandomizedDelaySec= that
3987 allows configuring some additional randomized delay to the configured
3988 time. This is useful to spread out timer events to avoid load peaks in
3989 clusters or larger setups.
3991 * Calendar time specifications now support sub-second accuracy.
3993 * Socket units now support listening on SCTP and UDP-lite protocol
3996 * The sd-event API now comes with a full set of man pages.
3998 * Older versions of systemd contained experimental support for
3999 compressing journal files and coredumps with the LZ4 compressor that
4000 was not compatible with the lz4 binary (due to API limitations of the
4001 lz4 library). This support has been removed; only support for files
4002 compatible with the lz4 binary remains. This LZ4 logic is now
4003 officially supported and no longer considered experimental.
4005 * The dkr image import logic has been removed again from importd. dkr's
4006 micro-services focus doesn't fit into the machine image focus of
4007 importd, and quickly got out of date with the upstream dkr API.
4009 * Creation of the /run/lock/lockdev/ directory was dropped from
4010 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
4011 been available for many years. If you still need this, you need to
4012 create your own tmpfiles.d config file with:
4014 d /run/lock/lockdev 0775 root lock -
4016 * The settings StartLimitBurst=, StartLimitInterval=, StartLimitAction=
4017 and RebootArgument= have been moved from the [Service] section of
4018 unit files to [Unit], and they are now supported on all unit types,
4019 not just service units. Of course, systemd will continue to
4020 understand these settings also at the old location, in order to
4021 maintain compatibility.
4023 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
4024 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
4025 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
4026 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
4027 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
4028 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
4029 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
4030 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
4031 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
4032 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
4033 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
4034 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
4035 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
4036 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
4037 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
4038 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
4039 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
4040 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
4041 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
4043 — Berlin, 2016-02-11
4047 * A number of properties previously only settable in unit
4048 files are now also available as properties to set when
4049 creating transient units programmatically via the bus, as it
4050 is exposed with systemd-run's --property=
4051 setting. Specifically, these are: SyslogIdentifier=,
4052 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
4053 EnvironmentFile=, ReadWriteDirectories=,
4054 ReadOnlyDirectories=, InaccessibleDirectories=,
4055 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
4057 * When creating transient services via the bus API it is now
4058 possible to pass in a set of file descriptors to use as
4059 STDIN/STDOUT/STDERR for the invoked process.
4061 * Slice units may now be created transiently via the bus APIs,
4062 similar to the way service and scope units may already be
4063 created transiently.
4065 * Wherever systemd expects a calendar timestamp specification
4066 (like in journalctl's --since= and --until= switches) UTC
4067 timestamps are now supported. Timestamps suffixed with "UTC"
4068 are now considered to be in Universal Time Coordinated
4069 instead of the local timezone. Also, timestamps may now
4070 optionally be specified with sub-second accuracy. Both of
4071 these additions also apply to recurring calendar event
4072 specification, such as OnCalendar= in timer units.
4074 * journalctl gained a new "--sync" switch that asks the
4075 journal daemon to write all so far unwritten log messages to
4076 disk and sync the files, before returning.
4078 * systemd-tmpfiles learned two new line types "q" and "Q" that
4079 operate like "v", but also set up a basic btrfs quota
4080 hierarchy when used on a btrfs file system with quota
4083 * tmpfiles' "v", "q" and "Q" will now create a plain directory
4084 instead of a subvolume (even on a btrfs file system) if the
4085 root directory is a plain directory, and not a
4086 subvolume. This should simplify things with certain chroot()
4087 environments which are not aware of the concept of btrfs
4090 * systemd-detect-virt gained a new --chroot switch to detect
4091 whether execution takes place in a chroot() environment.
4093 * CPUAffinity= now takes CPU index ranges in addition to
4096 * The various memory-related resource limit settings (such as
4097 LimitAS=) now understand the usual K, M, G, ... suffixes to
4098 the base of 1024 (IEC). Similar, the time-related resource
4099 limit settings understand the usual min, h, day, ...
4102 * There's a new system.conf setting DefaultTasksMax= to
4103 control the default TasksMax= setting for services and
4104 scopes running on the system. (TasksMax= is the primary
4105 setting that exposes the "pids" cgroup controller on systemd
4106 and was introduced in the previous systemd release.) The
4107 setting now defaults to 512, which means services that are
4108 not explicitly configured otherwise will only be able to
4109 create 512 processes or threads at maximum, from this
4110 version on. Note that this means that thread- or
4111 process-heavy services might need to be reconfigured to set
4112 TasksMax= to a higher value. It is sufficient to set
4113 TasksMax= in these specific unit files to a higher value, or
4114 even "infinity". Similar, there's now a logind.conf setting
4115 UserTasksMax= that defaults to 4096 and limits the total
4116 number of processes or tasks each user may own
4117 concurrently. nspawn containers also have the TasksMax=
4118 value set by default now, to 8192. Note that all of this
4119 only has an effect if the "pids" cgroup controller is
4120 enabled in the kernel. The general benefit of these changes
4121 should be a more robust and safer system, that provides a
4122 certain amount of per-service fork() bomb protection.
4124 * systemd-nspawn gained the new --network-veth-extra= switch
4125 to define additional and arbitrarily-named virtual Ethernet
4126 links between the host and the container.
4128 * A new service execution setting PassEnvironment= has been
4129 added that allows importing select environment variables
4130 from PID1's environment block into the environment block of
4133 * Timer units gained support for a new RemainAfterElapse=
4134 setting which takes a boolean argument. It defaults to on,
4135 exposing behaviour unchanged to previous releases. If set to
4136 off, timer units are unloaded after they elapsed if they
4137 cannot elapse again. This is particularly useful for
4138 transient timer units, which shall not stay around longer
4139 than until they first elapse.
4141 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
4142 default now (the kernel default is 16). This is beneficial
4143 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
4144 allows substantially larger numbers of queued
4145 datagrams. This should increase the capability of systemd to
4146 parallelize boot-up, as logging and sd_notify() are unlikely
4147 to stall execution anymore. If you need to change the value
4148 from the new defaults, use the usual sysctl.d/ snippets.
4150 * The compression framing format used by the journal or
4151 coredump processing has changed to be in line with what the
4152 official LZ4 tools generate. LZ4 compression support in
4153 systemd was considered unsupported previously, as the format
4154 was not compatible with the normal tools. With this release
4155 this has changed now, and it is hence safe for downstream
4156 distributions to turn it on. While not compressing as well
4157 as the XZ, LZ4 is substantially faster, which makes
4158 it a good default choice for the compression logic in the
4159 journal and in coredump handling.
4161 * Any reference to /etc/mtab has been dropped from
4162 systemd. The file has been obsolete since a while, but
4163 systemd refused to work on systems where it was incorrectly
4164 set up (it should be a symlink or non-existent). Please make
4165 sure to update to util-linux 2.27.1 or newer in conjunction
4166 with this systemd release, which also drops any reference to
4167 /etc/mtab. If you maintain a distribution make sure that no
4168 software you package still references it, as this is a
4169 likely source of bugs. There's also a glibc bug pending,
4170 asking for removal of any reference to this obsolete file:
4172 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
4174 Note that only util-linux versions built with
4175 --enable-libmount-force-mountinfo are supported.
4177 * Support for the ".snapshot" unit type has been removed. This
4178 feature turned out to be little useful and little used, and
4179 has now been removed from the core and from systemctl.
4181 * The dependency types RequiresOverridable= and
4182 RequisiteOverridable= have been removed from systemd. They
4183 have been used only very sparingly to our knowledge and
4184 other options that provide a similar effect (such as
4185 systemctl --mode=ignore-dependencies) are much more useful
4186 and commonly used. Moreover, they were only half-way
4187 implemented as the option to control behaviour regarding
4188 these dependencies was never added to systemctl. By removing
4189 these dependency types the execution engine becomes a bit
4190 simpler. Unit files that use these dependencies should be
4191 changed to use the non-Overridable dependency types
4192 instead. In fact, when parsing unit files with these
4193 options, that's what systemd will automatically convert them
4194 too, but it will also warn, asking users to fix the unit
4195 files accordingly. Removal of these dependency types should
4196 only affect a negligible number of unit files in the wild.
4198 * Behaviour of networkd's IPForward= option changed
4199 (again). It will no longer maintain a per-interface setting,
4200 but propagate one way from interfaces where this is enabled
4201 to the global kernel setting. The global setting will be
4202 enabled when requested by a network that is set up, but
4203 never be disabled again. This change was made to make sure
4204 IPv4 and IPv6 behaviour regarding packet forwarding is
4205 similar (as the Linux IPv6 stack does not support
4206 per-interface control of this setting) and to minimize
4209 * In unit files the behaviour of %u, %U, %h, %s has
4210 changed. These specifiers will now unconditionally resolve
4211 to the various user database fields of the user that the
4212 systemd instance is running as, instead of the user
4213 configured in the specific unit via User=. Note that this
4214 effectively doesn't change much, as resolving of these
4215 specifiers was already turned off in the --system instance
4216 of systemd, as we cannot do NSS lookups from PID 1. In the
4217 --user instance of systemd these specifiers where correctly
4218 resolved, but hardly made any sense, since the user instance
4219 lacks privileges to do user switches anyway, and User= is
4220 hence useless. Moreover, even in the --user instance of
4221 systemd behaviour was awkward as it would only take settings
4222 from User= assignment placed before the specifier into
4223 account. In order to unify and simplify the logic around
4224 this the specifiers will now always resolve to the
4225 credentials of the user invoking the manager (which in case
4226 of PID 1 is the root user).
4228 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
4229 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
4230 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
4231 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
4232 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
4233 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
4234 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
4235 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
4236 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
4237 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
4238 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
4239 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
4240 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
4241 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
4244 — Berlin, 2015-11-18
4248 * systemd now depends on util-linux v2.27. More specifically,
4249 the newly added mount monitor feature in libmount now
4250 replaces systemd's former own implementation.
4252 * libmount mandates /etc/mtab not to be regular file, and
4253 systemd now enforces this condition at early boot.
4254 /etc/mtab has been deprecated and warned about for a very
4255 long time, so systems running systemd should already have
4256 stopped having this file around as anything else than a
4257 symlink to /proc/self/mounts.
4259 * Support for the "pids" cgroup controller has been added. It
4260 allows accounting the number of tasks in a cgroup and
4261 enforcing limits on it. This adds two new setting
4262 TasksAccounting= and TasksMax= to each unit, as well as a
4263 global option DefaultTasksAccounting=.
4265 * Support for the "net_cls" cgroup controller has been added.
4266 It allows assigning a net class ID to each task in the
4267 cgroup, which can then be used in firewall rules and traffic
4268 shaping configurations. Note that the kernel netfilter net
4269 class code does not currently work reliably for ingress
4270 packets on unestablished sockets.
4272 This adds a new config directive called NetClass= to CGroup
4273 enabled units. Allowed values are positive numbers for fixed
4274 assignments and "auto" for picking a free value
4277 * 'systemctl is-system-running' now returns 'offline' if the
4278 system is not booted with systemd. This command can now be
4279 used as a substitute for 'systemd-notify --booted'.
4281 * Watchdog timeouts have been increased to 3 minutes for all
4282 in-tree service files. Apparently, disk IO issues are more
4283 frequent than we hoped, and user reported >1 minute waiting
4286 * 'machine-id-commit' functionality has been merged into
4287 'machine-id-setup --commit'. The separate binary has been
4290 * The WorkingDirectory= directive in unit files may now be set
4291 to the special value '~'. In this case, the working
4292 directory is set to the home directory of the user
4293 configured in User=.
4295 * "machinectl shell" will now open the shell in the home
4296 directory of the selected user by default.
4298 * The CrashChVT= configuration file setting is renamed to
4299 CrashChangeVT=, following our usual logic of not
4300 abbreviating unnecessarily. The old directive is still
4301 supported for compat reasons. Also, this directive now takes
4302 an integer value between 1 and 63, or a boolean value. The
4303 formerly supported '-1' value for disabling stays around for
4306 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
4307 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
4308 RootDirectory= properties can now be set for transient
4311 * The systemd-analyze tool gained a new "set-log-target" verb
4312 to change the logging target the system manager logs to
4313 dynamically during runtime. This is similar to how
4314 "systemd-analyze set-log-level" already changes the log
4317 * In nspawn /sys is now mounted as tmpfs, with only a selected
4318 set of subdirectories mounted in from the real sysfs. This
4319 enhances security slightly, and is useful for ensuring user
4320 namespaces work correctly.
4322 * Support for USB FunctionFS activation has been added. This
4323 allows implementation of USB gadget services that are
4324 activated as soon as they are requested, so that they don't
4325 have to run continuously, similar to classic socket
4328 * The "systemctl exit" command now optionally takes an
4329 additional parameter that sets the exit code to return from
4330 the systemd manager when exiting. This is only relevant when
4331 running the systemd user instance, or when running the
4332 system instance in a container.
4334 * sd-bus gained the new API calls sd_bus_path_encode_many()
4335 and sd_bus_path_decode_many() that allow easy encoding and
4336 decoding of multiple identifier strings inside a D-Bus
4337 object path. Another new call sd_bus_default_flush_close()
4338 has been added to flush and close per-thread default
4341 * systemd-cgtop gained support for a -M/--machine= switch to
4342 show the control groups within a certain container only.
4344 * "systemctl kill" gained support for an optional --fail
4345 switch. If specified the requested operation will fail of no
4346 processes have been killed, because the unit had no
4347 processes attached, or similar.
4349 * A new systemd.crash_reboot=1 kernel command line option has
4350 been added that triggers a reboot after crashing. This can
4351 also be set through CrashReboot= in systemd.conf.
4353 * The RuntimeDirectory= setting now understands unit
4354 specifiers like %i or %f.
4356 * A new (still internal) library API sd-ipv4acd has been added,
4357 that implements address conflict detection for IPv4. It's
4358 based on code from sd-ipv4ll, and will be useful for
4359 detecting DHCP address conflicts.
4361 * File descriptors passed during socket activation may now be
4362 named. A new API sd_listen_fds_with_names() is added to
4363 access the names. The default names may be overridden,
4364 either in the .socket file using the FileDescriptorName=
4365 parameter, or by passing FDNAME= when storing the file
4366 descriptors using sd_notify().
4368 * systemd-networkd gained support for:
4370 - Setting the IPv6 Router Advertisement settings via
4371 IPv6AcceptRouterAdvertisements= in .network files.
4373 - Configuring the HelloTimeSec=, MaxAgeSec= and
4374 ForwardDelaySec= bridge parameters in .netdev files.
4376 - Configuring PreferredSource= for static routes in
4379 * The "ask-password" framework used to query for LUKS harddisk
4380 passwords or SSL passwords during boot gained support for
4381 caching passwords in the kernel keyring, if it is
4382 available. This makes sure that the user only has to type in
4383 a passphrase once if there are multiple objects to unlock
4384 with the same one. Previously, such password caching was
4385 available only when Plymouth was used; this moves the
4386 caching logic into the systemd codebase itself. The
4387 "systemd-ask-password" utility gained a new --keyname=
4388 switch to control which kernel keyring key to use for
4389 caching a password in. This functionality is also useful for
4390 enabling display managers such as gdm to automatically
4391 unlock the user's GNOME keyring if its passphrase, the
4392 user's password and the harddisk password are the same, if
4393 gdm-autologin is used.
4395 * When downloading tar or raw images using "machinectl
4396 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
4397 file is now also downloaded, if it is available and stored
4398 next to the image file.
4400 * Units of type ".socket" gained a new boolean setting
4401 Writable= which is only useful in conjunction with
4402 ListenSpecial=. If true, enables opening the specified
4403 special file in O_RDWR mode rather than O_RDONLY mode.
4405 * systemd-rfkill has been reworked to become a singleton
4406 service that is activated through /dev/rfkill on each rfkill
4407 state change and saves the settings to disk. This way,
4408 systemd-rfkill is now compatible with devices that exist
4409 only intermittendly, and even restores state if the previous
4410 system shutdown was abrupt rather than clean.
4412 * The journal daemon gained support for vacuuming old journal
4413 files controlled by the number of files that shall remain,
4414 in addition to the already existing control by size and by
4415 date. This is useful as journal interleaving performance
4416 degrades with too many separate journal files, and allows
4417 putting an effective limit on them. The new setting defaults
4418 to 100, but this may be changed by setting SystemMaxFiles=
4419 and RuntimeMaxFiles= in journald.conf. Also, the
4420 "journalctl" tool gained the new --vacuum-files= switch to
4421 manually vacuum journal files to leave only the specified
4422 number of files in place.
4424 * udev will now create /dev/disk/by-path links for ATA devices
4425 on kernels where that is supported.
4427 * Galician, Serbian, Turkish and Korean translations were added.
4429 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
4430 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
4431 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
4432 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
4433 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
4434 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
4435 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
4436 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
4437 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
4438 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
4439 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
4440 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
4441 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
4442 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
4443 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
4444 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
4445 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
4446 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
4448 — Berlin, 2015-10-07
4452 * The DHCP implementation of systemd-networkd gained a set of
4455 - The DHCP server now supports emitting DNS and NTP
4456 information. It may be enabled and configured via
4457 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
4458 and NTP information is enabled, but no servers are
4459 configured, the corresponding uplink information (if there
4460 is any) is propagated.
4462 - Server and client now support transmission and reception
4463 of timezone information. It can be configured via the
4464 newly introduced network options UseTimezone=,
4465 EmitTimezone=, and Timezone=. Transmission of timezone
4466 information is enabled between host and containers by
4467 default now: the container will change its local timezone
4468 to what the host has set.
4470 - Lease timeouts can now be configured via
4471 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
4473 - The DHCP server improved on the stability of
4474 leases. Clients are more likely to get the same lease
4475 information back, even if the server loses state.
4477 - The DHCP server supports two new configuration options to
4478 control the lease address pool metrics, PoolOffset= and
4481 * The encapsulation limit of tunnels in systemd-networkd may
4482 now be configured via 'EncapsulationLimit='. It allows
4483 modifying the maximum additional levels of encapsulation
4484 that are permitted to be prepended to a packet.
4486 * systemd now supports the concept of user buses replacing
4487 session buses, if used with dbus-1.10 (and enabled via dbus
4488 --enable-user-session). It previously only supported this on
4489 kdbus-enabled systems, and this release expands this to
4490 'dbus-daemon' systems.
4492 * systemd-networkd now supports predictable interface names
4495 * systemd now optionally supports the new Linux kernel
4496 "unified" control group hierarchy. If enabled via the kernel
4497 command-line option 'systemd.unified_cgroup_hierarchy=1',
4498 systemd will try to mount the unified cgroup hierarchy
4499 directly on /sys/fs/cgroup. If not enabled, or not
4500 available, systemd will fall back to the legacy cgroup
4501 hierarchy setup, as before. Host system and containers can
4502 mix and match legacy and unified hierarchies as they
4503 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
4504 environment variable to individually select the hierarchy to
4505 use for executed containers. By default, nspawn will use the
4506 unified hierarchy for the containers if the host uses the
4507 unified hierarchy, and the legacy hierarchy otherwise.
4508 Please note that at this point the unified hierarchy is an
4509 experimental kernel feature and is likely to change in one
4510 of the next kernel releases. Therefore, it should not be
4511 enabled by default in downstream distributions yet. The
4512 minimum required kernel version for the unified hierarchy to
4513 work is 4.2. Note that when the unified hierarchy is used
4514 for the first time delegated access to controllers is
4515 safe. Because of this systemd-nspawn containers will get
4516 access to controllers now, as will systemd user
4517 sessions. This means containers and user sessions may now
4518 manage their own resources, partitioning up what the system
4521 * A new special scope unit "init.scope" has been introduced
4522 that encapsulates PID 1 of the system. It may be used to
4523 determine resource usage and enforce resource limits on PID
4524 1 itself. PID 1 hence moved out of the root of the control
4527 * The cgtop tool gained support for filtering out kernel
4528 threads when counting tasks in a control group. Also, the
4529 count of processes is now recursively summed up by
4530 default. Two options -k and --recursive= have been added to
4531 revert to old behaviour. The tool has also been updated to
4532 work correctly in containers now.
4534 * systemd-nspawn's --bind= and --bind-ro= options have been
4535 extended to allow creation of non-recursive bind mounts.
4537 * libsystemd gained two new calls sd_pid_get_cgroup() and
4538 sd_peer_get_cgroup() which return the control group path of
4539 a process or peer of a connected AF_UNIX socket. This
4540 function call is particularly useful when implementing
4541 delegated subtrees support in the control group hierarchy.
4543 * The "sd-event" event loop API of libsystemd now supports
4544 correct dequeuing of real-time signals, without losing
4547 * When systemd requests a polkit decision when managing units it
4548 will now add additional fields to the request, including unit
4549 name and desired operation. This enables more powerful polkit
4550 policies, that make decisions depending on these parameters.
4552 * nspawn learnt support for .nspawn settings files, that may
4553 accompany the image files or directories of containers, and
4554 may contain additional settings for the container. This is
4555 an alternative to configuring container parameters via the
4556 nspawn command line.
4558 Contributions from: Cristian Rodríguez, Daniel Mack, David
4559 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
4560 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
4561 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
4562 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
4563 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
4564 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
4565 Andersen, Tom Gundersen, Torstein Husebø
4567 — Berlin, 2015-09-08
4571 * machinectl gained a new verb 'shell' which opens a fresh
4572 shell on the target container or the host. It is similar to
4573 the existing 'login' command of machinectl, but spawns the
4574 shell directly without prompting for username or
4575 password. The pseudo machine '.host' now refers to the local
4576 host and is used by default. Hence, 'machinectl shell' can
4577 be used as replacement for 'su -' which spawns a session as
4578 a fresh systemd unit in a way that is fully isolated from
4579 the originating session.
4581 * systemd-networkd learned to cope with private-zone DHCP
4582 options and allows other programs to query the values.
4584 * SELinux access control when enabling/disabling units is no
4585 longer enforced with this release. The previous implementation
4586 was incorrect, and a new corrected implementation is not yet
4587 available. As unit file operations are still protected via
4588 polkit and D-Bus policy this is not a security problem. Yet,
4589 distributions which care about optimal SELinux support should
4590 probably not stabilize on this release.
4592 * sd-bus gained support for matches of type "arg0has=", that
4593 test for membership of strings in string arrays sent in bus
4596 * systemd-resolved now dumps the contents of its DNS and LLMNR
4597 caches to the logs on reception of the SIGUSR1 signal. This
4598 is useful to debug DNS behaviour.
4600 * The coredumpctl tool gained a new --directory= option to
4601 operate on journal files in a specific directory.
4603 * "systemctl reboot" and related commands gained a new
4604 "--message=" option which may be used to set a free-text
4605 wall message when shutting down or rebooting the
4606 system. This message is also logged, which is useful for
4607 figuring out the reason for a reboot or shutdown a
4610 * The "systemd-resolve-host" tool's -i switch now takes
4611 network interface numbers as alternative to interface names.
4613 * A new unit file setting for services has been introduced:
4614 UtmpMode= allows configuration of how precisely systemd
4615 handles utmp and wtmp entries for the service if this is
4616 enabled. This allows writing services that appear similar to
4617 user sessions in the output of the "w", "who", "last" and
4620 * systemd-resolved will now locally synthesize DNS resource
4621 records for the "localhost" and "gateway" domains as well as
4622 the local hostname. This should ensure that clients querying
4623 RRs via resolved will get similar results as those going via
4624 NSS, if nss-myhostname is enabled.
4626 Contributions from: Alastair Hughes, Alex Crawford, Daniel
4627 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
4628 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
4629 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
4630 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
4631 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
4632 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
4633 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
4634 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
4635 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
4636 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
4637 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
4639 — Berlin, 2015-08-27
4643 * The systemd-efi-boot-generator functionality was merged into
4644 systemd-gpt-auto-generator.
4646 * systemd-networkd now supports Group Policy for vxlan
4647 devices. It can be enabled via the new boolean configuration
4648 option called 'GroupPolicyExtension='.
4650 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
4651 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
4652 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
4654 — Berlin, 2015-07-31
4658 * The python-systemd code has been removed from the systemd repository.
4659 A new repository has been created which accommodates the code from
4660 now on, and we kindly ask distributions to create a separate package
4661 for this: https://github.com/systemd/python-systemd
4663 * The systemd daemon will now reload its main configuration
4664 (/etc/systemd/system.conf) on daemon-reload.
4666 * sd-dhcp now exposes vendor specific extensions via
4667 sd_dhcp_lease_get_vendor_specific().
4669 * systemd-networkd gained a number of new configuration options.
4671 - A new boolean configuration option for TAP devices called
4672 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
4673 device, thus allowing to send and receive GSO packets.
4675 - A new tunnel configuration option called 'CopyDSCP='.
4676 If enabled, the DSCP field of ip6 tunnels is copied into the
4677 decapsulated packet.
4679 - A set of boolean bridge configuration options were added.
4680 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
4681 and 'UnicastFlood=' are now parsed by networkd and applied to the
4682 respective bridge link device via the respective IFLA_BRPORT_*
4685 - A new string configuration option to override the hostname sent
4686 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
4687 is true, networkd will use the configured hostname instead of the
4688 system hostname when sending DHCP requests.
4690 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
4691 networkd will configure the IPv6 flow-label of the tunnel device
4692 according to RFC2460.
4694 - The 'macvtap' virtual network devices are now supported, similar to
4695 the already supported 'macvlan' devices.
4697 * systemd-resolved now implements RFC5452 to improve resilience against
4698 cache poisoning. Additionally, source port randomization is enabled
4699 by default to further protect against DNS spoofing attacks.
4701 * nss-mymachines now supports translating UIDs and GIDs of running
4702 containers with user-namespaces enabled. If a container 'foo'
4703 translates a host uid 'UID' to the container uid 'TUID', then
4704 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
4705 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
4706 mapped as 'vg-foo-TGID'.
4708 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
4709 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
4710 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
4711 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
4712 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
4713 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
4714 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
4715 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
4716 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
4717 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
4719 — Berlin, 2015-07-29
4723 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
4724 There are no known issues with current sysfs, and udev does not need
4725 or should be used to work around such bugs.
4727 * udev does no longer enable USB HID power management. Several reports
4728 indicate, that some devices cannot handle that setting.
4730 * The udev accelerometer helper was removed. The functionality
4731 is now fully included in iio-sensor-proxy. But this means,
4732 older iio-sensor-proxy versions will no longer provide
4733 accelerometer/orientation data with this systemd version.
4734 Please upgrade iio-sensor-proxy to version 1.0.
4736 * networkd gained a new configuration option IPv6PrivacyExtensions=
4737 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
4738 for Stateless Address") on selected networks.
4740 * For the sake of fewer build-time dependencies and less code in the
4741 main repository, the python bindings are about to be removed in the
4742 next release. A new repository has been created which accommodates
4743 the code from now on, and we kindly ask distributions to create a
4744 separate package for this. The removal will take place in v223.
4746 https://github.com/systemd/python-systemd
4748 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
4749 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
4750 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
4751 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
4752 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
4753 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
4754 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
4755 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
4756 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
4757 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
4759 — Berlin, 2015-07-07
4763 * The sd-bus.h and sd-event.h APIs have now been declared
4764 stable and have been added to the official interface of
4765 libsystemd.so. sd-bus implements an alternative D-Bus client
4766 library, that is relatively easy to use, very efficient and
4767 supports both classic D-Bus as well as kdbus as transport
4768 backend. sd-event is a generic event loop abstraction that
4769 is built around Linux epoll, but adds features such as event
4770 prioritization or efficient timer handling. Both APIs are good
4771 choices for C programs looking for a bus and/or event loop
4772 implementation that is minimal and does not have to be
4773 portable to other kernels.
4775 * kdbus support is no longer compile-time optional. It is now
4776 always built-in. However, it can still be disabled at
4777 runtime using the kdbus=0 kernel command line setting, and
4778 that setting may be changed to default to off, by specifying
4779 --disable-kdbus at build-time. Note though that the kernel
4780 command line setting has no effect if the kdbus.ko kernel
4781 module is not installed, in which case kdbus is (obviously)
4782 also disabled. We encourage all downstream distributions to
4783 begin testing kdbus by adding it to the kernel images in the
4784 development distributions, and leaving kdbus support in
4787 * The minimal required util-linux version has been bumped to
4790 * Support for chkconfig (--enable-chkconfig) was removed in
4791 favor of calling an abstraction tool
4792 /lib/systemd/systemd-sysv-install. This needs to be
4793 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
4794 in README for details.
4796 * If there's a systemd unit and a SysV init script for the
4797 same service name, and the user executes "systemctl enable"
4798 for it (or a related call), then this will now enable both
4799 (or execute the related operation on both), not just the
4802 * The libudev API documentation has been converted from gtkdoc
4805 * gudev has been removed from the systemd tree, it is now an
4808 * The systemd-cgtop tool learnt a new --raw switch to generate
4809 "raw" (machine parsable) output.
4811 * networkd's IPForwarding= .network file setting learnt the
4812 new setting "kernel", which ensures that networkd does not
4813 change the IP forwarding sysctl from the default kernel
4816 * The systemd-logind bus API now exposes a new boolean
4817 property "Docked" that reports whether logind considers the
4818 system "docked", i.e. connected to a docking station or not.
4820 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
4821 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
4822 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
4823 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
4824 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
4825 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
4826 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
4827 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
4828 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
4829 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
4830 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
4831 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
4832 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
4833 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
4834 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
4835 Fink, Zbigniew Jędrzejewski-Szmek
4837 — Berlin, 2015-06-19
4841 * The gudev library has been extracted into a separate repository
4842 available at: https://git.gnome.org/browse/libgudev/
4843 It is now managed as part of the Gnome project. Distributions
4844 are recommended to pass --disable-gudev to systemd and use
4845 gudev from the Gnome project instead. gudev is still included
4846 in systemd, for now. It will be removed soon, though. Please
4847 also see the announcement-thread on systemd-devel:
4848 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
4850 * systemd now exposes a CPUUsageNSec= property for each
4851 service unit on the bus, that contains the overall consumed
4852 CPU time of a service (the sum of what each process of the
4853 service consumed). This value is only available if
4854 CPUAccounting= is turned on for a service, and is then shown
4855 in the "systemctl status" output.
4857 * Support for configuring alternative mappings of the old SysV
4858 runlevels to systemd targets has been removed. They are now
4859 hardcoded in a way that runlevels 2, 3, 4 all map to
4860 multi-user.target and 5 to graphical.target (which
4861 previously was already the default behaviour).
4863 * The auto-mounter logic gained support for mount point
4864 expiry, using a new TimeoutIdleSec= setting in .automount
4865 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
4867 * The EFI System Partition (ESP) as mounted to /boot by
4868 systemd-efi-boot-generator will now be unmounted
4869 automatically after 2 minutes of not being used. This should
4870 minimize the risk of ESP corruptions.
4872 * New /etc/fstab options x-systemd.requires= and
4873 x-systemd.requires-mounts-for= are now supported to express
4874 additional dependencies for mounts. This is useful for
4875 journalling file systems that support external journal
4876 devices or overlay file systems that require underlying file
4877 systems to be mounted.
4879 * systemd does not support direct live-upgrades (via systemctl
4880 daemon-reexec) from versions older than v44 anymore. As no
4881 distribution we are aware of shipped such old versions in a
4882 stable release this should not be problematic.
4884 * When systemd forks off a new per-connection service instance
4885 it will now set the $REMOTE_ADDR environment variable to the
4886 remote IP address, and $REMOTE_PORT environment variable to
4887 the remote IP port. This behaviour is similar to the
4888 corresponding environment variables defined by CGI.
4890 * systemd-networkd gained support for uplink failure
4891 detection. The BindCarrier= option allows binding interface
4892 configuration dynamically to the link sense of other
4893 interfaces. This is useful to achieve behaviour like in
4896 * systemd-networkd gained support for configuring the DHCP
4897 client identifier to use when requesting leases.
4899 * systemd-networkd now has a per-network UseNTP= option to
4900 configure whether NTP server information acquired via DHCP
4901 is passed on to services like systemd-timesyncd.
4903 * systemd-networkd gained support for vti6 tunnels.
4905 * Note that systemd-networkd manages the sysctl variable
4906 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
4907 it is configured for since v219. The variable controls IP
4908 forwarding, and is a per-interface alternative to the global
4909 /proc/sys/net/ipv[46]/ip_forward. This setting is
4910 configurable in the IPForward= option, which defaults to
4911 "no". This means if networkd is used for an interface it is
4912 no longer sufficient to set the global sysctl option to turn
4913 on IP forwarding! Instead, the .network file option
4914 IPForward= needs to be turned on! Note that the
4915 implementation of this behaviour was broken in v219 and has
4918 * Many bonding and vxlan options are now configurable in
4921 * systemd-nspawn gained a new --property= setting to set unit
4922 properties for the container scope. This is useful for
4923 setting resource parameters (e.g. "CPUShares=500") on
4924 containers started from the command line.
4926 * systemd-nspawn gained a new --private-users= switch to make
4927 use of user namespacing available on recent Linux kernels.
4929 * systemd-nspawn may now be called as part of a shell pipeline
4930 in which case the pipes used for stdin and stdout are passed
4931 directly to the process invoked in the container, without
4932 indirection via a pseudo tty.
4934 * systemd-nspawn gained a new switch to control the UNIX
4935 signal to use when killing the init process of the container
4938 * systemd-nspawn gained a new --overlay= switch for mounting
4939 overlay file systems into the container using the new kernel
4942 * When a container image is imported via systemd-importd and
4943 the host file system is not btrfs, a loopback block device
4944 file is created in /var/lib/machines.raw with a btrfs file
4945 system inside. It is then mounted to /var/lib/machines to
4946 enable btrfs features for container management. The loopback
4947 file and btrfs file system is grown as needed when container
4948 images are imported via systemd-importd.
4950 * systemd-machined/systemd-importd gained support for btrfs
4951 quota, to enforce container disk space limits on disk. This
4952 is exposed in "machinectl set-limit".
4954 * systemd-importd now can import containers from local .tar,
4955 .raw and .qcow2 images, and export them to .tar and .raw. It
4956 can also import dkr v2 images now from the network (on top
4959 * systemd-importd gained support for verifying downloaded
4960 images with gpg2 (previously only gpg1 was supported).
4962 * systemd-machined, systemd-logind, systemd: most bus calls are
4963 now accessible to unprivileged processes via polkit. Also,
4964 systemd-logind will now allow users to kill their own sessions
4965 without further privileges or authorization.
4967 * systemd-shutdownd has been removed. This service was
4968 previously responsible for implementing scheduled shutdowns
4969 as exposed in /usr/bin/shutdown's time parameter. This
4970 functionality has now been moved into systemd-logind and is
4971 accessible via a bus interface.
4973 * "systemctl reboot" gained a new switch --firmware-setup that
4974 can be used to reboot into the EFI firmware setup, if that
4975 is available. systemd-logind now exposes an API on the bus
4976 to trigger such reboots, in case graphical desktop UIs want
4977 to cover this functionality.
4979 * "systemctl enable", "systemctl disable" and "systemctl mask"
4980 now support a new "--now" switch. If specified the units
4981 that are enabled will also be started, and the ones
4982 disabled/masked also stopped.
4984 * The Gummiboot EFI boot loader tool has been merged into
4985 systemd, and renamed to "systemd-boot". The bootctl tool has been
4986 updated to support systemd-boot.
4988 * An EFI kernel stub has been added that may be used to create
4989 kernel EFI binaries that contain not only the actual kernel,
4990 but also an initrd, boot splash, command line and OS release
4991 information. This combined binary can then be signed as a
4992 single image, so that the firmware can verify it all in one
4993 step. systemd-boot has special support for EFI binaries created
4994 like this and can extract OS release information from them
4995 and show them in the boot menu. This functionality is useful
4996 to implement cryptographically verified boot schemes.
4998 * Optional support has been added to systemd-fsck to pass
4999 fsck's progress report to an AF_UNIX socket in the file
5002 * udev will no longer create device symlinks for all block
5003 devices by default. A blacklist for excluding special block
5004 devices from this logic has been turned into a whitelist
5005 that requires picking block devices explicitly that require
5008 * A new (currently still internal) API sd-device.h has been
5009 added to libsystemd. This modernized API is supposed to
5010 replace libudev eventually. In fact, already much of libudev
5011 is now just a wrapper around sd-device.h.
5013 * A new hwdb database for storing metadata about pointing
5014 stick devices has been added.
5016 * systemd-tmpfiles gained support for setting file attributes
5017 similar to the "chattr" tool with new 'h' and 'H' lines.
5019 * systemd-journald will no longer unconditionally set the
5020 btrfs NOCOW flag on new journal files. This is instead done
5021 with tmpfiles snippet using the new 'h' line type. This
5022 allows easy disabling of this logic, by masking the
5023 journal-nocow.conf tmpfiles file.
5025 * systemd-journald will now translate audit message types to
5026 human readable identifiers when writing them to the
5027 journal. This should improve readability of audit messages.
5029 * The LUKS logic gained support for the offset= and skip=
5030 options in /etc/crypttab, as previously implemented by
5033 * /usr/lib/os-release gained a new optional field VARIANT= for
5034 distributions that support multiple variants (such as a
5035 desktop edition, a server edition, ...)
5037 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
5038 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
5039 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
5040 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
5041 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
5042 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
5043 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
5044 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
5045 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
5046 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
5047 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
5048 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
5049 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
5050 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
5051 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
5052 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
5053 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
5054 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
5055 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
5056 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
5057 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
5058 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
5059 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
5060 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
5061 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
5062 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
5063 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
5065 — Berlin, 2015-05-22
5069 * Introduce a new API "sd-hwdb.h" for querying the hardware
5070 metadata database. With this minimal interface one can query
5071 and enumerate the udev hwdb, decoupled from the old libudev
5072 library. libudev's interface for this is now only a wrapper
5073 around sd-hwdb. A new tool systemd-hwdb has been added to
5074 interface with and update the database.
5076 * When any of systemd's tools copies files (for example due to
5077 tmpfiles' C lines) a btrfs reflink will attempted first,
5078 before bytewise copying is done.
5080 * systemd-nspawn gained a new --ephemeral switch. When
5081 specified a btrfs snapshot is taken of the container's root
5082 directory, and immediately removed when the container
5083 terminates again. Thus, a container can be started whose
5084 changes never alter the container's root directory, and are
5085 lost on container termination. This switch can also be used
5086 for starting a container off the root file system of the
5087 host without affecting the host OS. This switch is only
5088 available on btrfs file systems.
5090 * systemd-nspawn gained a new --template= switch. It takes the
5091 path to a container tree to use as template for the tree
5092 specified via --directory=, should that directory be
5093 missing. This allows instantiating containers dynamically,
5094 on first run. This switch is only available on btrfs file
5097 * When a .mount unit refers to a mount point on which multiple
5098 mounts are stacked, and the .mount unit is stopped all of
5099 the stacked mount points will now be unmounted until no
5100 mount point remains.
5102 * systemd now has an explicit notion of supported and
5103 unsupported unit types. Jobs enqueued for unsupported unit
5104 types will now fail with an "unsupported" error code. More
5105 specifically .swap, .automount and .device units are not
5106 supported in containers, .busname units are not supported on
5107 non-kdbus systems. .swap and .automount are also not
5108 supported if their respective kernel compile time options
5111 * machinectl gained support for two new "copy-from" and
5112 "copy-to" commands for copying files from a running
5113 container to the host or vice versa.
5115 * machinectl gained support for a new "bind" command to bind
5116 mount host directories into local containers. This is
5117 currently only supported for nspawn containers.
5119 * networkd gained support for configuring bridge forwarding
5120 database entries (fdb) from .network files.
5122 * A new tiny daemon "systemd-importd" has been added that can
5123 download container images in tar, raw, qcow2 or dkr formats,
5124 and make them available locally in /var/lib/machines, so
5125 that they can run as nspawn containers. The daemon can GPG
5126 verify the downloads (not supported for dkr, since it has no
5127 provisions for verifying downloads). It will transparently
5128 decompress bz2, xz, gzip compressed downloads if necessary,
5129 and restore sparse files on disk. The daemon uses privilege
5130 separation to ensure the actual download logic runs with
5131 fewer privileges than the daemon itself. machinectl has
5132 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
5133 make the functionality of importd available to the
5134 user. With this in place the Fedora and Ubuntu "Cloud"
5135 images can be downloaded and booted as containers unmodified
5136 (the Fedora images lack the appropriate GPG signature files
5137 currently, so they cannot be verified, but this will change
5138 soon, hopefully). Note that downloading images is currently
5139 only fully supported on btrfs.
5141 * machinectl is now able to list container images found in
5142 /var/lib/machines, along with some metadata about sizes of
5143 disk and similar. If the directory is located on btrfs and
5144 quota is enabled, this includes quota display. A new command
5145 "image-status" has been added that shows additional
5146 information about images.
5148 * machinectl is now able to clone container images
5149 efficiently, if the underlying file system (btrfs) supports
5150 it, with the new "machinectl clone" command. It also
5151 gained commands for renaming and removing images, as well as
5152 marking them read-only or read-write (supported also on
5153 legacy file systems).
5155 * networkd gained support for collecting LLDP network
5156 announcements, from hardware that supports this. This is
5157 shown in networkctl output.
5159 * systemd-run gained support for a new -t (--pty) switch for
5160 invoking a binary on a pty whose input and output is
5161 connected to the invoking terminal. This allows executing
5162 processes as system services while interactively
5163 communicating with them via the terminal. Most interestingly
5164 this is supported across container boundaries. Invoking
5165 "systemd-run -t /bin/bash" is an alternative to running a
5166 full login session, the difference being that the former
5167 will not register a session, nor go through the PAM session
5170 * tmpfiles gained support for a new "v" line type for creating
5171 btrfs subvolumes. If the underlying file system is a legacy
5172 file system, this automatically degrades to creating a
5173 normal directory. Among others /var/lib/machines is now
5174 created like this at boot, should it be missing.
5176 * The directory /var/lib/containers/ has been deprecated and
5177 been replaced by /var/lib/machines. The term "machines" has
5178 been used in the systemd context as generic term for both
5179 VMs and containers, and hence appears more appropriate for
5180 this, as the directory can also contain raw images bootable
5183 * systemd-nspawn when invoked with -M but without --directory=
5184 or --image= is now capable of searching for the container
5185 root directory, subvolume or disk image automatically, in
5186 /var/lib/machines. systemd-nspawn@.service has been updated
5187 to make use of this, thus allowing it to be used for raw
5190 * A new machines.target unit has been introduced that is
5191 supposed to group all containers/VMs invoked as services on
5192 the system. systemd-nspawn@.service has been updated to
5193 integrate with that.
5195 * machinectl gained a new "start" command, for invoking a
5196 container as a service. "machinectl start foo" is mostly
5197 equivalent to "systemctl start systemd-nspawn@foo.service",
5198 but handles escaping in a nicer way.
5200 * systemd-nspawn will now mount most of the cgroupfs tree
5201 read-only into each container, with the exception of the
5202 container's own subtree in the name=systemd hierarchy.
5204 * journald now sets the special FS_NOCOW file flag for its
5205 journal files. This should improve performance on btrfs, by
5206 avoiding heavy fragmentation when journald's write-pattern
5207 is used on COW file systems. It degrades btrfs' data
5208 integrity guarantees for the files to the same levels as for
5209 ext3/ext4 however. This should be OK though as journald does
5210 its own data integrity checks and all its objects are
5211 checksummed on disk. Also, journald should handle btrfs disk
5212 full events a lot more gracefully now, by processing SIGBUS
5213 errors, and not relying on fallocate() anymore.
5215 * When journald detects that journal files it is writing to
5216 have been deleted it will immediately start new journal
5219 * systemd now provides a way to store file descriptors
5220 per-service in PID 1. This is useful for daemons to ensure
5221 that fds they require are not lost during a daemon
5222 restart. The fds are passed to the daemon on the next
5223 invocation in the same way socket activation fds are
5224 passed. This is now used by journald to ensure that the
5225 various sockets connected to all the system's stdout/stderr
5226 are not lost when journald is restarted. File descriptors
5227 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
5228 an extension to sd_notify(). Note that a limit is enforced
5229 on the number of fds a service can store in PID 1, and it
5230 defaults to 0, so that no fds may be stored, unless this is
5231 explicitly turned on.
5233 * The default TERM variable to use for units connected to a
5234 terminal, when no other value is explicitly is set is now
5235 vt220 rather than vt102. This should be fairly safe still,
5236 but allows PgUp/PgDn work.
5238 * The /etc/crypttab option header= as known from Debian is now
5241 * "loginctl user-status" and "loginctl session-status" will
5242 now show the last 10 lines of log messages of the
5243 user/session following the status output. Similar,
5244 "machinectl status" will show the last 10 log lines
5245 associated with a virtual machine or container
5246 service. (Note that this is usually not the log messages
5247 done in the VM/container itself, but simply what the
5248 container manager logs. For nspawn this includes all console
5251 * "loginctl session-status" without further argument will now
5252 show the status of the session of the caller. Similar,
5253 "lock-session", "unlock-session", "activate",
5254 "enable-linger", "disable-linger" may now be called without
5255 session/user parameter in which case they apply to the
5256 caller's session/user.
5258 * An X11 session scriptlet is now shipped that uploads
5259 $DISPLAY and $XAUTHORITY into the environment of the systemd
5260 --user daemon if a session begins. This should improve
5261 compatibility with X11 enabled applications run as systemd
5264 * Generators are now subject to masking via /etc and /run, the
5265 same way as unit files.
5267 * networkd .network files gained support for configuring
5268 per-link IPv4/IPv6 packet forwarding as well as IPv4
5269 masquerading. This is by default turned on for veth links to
5270 containers, as registered by systemd-nspawn. This means that
5271 nspawn containers run with --network-veth will now get
5272 automatic routed access to the host's networks without any
5273 further configuration or setup, as long as networkd runs on
5276 * systemd-nspawn gained the --port= (-p) switch to expose TCP
5277 or UDP posts of a container on the host. With this in place
5278 it is possible to run containers with private veth links
5279 (--network-veth), and have their functionality exposed on
5280 the host as if their services were running directly on the
5283 * systemd-nspawn's --network-veth switch now gained a short
5284 version "-n", since with the changes above it is now truly
5285 useful out-of-the-box. The systemd-nspawn@.service has been
5286 updated to make use of it too by default.
5288 * systemd-nspawn will now maintain a per-image R/W lock, to
5289 ensure that the same image is not started more than once
5290 writable. (It's OK to run an image multiple times
5291 simultaneously in read-only mode.)
5293 * systemd-nspawn's --image= option is now capable of
5294 dissecting and booting MBR and GPT disk images that contain
5295 only a single active Linux partition. Previously it
5296 supported only GPT disk images with proper GPT type
5297 IDs. This allows running cloud images from major
5298 distributions directly with systemd-nspawn, without
5301 * In addition to collecting mouse dpi data in the udev
5302 hardware database, there's now support for collecting angle
5303 information for mouse scroll wheels. The database is
5304 supposed to guarantee similar scrolling behavior on mice
5305 that it knows about. There's also support for collecting
5306 information about Touchpad types.
5308 * udev's input_id built-in will now also collect touch screen
5309 dimension data and attach it to probed devices.
5311 * /etc/os-release gained support for a Distribution Privacy
5314 * networkd gained support for creating "ipvlan", "gretap",
5315 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
5317 * systemd-tmpfiles gained support for "a" lines for setting
5320 * systemd-nspawn will now mount /tmp in the container to
5321 tmpfs, automatically.
5323 * systemd now exposes the memory.usage_in_bytes cgroup
5324 attribute and shows it for each service in the "systemctl
5325 status" output, if available.
5327 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
5328 immediate reboot is triggered. This useful if shutdown is
5329 hung and is unable to complete, to expedite the
5330 operation. Note that this kind of reboot will still unmount
5331 all file systems, and hence should not result in fsck being
5334 * A .device unit for an optical block device will now be
5335 considered active only when a medium is in the drive. Also,
5336 mount units are now bound to their backing devices thus
5337 triggering automatic unmounting when devices become
5338 unavailable. With this in place systemd will now
5339 automatically unmount left-over mounts when a CD-ROM is
5340 ejected or an USB stick is yanked from the system.
5342 * networkd-wait-online now has support for waiting for
5343 specific interfaces only (with globbing), and for giving up
5344 after a configurable timeout.
5346 * networkd now exits when idle. It will be automatically
5347 restarted as soon as interfaces show up, are removed or
5348 change state. networkd will stay around as long as there is
5349 at least one DHCP state machine or similar around, that keep
5352 * networkd may now configure IPv6 link-local addressing in
5353 addition to IPv4 link-local addressing.
5355 * The IPv6 "token" for use in SLAAC may now be configured for
5356 each .network interface in networkd.
5358 * Routes configured with networkd may now be assigned a scope
5361 * networkd's [Match] sections now support globbing and lists
5362 of multiple space-separated matches per item.
5364 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
5365 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
5366 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
5367 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
5368 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
5369 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
5370 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
5371 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
5372 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
5373 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
5374 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
5375 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
5376 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
5377 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
5378 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
5379 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
5380 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
5381 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
5382 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
5383 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
5384 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
5385 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
5386 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
5387 Hoffmann, Zbigniew Jędrzejewski-Szmek
5389 — Berlin, 2015-02-16
5393 * When querying unit file enablement status (for example via
5394 "systemctl is-enabled"), a new state "indirect" is now known
5395 which indicates that a unit might not be enabled itself, but
5396 another unit listed in its Also= setting might be.
5398 * Similar to the various existing ConditionXYZ= settings for
5399 units, there are now matching AssertXYZ= settings. While
5400 failing conditions cause a unit to be skipped, but its job
5401 to succeed, failing assertions declared like this will cause
5402 a unit start operation and its job to fail.
5404 * hostnamed now knows a new chassis type "embedded".
5406 * systemctl gained a new "edit" command. When used on a unit
5407 file, this allows extending unit files with .d/ drop-in
5408 configuration snippets or editing the full file (after
5409 copying it from /usr/lib to /etc). This will invoke the
5410 user's editor (as configured with $EDITOR), and reload the
5411 modified configuration after editing.
5413 * "systemctl status" now shows the suggested enablement state
5414 for a unit, as declared in the (usually vendor-supplied)
5415 system preset files.
5417 * nss-myhostname will now resolve the single-label host name
5418 "gateway" to the locally configured default IP routing
5419 gateways, ordered by their metrics. This assigns a stable
5420 name to the used gateways, regardless which ones are
5421 currently configured. Note that the name will only be
5422 resolved after all other name sources (if nss-myhostname is
5423 configured properly) and should hence not negatively impact
5424 systems that use the single-label host name "gateway" in
5427 * systemd-inhibit now allows filtering by mode when listing
5430 * Scope and service units gained a new "Delegate" boolean
5431 property, which, when set, allows processes running inside the
5432 unit to further partition resources. This is primarily
5433 useful for systemd user instances as well as container
5436 * journald will now pick up audit messages directly from
5437 the kernel, and log them like any other log message. The
5438 audit fields are split up and fully indexed. This means that
5439 journalctl in many ways is now a (nicer!) alternative to
5440 ausearch, the traditional audit client. Note that this
5441 implements only a minimal audit client. If you want the
5442 special audit modes like reboot-on-log-overflow, please use
5443 the traditional auditd instead, which can be used in
5444 parallel to journald.
5446 * The ConditionSecurity= unit file option now understands the
5447 special string "audit" to check whether auditing is
5450 * journalctl gained two new commands --vacuum-size= and
5451 --vacuum-time= to delete old journal files until the
5452 remaining ones take up no more than the specified size on disk,
5453 or are not older than the specified time.
5455 * A new, native PPPoE library has been added to sd-network,
5456 systemd's library of light-weight networking protocols. This
5457 library will be used in a future version of networkd to
5458 enable PPPoE communication without an external pppd daemon.
5460 * The busctl tool now understands a new "capture" verb that
5461 works similar to "monitor", but writes a packet capture
5462 trace to STDOUT that can be redirected to a file which is
5463 compatible with libcap's capture file format. This can then
5464 be loaded in Wireshark and similar tools to inspect bus
5467 * The busctl tool now understands a new "tree" verb that shows
5468 the object trees of a specific service on the bus, or of all
5471 * The busctl tool now understands a new "introspect" verb that
5472 shows all interfaces and members of objects on the bus,
5473 including their signature and values. This is particularly
5474 useful to get more information about bus objects shown by
5475 the new "busctl tree" command.
5477 * The busctl tool now understands new verbs "call",
5478 "set-property" and "get-property" for invoking bus method
5479 calls, setting and getting bus object properties in a
5482 * busctl gained a new --augment-creds= argument that controls
5483 whether the tool shall augment credential information it
5484 gets from the bus with data from /proc, in a possibly
5487 * nspawn's --link-journal= switch gained two new values
5488 "try-guest" and "try-host" that work like "guest" and
5489 "host", but do not fail if the host has no persistent
5490 journalling enabled. -j is now equivalent to
5491 --link-journal=try-guest.
5493 * macvlan network devices created by nspawn will now have
5494 stable MAC addresses.
5496 * A new SmackProcessLabel= unit setting has been added, which
5497 controls the SMACK security label processes forked off by
5498 the respective unit shall use.
5500 * If compiled with --enable-xkbcommon, systemd-localed will
5501 verify x11 keymap settings by compiling the given keymap. It
5502 will spew out warnings if the compilation fails. This
5503 requires libxkbcommon to be installed.
5505 * When a coredump is collected, a larger number of metadata
5506 fields is now collected and included in the journal records
5507 created for it. More specifically, control group membership,
5508 environment variables, memory maps, working directory,
5509 chroot directory, /proc/$PID/status, and a list of open file
5510 descriptors is now stored in the log entry.
5512 * The udev hwdb now contains DPI information for mice. For
5515 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
5517 * All systemd programs that read standalone configuration
5518 files in /etc now also support a corresponding series of
5519 .conf.d configuration directories in /etc/, /run/,
5520 /usr/local/lib/, /usr/lib/, and (if configured with
5521 --enable-split-usr) /lib/. In particular, the following
5522 configuration files now have corresponding configuration
5523 directories: system.conf user.conf, logind.conf,
5524 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
5525 resolved.conf, timesyncd.conf, journal-remote.conf, and
5526 journal-upload.conf. Note that distributions should use the
5527 configuration directories in /usr/lib/; the directories in
5528 /etc/ are reserved for the system administrator.
5530 * systemd-rfkill will no longer take the rfkill device name
5531 into account when storing rfkill state on disk, as the name
5532 might be dynamically assigned and not stable. Instead, the
5533 ID_PATH udev variable combined with the rfkill type (wlan,
5534 bluetooth, ...) is used.
5536 * A new service systemd-machine-id-commit.service has been
5537 added. When used on systems where /etc is read-only during
5538 boot, and /etc/machine-id is not initialized (but an empty
5539 file), this service will copy the temporary machine ID
5540 created as replacement into /etc after the system is fully
5541 booted up. This is useful for systems that are freshly
5542 installed with a non-initialized machine ID, but should get
5543 a fixed machine ID for subsequent boots.
5545 * networkd's .netdev files now provide a large set of
5546 configuration parameters for VXLAN devices. Similarly, the
5547 bridge port cost parameter is now configurable in .network
5548 files. There's also new support for configuring IP source
5549 routing. networkd .link files gained support for a new
5550 OriginalName= match that is useful to match against the
5551 original interface name the kernel assigned. .network files
5552 may include MTU= and MACAddress= fields for altering the MTU
5553 and MAC address while being connected to a specific network
5556 * The LUKS logic gained supported for configuring
5557 UUID-specific key files. There's also new support for naming
5558 LUKS device from the kernel command line, using the new
5559 luks.name= argument.
5561 * Timer units may now be transiently created via the bus API
5562 (this was previously already available for scope and service
5563 units). In addition it is now possible to create multiple
5564 transient units at the same time with a single bus call. The
5565 "systemd-run" tool has been updated to make use of this for
5566 running commands on a specified time, in at(1)-style.
5568 * tmpfiles gained support for "t" lines, for assigning
5569 extended attributes to files. Among other uses this may be
5570 used to assign SMACK labels to files.
5572 Contributions from: Alin Rauta, Alison Chaiken, Andrej
5573 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
5574 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
5575 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
5576 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
5577 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
5578 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
5579 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
5580 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
5581 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
5582 Michael Biebl, Michael Chapman, Michael Marineau, Michal
5583 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
5584 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
5585 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
5586 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
5587 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
5588 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
5589 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
5591 — Berlin, 2014-12-10
5595 * journalctl gained the new options -t/--identifier= to match
5596 on the syslog identifier (aka "tag"), as well as --utc to
5597 show log timestamps in the UTC timezone. journalctl now also
5598 accepts -n/--lines=all to disable line capping in a pager.
5600 * journalctl gained a new switch, --flush, that synchronously
5601 flushes logs from /run/log/journal to /var/log/journal if
5602 persistent storage is enabled. systemd-journal-flush.service
5603 now waits until the operation is complete.
5605 * Services can notify the manager before they start a reload
5606 (by sending RELOADING=1) or shutdown (by sending
5607 STOPPING=1). This allows the manager to track and show the
5608 internal state of daemons and closes a race condition when
5609 the process is still running but has closed its D-Bus
5612 * Services with Type=oneshot do not have to have any ExecStart
5615 * User units are now loaded also from
5616 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
5617 /run/systemd/user directory that was already previously
5618 supported, but is under the control of the user.
5620 * Job timeouts (i.e. timeouts on the time a job that is
5621 queued stays in the run queue) can now optionally result in
5622 immediate reboot or power-off actions (JobTimeoutAction= and
5623 JobTimeoutRebootArgument=). This is useful on ".target"
5624 units, to limit the maximum time a target remains
5625 undispatched in the run queue, and to trigger an emergency
5626 operation in such a case. This is now used by default to
5627 turn off the system if boot-up (as defined by everything in
5628 basic.target) hangs and does not complete for at least
5629 15min. Also, if power-off or reboot hang for at least 30min
5630 an immediate power-off/reboot operation is triggered. This
5631 functionality is particularly useful to increase reliability
5632 on embedded devices, but also on laptops which might
5633 accidentally get powered on when carried in a backpack and
5634 whose boot stays stuck in a hard disk encryption passphrase
5637 * systemd-logind can be configured to also handle lid switch
5638 events even when the machine is docked or multiple displays
5639 are attached (HandleLidSwitchDocked= option).
5641 * A helper binary and a service have been added which can be
5642 used to resume from hibernation in the initramfs. A
5643 generator will parse the resume= option on the kernel
5644 command line to trigger resume.
5646 * A user console daemon systemd-consoled has been
5647 added. Currently, it is a preview, and will so far open a
5648 single terminal on each session of the user marked as
5649 Desktop=systemd-console.
5651 * Route metrics can be specified for DHCP routes added by
5654 * The SELinux context of socket-activated services can be set
5655 from the information provided by the networking stack
5656 (SELinuxContextFromNet= option).
5658 * Userspace firmware loading support has been removed and
5659 the minimum supported kernel version is thus bumped to 3.7.
5661 * Timeout for udev workers has been increased from 1 to 3
5662 minutes, but a warning will be printed after 1 minute to
5663 help diagnose kernel modules that take a long time to load.
5665 * Udev rules can now remove tags on devices with TAG-="foobar".
5667 * systemd's readahead implementation has been removed. In many
5668 circumstances it didn't give expected benefits even for
5669 rotational disk drives and was becoming less relevant in the
5670 age of SSDs. As none of the developers has been using
5671 rotating media anymore, and nobody stepped up to actively
5672 maintain this component of systemd it has now been removed.
5674 * Swap units can use Options= to specify discard options.
5675 Discard options specified for swaps in /etc/fstab are now
5678 * Docker containers are now detected as a separate type of
5681 * The Password Agent protocol gained support for queries where
5682 the user input is shown, useful e.g. for user names.
5683 systemd-ask-password gained a new --echo option to turn that
5686 * The default sysctl.d/ snippets will now set:
5688 net.core.default_qdisc = fq_codel
5690 This selects Fair Queuing Controlled Delay as the default
5691 queuing discipline for network interfaces. fq_codel helps
5692 fight the network bufferbloat problem. It is believed to be
5693 a good default with no tuning required for most workloads.
5694 Downstream distributions may override this choice. On 10Gbit
5695 servers that do not do forwarding, "fq" may perform better.
5696 Systems without a good clocksource should use "pfifo_fast".
5698 * If kdbus is enabled during build a new option BusPolicy= is
5699 available for service units, that allows locking all service
5700 processes into a stricter bus policy, in order to limit
5701 access to various bus services, or even hide most of them
5702 from the service's view entirely.
5704 * networkctl will now show the .network and .link file
5705 networkd has applied to a specific interface.
5707 * sd-login gained a new API call sd_session_get_desktop() to
5708 query which desktop environment has been selected for a
5711 * UNIX utmp support is now compile-time optional to support
5712 legacy-free systems.
5714 * systemctl gained two new commands "add-wants" and
5715 "add-requires" for pulling in units from specific targets
5718 * If the word "rescue" is specified on the kernel command line
5719 the system will now boot into rescue mode (aka
5720 rescue.target), which was previously available only by
5721 specifying "1" or "systemd.unit=rescue.target" on the kernel
5722 command line. This new kernel command line option nicely
5723 mirrors the already existing "emergency" kernel command line
5726 * New kernel command line options mount.usr=, mount.usrflags=,
5727 mount.usrfstype= have been added that match root=, rootflags=,
5728 rootfstype= but allow mounting a specific file system to
5731 * The $NOTIFY_SOCKET is now also passed to control processes of
5732 services, not only the main process.
5734 * This version reenables support for fsck's -l switch. This
5735 means at least version v2.25 of util-linux is required for
5736 operation, otherwise dead-locks on device nodes may
5737 occur. Again: you need to update util-linux to at least
5738 v2.25 when updating systemd to v217.
5740 * The "multi-seat-x" tool has been removed from systemd, as
5741 its functionality has been integrated into X servers 1.16,
5742 and the tool is hence redundant. It is recommended to update
5743 display managers invoking this tool to simply invoke X
5744 directly from now on, again.
5746 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
5747 message flag has been added for all of systemd's polkit
5748 authenticated method calls has been added. In particular this
5749 now allows optional interactive authorization via polkit for
5750 many of PID1's privileged operations such as unit file
5751 enabling and disabling.
5753 * "udevadm hwdb --update" learnt a new switch "--usr" for
5754 placing the rebuilt hardware database in /usr instead of
5755 /etc. When used only hardware database entries stored in
5756 /usr will be used, and any user database entries in /etc are
5757 ignored. This functionality is useful for vendors to ship a
5758 pre-built database on systems where local configuration is
5759 unnecessary or unlikely.
5761 * Calendar time specifications in .timer units now also
5762 understand the strings "semi-annually", "quarterly" and
5763 "minutely" as shortcuts (in addition to the preexisting
5764 "anually", "hourly", ...).
5766 * systemd-tmpfiles will now correctly create files in /dev
5767 at boot which are marked for creation only at boot. It is
5768 recommended to always create static device nodes with 'c!'
5769 and 'b!', so that they are created only at boot and not
5770 overwritten at runtime.
5772 * When the watchdog logic is used for a service (WatchdogSec=)
5773 and the watchdog timeout is hit the service will now be
5774 terminated with SIGABRT (instead of just SIGTERM), in order
5775 to make sure a proper coredump and backtrace is
5776 generated. This ensures that hanging services will result in
5777 similar coredump/backtrace behaviour as services that hit a
5780 Contributions from: Andreas Henriksson, Andrei Borzenkov,
5781 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
5782 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
5783 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
5784 Herrmann, David Sommerseth, David Strauss, Emil Renner
5785 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
5786 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
5787 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
5788 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
5789 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
5790 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
5791 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
5792 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
5793 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
5794 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
5795 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
5796 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
5797 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
5798 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
5799 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
5800 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
5803 — Berlin, 2014-10-28
5807 * timedated no longer reads NTP implementation unit names from
5808 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
5809 implementations should add a
5811 Conflicts=systemd-timesyncd.service
5813 to their unit files to take over and replace systemd's NTP
5814 default functionality.
5816 * systemd-sysusers gained a new line type "r" for configuring
5817 which UID/GID ranges to allocate system users/groups
5818 from. Lines of type "u" may now add an additional column
5819 that specifies the home directory for the system user to be
5820 created. Also, systemd-sysusers may now optionally read user
5821 information from STDIN instead of a file. This is useful for
5822 invoking it from RPM preinst scriptlets that need to create
5823 users before the first RPM file is installed since these
5824 files might need to be owned by them. A new
5825 %sysusers_create_inline RPM macro has been introduced to do
5826 just that. systemd-sysusers now updates the shadow files as
5827 well as the user/group databases, which should enhance
5828 compatibility with certain tools like grpck.
5830 * A number of bus APIs of PID 1 now optionally consult polkit to
5831 permit access for otherwise unprivileged clients under certain
5832 conditions. Note that this currently doesn't support
5833 interactive authentication yet, but this is expected to be
5834 added eventually, too.
5836 * /etc/machine-info now has new fields for configuring the
5837 deployment environment of the machine, as well as the
5838 location of the machine. hostnamectl has been updated with
5839 new command to update these fields.
5841 * systemd-timesyncd has been updated to automatically acquire
5842 NTP server information from systemd-networkd, which might
5843 have been discovered via DHCP.
5845 * systemd-resolved now includes a caching DNS stub resolver
5846 and a complete LLMNR name resolution implementation. A new
5847 NSS module "nss-resolve" has been added which can be used
5848 instead of glibc's own "nss-dns" to resolve hostnames via
5849 systemd-resolved. Hostnames, addresses and arbitrary RRs may
5850 be resolved via systemd-resolved D-Bus APIs. In contrast to
5851 the glibc internal resolver systemd-resolved is aware of
5852 multi-homed system, and keeps DNS server and caches separate
5853 and per-interface. Queries are sent simultaneously on all
5854 interfaces that have DNS servers configured, in order to
5855 properly handle VPNs and local LANs which might resolve
5856 separate sets of domain names. systemd-resolved may acquire
5857 DNS server information from systemd-networkd automatically,
5858 which in turn might have discovered them via DHCP. A tool
5859 "systemd-resolve-host" has been added that may be used to
5860 query the DNS logic in resolved. systemd-resolved implements
5861 IDNA and automatically uses IDNA or UTF-8 encoding depending
5862 on whether classic DNS or LLMNR is used as transport. In the
5863 next releases we intend to add a DNSSEC and mDNS/DNS-SD
5864 implementation to systemd-resolved.
5866 * A new NSS module nss-mymachines has been added, that
5867 automatically resolves the names of all local registered
5868 containers to their respective IP addresses.
5870 * A new client tool "networkctl" for systemd-networkd has been
5871 added. It currently is entirely passive and will query
5872 networking configuration from udev, rtnetlink and networkd,
5873 and present it to the user in a very friendly
5874 way. Eventually, we hope to extend it to become a full
5875 control utility for networkd.
5877 * .socket units gained a new DeferAcceptSec= setting that
5878 controls the kernels' TCP_DEFER_ACCEPT sockopt for
5879 TCP. Similarly, support for controlling TCP keep-alive
5880 settings has been added (KeepAliveTimeSec=,
5881 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
5882 turning off Nagle's algorithm on TCP has been added
5885 * logind learned a new session type "web", for use in projects
5886 like Cockpit which register web clients as PAM sessions.
5888 * timer units with at least one OnCalendar= setting will now
5889 be started only after time-sync.target has been
5890 reached. This way they will not elapse before the system
5891 clock has been corrected by a local NTP client or
5892 similar. This is particular useful on RTC-less embedded
5893 machines, that come up with an invalid system clock.
5895 * systemd-nspawn's --network-veth= switch should now result in
5896 stable MAC addresses for both the outer and the inner side
5899 * systemd-nspawn gained a new --volatile= switch for running
5900 container instances with /etc or /var unpopulated.
5902 * The kdbus client code has been updated to use the new Linux
5903 3.17 memfd subsystem instead of the old kdbus-specific one.
5905 * systemd-networkd's DHCP client and server now support
5906 FORCERENEW. There are also new configuration options to
5907 configure the vendor client identifier and broadcast mode
5910 * systemd will no longer inform the kernel about the current
5911 timezone, as this is necessarily incorrect and racy as the
5912 kernel has no understanding of DST and similar
5913 concepts. This hence means FAT timestamps will be always
5914 considered UTC, similar to what Android is already
5915 doing. Also, when the RTC is configured to the local time
5916 (rather than UTC) systemd will never synchronize back to it,
5917 as this might confuse Windows at a later boot.
5919 * systemd-analyze gained a new command "verify" for offline
5920 validation of unit files.
5922 * systemd-networkd gained support for a couple of additional
5923 settings for bonding networking setups. Also, the metric for
5924 statically configured routes may now be configured. For
5925 network interfaces where this is appropriate the peer IP
5926 address may now be configured.
5928 * systemd-networkd's DHCP client will no longer request
5929 broadcasting by default, as this tripped up some networks.
5930 For hardware where broadcast is required the feature should
5931 be switched back on using RequestBroadcast=yes.
5933 * systemd-networkd will now set up IPv4LL addresses (when
5934 enabled) even if DHCP is configured successfully.
5936 * udev will now default to respect network device names given
5937 by the kernel when the kernel indicates that these are
5938 predictable. This behavior can be tweaked by changing
5939 NamePolicy= in the relevant .link file.
5941 * A new library systemd-terminal has been added that
5942 implements full TTY stream parsing and rendering. This
5943 library is supposed to be used later on for implementing a
5944 full userspace VT subsystem, replacing the current kernel
5947 * A new tool systemd-journal-upload has been added to push
5948 journal data to a remote system running
5949 systemd-journal-remote.
5951 * journald will no longer forward all local data to another
5952 running syslog daemon. This change has been made because
5953 rsyslog (which appears to be the most commonly used syslog
5954 implementation these days) no longer makes use of this, and
5955 instead pulls the data out of the journal on its own. Since
5956 forwarding the messages to a non-existent syslog server is
5957 more expensive than we assumed we have now turned this
5958 off. If you run a syslog server that is not a recent rsyslog
5959 version, you have to turn this option on again
5960 (ForwardToSyslog= in journald.conf).
5962 * journald now optionally supports the LZ4 compressor for
5963 larger journal fields. This compressor should perform much
5964 better than XZ which was the previous default.
5966 * machinectl now shows the IP addresses of local containers,
5967 if it knows them, plus the interface name of the container.
5969 * A new tool "systemd-escape" has been added that makes it
5970 easy to escape strings to build unit names and similar.
5972 * sd_notify() messages may now include a new ERRNO= field
5973 which is parsed and collected by systemd and shown among the
5974 "systemctl status" output for a service.
5976 * A new component "systemd-firstboot" has been added that
5977 queries the most basic systemd information (timezone,
5978 hostname, root password) interactively on first
5979 boot. Alternatively it may also be used to provision these
5980 things offline on OS images installed into directories.
5982 * The default sysctl.d/ snippets will now set
5984 net.ipv4.conf.default.promote_secondaries=1
5986 This has the benefit of no flushing secondary IP addresses
5987 when primary addresses are removed.
5989 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
5990 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
5991 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
5992 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
5993 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
5994 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
5995 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
5996 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
5997 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
5998 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
5999 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
6000 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
6001 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
6002 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
6003 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
6005 — Berlin, 2014-08-19
6009 * A new tool systemd-sysusers has been added. This tool
6010 creates system users and groups in /etc/passwd and
6011 /etc/group, based on static declarative system user/group
6012 definitions in /usr/lib/sysusers.d/. This is useful to
6013 enable factory resets and volatile systems that boot up with
6014 an empty /etc directory, and thus need system users and
6015 groups created during early boot. systemd now also ships
6016 with two default sysusers.d/ files for the most basic
6017 users and groups systemd and the core operating system
6020 * A new tmpfiles snippet has been added that rebuilds the
6021 essential files in /etc on boot, should they be missing.
6023 * A directive for ensuring automatic clean-up of
6024 /var/cache/man/ has been removed from the default
6025 configuration. This line should now be shipped by the man
6026 implementation. The necessary change has been made to the
6027 man-db implementation. Note that you need to update your man
6028 implementation to one that ships this line, otherwise no
6029 automatic clean-up of /var/cache/man will take place.
6031 * A new condition ConditionNeedsUpdate= has been added that
6032 may conditionalize services to only run when /etc or /var
6033 are "older" than the vendor operating system resources in
6034 /usr. This is useful for reconstructing or updating /etc
6035 after an offline update of /usr or a factory reset, on the
6036 next reboot. Services that want to run once after such an
6037 update or reset should use this condition and order
6038 themselves before the new systemd-update-done.service, which
6039 will mark the two directories as fully updated. A number of
6040 service files have been added making use of this, to rebuild
6041 the udev hardware database, the journald message catalog and
6042 dynamic loader cache (ldconfig). The systemd-sysusers tool
6043 described above also makes use of this now. With this in
6044 place it is now possible to start up a minimal operating
6045 system with /etc empty cleanly. For more information on the
6046 concepts involved see this recent blog story:
6048 http://0pointer.de/blog/projects/stateless.html
6050 * A new system group "input" has been introduced, and all
6051 input device nodes get this group assigned. This is useful
6052 for system-level software to get access to input devices. It
6053 complements what is already done for "audio" and "video".
6055 * systemd-networkd learnt minimal DHCPv4 server support in
6056 addition to the existing DHCPv4 client support. It also
6057 learnt DHCPv6 client and IPv6 Router Solicitation client
6058 support. The DHCPv4 client gained support for static routes
6059 passed in from the server. Note that the [DHCPv4] section
6060 known in older systemd-networkd versions has been renamed to
6061 [DHCP] and is now also used by the DHCPv6 client. Existing
6062 .network files using settings of this section should be
6063 updated, though compatibility is maintained. Optionally, the
6064 client hostname may now be sent to the DHCP server.
6066 * networkd gained support for vxlan virtual networks as well
6067 as tun/tap and dummy devices.
6069 * networkd gained support for automatic allocation of address
6070 ranges for interfaces from a system-wide pool of
6071 addresses. This is useful for dynamically managing a large
6072 number of interfaces with a single network configuration
6073 file. In particular this is useful to easily assign
6074 appropriate IP addresses to the veth links of a large number
6075 of nspawn instances.
6077 * RPM macros for processing sysusers, sysctl and binfmt
6078 drop-in snippets at package installation time have been
6081 * The /etc/os-release file should now be placed in
6082 /usr/lib/os-release. The old location is automatically
6083 created as symlink. /usr/lib is the more appropriate
6084 location of this file, since it shall actually describe the
6085 vendor operating system shipped in /usr, and not the
6086 configuration stored in /etc.
6088 * .mount units gained a new boolean SloppyOptions= setting
6089 that maps to mount(8)'s -s option which enables permissive
6090 parsing of unknown mount options.
6092 * tmpfiles learnt a new "L+" directive which creates a symlink
6093 but (unlike "L") deletes a pre-existing file first, should
6094 it already exist and not already be the correct
6095 symlink. Similarly, "b+", "c+" and "p+" directives have been
6096 added as well, which create block and character devices, as
6097 well as fifos in the filesystem, possibly removing any
6098 pre-existing files of different types.
6100 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
6101 'argument' field (which so far specified the source to
6102 symlink/copy the files from) is now optional. If omitted the
6103 same file os copied from /usr/share/factory/ suffixed by the
6104 full destination path. This is useful for populating /etc
6105 with essential files, by copying them from vendor defaults
6106 shipped in /usr/share/factory/etc.
6108 * A new command "systemctl preset-all" has been added that
6109 applies the service preset settings to all installed unit
6110 files. A new switch --preset-mode= has been added that
6111 controls whether only enable or only disable operations
6114 * A new command "systemctl is-system-running" has been added
6115 that allows checking the overall state of the system, for
6116 example whether it is fully up and running.
6118 * When the system boots up with an empty /etc, the equivalent
6119 to "systemctl preset-all" is executed during early boot, to
6120 make sure all default services are enabled after a factory
6123 * systemd now contains a minimal preset file that enables the
6124 most basic services systemd ships by default.
6126 * Unit files' [Install] section gained a new DefaultInstance=
6127 field for defining the default instance to create if a
6128 template unit is enabled with no instance specified.
6130 * A new passive target cryptsetup-pre.target has been added
6131 that may be used by services that need to make they run and
6132 finish before the first LUKS cryptographic device is set up.
6134 * The /dev/loop-control and /dev/btrfs-control device nodes
6135 are now owned by the "disk" group by default, opening up
6136 access to this group.
6138 * systemd-coredump will now automatically generate a
6139 stack trace of all core dumps taking place on the system,
6140 based on elfutils' libdw library. This stack trace is logged
6143 * systemd-coredump may now optionally store coredumps directly
6144 on disk (in /var/lib/systemd/coredump, possibly compressed),
6145 instead of storing them unconditionally in the journal. This
6146 mode is the new default. A new configuration file
6147 /etc/systemd/coredump.conf has been added to configure this
6148 and other parameters of systemd-coredump.
6150 * coredumpctl gained a new "info" verb to show details about a
6151 specific coredump. A new switch "-1" has also been added
6152 that makes sure to only show information about the most
6153 recent entry instead of all entries. Also, as the tool is
6154 generally useful now the "systemd-" prefix of the binary
6155 name has been removed. Distributions that want to maintain
6156 compatibility with the old name should add a symlink from
6157 the old name to the new name.
6159 * journald's SplitMode= now defaults to "uid". This makes sure
6160 that unprivileged users can access their own coredumps with
6161 coredumpctl without restrictions.
6163 * New kernel command line options "systemd.wants=" (for
6164 pulling an additional unit during boot), "systemd.mask="
6165 (for masking a specific unit for the boot), and
6166 "systemd.debug-shell" (for enabling the debug shell on tty9)
6167 have been added. This is implemented in the new generator
6168 "systemd-debug-generator".
6170 * systemd-nspawn will now by default filter a couple of
6171 syscalls for containers, among them those required for
6172 kernel module loading, direct x86 IO port access, swap
6173 management, and kexec. Most importantly though
6174 open_by_handle_at() is now prohibited for containers,
6175 closing a hole similar to a recently discussed vulnerability
6176 in docker regarding access to files on file hierarchies the
6177 container should normally not have access to. Note that, for
6178 nspawn, we generally make no security claims anyway (and
6179 this is explicitly documented in the man page), so this is
6180 just a fix for one of the most obvious problems.
6182 * A new man page file-hierarchy(7) has been added that
6183 contains a minimized, modernized version of the file system
6184 layout systemd expects, similar in style to the FHS
6185 specification or hier(5). A new tool systemd-path(1) has
6186 been added to query many of these paths for the local
6189 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
6190 longer done. Since the directory now has a per-user size
6191 limit, and is cleaned on logout this appears unnecessary,
6192 in particular since this now brings the lifecycle of this
6193 directory closer in line with how IPC objects are handled.
6195 * systemd.pc now exports a number of additional directories,
6196 including $libdir (which is useful to identify the library
6197 path for the primary architecture of the system), and a
6198 couple of drop-in directories.
6200 * udev's predictable network interface names now use the dev_port
6201 sysfs attribute, introduced in linux 3.15 instead of dev_id to
6202 distinguish between ports of the same PCI function. dev_id should
6203 only be used for ports using the same HW address, hence the need
6206 * machined has been updated to export the OS version of a
6207 container (read from /etc/os-release and
6208 /usr/lib/os-release) on the bus. This is now shown in
6209 "machinectl status" for a machine.
6211 * A new service setting RestartForceExitStatus= has been
6212 added. If configured to a set of exit signals or process
6213 return values, the service will be restarted when the main
6214 daemon process exits with any of them, regardless of the
6217 * systemctl's -H switch for connecting to remote systemd
6218 machines has been extended so that it may be used to
6219 directly connect to a specific container on the
6220 host. "systemctl -H root@foobar:waldi" will now connect as
6221 user "root" to host "foobar", and then proceed directly to
6222 the container named "waldi". Note that currently you have to
6223 authenticate as user "root" for this to work, as entering
6224 containers is a privileged operation.
6226 Contributions from: Andreas Henriksson, Benjamin Steinwender,
6227 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
6228 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
6229 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
6230 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
6231 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
6232 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
6233 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
6234 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
6235 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
6236 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
6237 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
6239 — Berlin, 2014-07-03
6243 * As an experimental feature, udev now tries to lock the
6244 disk device node (flock(LOCK_SH|LOCK_NB)) while it
6245 executes events for the disk or any of its partitions.
6246 Applications like partitioning programs can lock the
6247 disk device node (flock(LOCK_EX)) and claim temporary
6248 device ownership that way; udev will entirely skip all event
6249 handling for this disk and its partitions. If the disk
6250 was opened for writing, the close will trigger a partition
6251 table rescan in udev's "watch" facility, and if needed
6252 synthesize "change" events for the disk and all its partitions.
6253 This is now unconditionally enabled, and if it turns out to
6254 cause major problems, we might turn it on only for specific
6255 devices, or might need to disable it entirely. Device Mapper
6256 devices are excluded from this logic.
6258 * We temporarily dropped the "-l" switch for fsck invocations,
6259 since they collide with the flock() logic above. util-linux
6260 upstream has been changed already to avoid this conflict,
6261 and we will re-add "-l" as soon as util-linux with this
6262 change has been released.
6264 * The dependency on libattr has been removed. Since a long
6265 time, the extended attribute calls have moved to glibc, and
6266 libattr is thus unnecessary.
6268 * Virtualization detection works without privileges now. This
6269 means the systemd-detect-virt binary no longer requires
6270 CAP_SYS_PTRACE file capabilities, and our daemons can run
6271 with fewer privileges.
6273 * systemd-networkd now runs under its own "systemd-network"
6274 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
6275 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
6276 loses the ability to write to files owned by root this way.
6278 * Similarly, systemd-resolved now runs under its own
6279 "systemd-resolve" user with no capabilities remaining.
6281 * Similarly, systemd-bus-proxyd now runs under its own
6282 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
6284 * systemd-networkd gained support for setting up "veth"
6285 virtual Ethernet devices for container connectivity, as well
6286 as GRE and VTI tunnels.
6288 * systemd-networkd will no longer automatically attempt to
6289 manually load kernel modules necessary for certain tunnel
6290 transports. Instead, it is assumed the kernel loads them
6291 automatically when required. This only works correctly on
6292 very new kernels. On older kernels, please consider adding
6293 the kernel modules to /etc/modules-load.d/ as a work-around.
6295 * The resolv.conf file systemd-resolved generates has been
6296 moved to /run/systemd/resolve/. If you have a symlink from
6297 /etc/resolv.conf, it might be necessary to correct it.
6299 * Two new service settings, ProtectHome= and ProtectSystem=,
6300 have been added. When enabled, they will make the user data
6301 (such as /home) inaccessible or read-only and the system
6302 (such as /usr) read-only, for specific services. This allows
6303 very light-weight per-service sandboxing to avoid
6304 modifications of user data or system files from
6305 services. These two new switches have been enabled for all
6306 of systemd's long-running services, where appropriate.
6308 * Socket units gained new SocketUser= and SocketGroup=
6309 settings to set the owner user and group of AF_UNIX sockets
6310 and FIFOs in the file system.
6312 * Socket units gained a new RemoveOnStop= setting. If enabled,
6313 all FIFOS and sockets in the file system will be removed
6314 when the specific socket unit is stopped.
6316 * Socket units gained a new Symlinks= setting. It takes a list
6317 of symlinks to create to file system sockets or FIFOs
6318 created by the specific Unix sockets. This is useful to
6319 manage symlinks to socket nodes with the same lifecycle as
6322 * The /dev/log socket and /dev/initctl FIFO have been moved to
6323 /run, and have been replaced by symlinks. This allows
6324 connecting to these facilities even if PrivateDevices=yes is
6325 used for a service (which makes /dev/log itself unavailable,
6326 but /run is left). This also has the benefit of ensuring
6327 that /dev only contains device nodes, directories and
6328 symlinks, and nothing else.
6330 * sd-daemon gained two new calls sd_pid_notify() and
6331 sd_pid_notifyf(). They are similar to sd_notify() and
6332 sd_notifyf(), but allow overriding of the source PID of
6333 notification messages if permissions permit this. This is
6334 useful to send notify messages on behalf of a different
6335 process (for example, the parent process). The
6336 systemd-notify tool has been updated to make use of this
6337 when sending messages (so that notification messages now
6338 originate from the shell script invoking systemd-notify and
6339 not the systemd-notify process itself. This should minimize
6340 a race where systemd fails to associate notification
6341 messages to services when the originating process already
6344 * A new "on-abnormal" setting for Restart= has been added. If
6345 set, it will result in automatic restarts on all "abnormal"
6346 reasons for a process to exit, which includes unclean
6347 signals, core dumps, timeouts and watchdog timeouts, but
6348 does not include clean and unclean exit codes or clean
6349 signals. Restart=on-abnormal is an alternative for
6350 Restart=on-failure for services that shall be able to
6351 terminate and avoid restarts on certain errors, by
6352 indicating so with an unclean exit code. Restart=on-failure
6353 or Restart=on-abnormal is now the recommended setting for
6354 all long-running services.
6356 * If the InaccessibleDirectories= service setting points to a
6357 mount point (or if there are any submounts contained within
6358 it), it is now attempted to completely unmount it, to make
6359 the file systems truly unavailable for the respective
6362 * The ReadOnlyDirectories= service setting and
6363 systemd-nspawn's --read-only parameter are now recursively
6364 applied to all submounts, too.
6366 * Mount units may now be created transiently via the bus APIs.
6368 * The support for SysV and LSB init scripts has been removed
6369 from the systemd daemon itself. Instead, it is now
6370 implemented as a generator that creates native systemd units
6371 from these scripts when needed. This enables us to remove a
6372 substantial amount of legacy code from PID 1, following the
6373 fact that many distributions only ship a very small number
6374 of LSB/SysV init scripts nowadays.
6376 * Privileged Xen (dom0) domains are not considered
6377 virtualization anymore by the virtualization detection
6378 logic. After all, they generally have unrestricted access to
6379 the hardware and usually are used to manage the unprivileged
6382 * systemd-tmpfiles gained a new "C" line type, for copying
6383 files or entire directories.
6385 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
6386 lines. So far, they have been non-globbing versions of the
6387 latter, and have thus been redundant. In future, it is
6388 recommended to only use "z". "m" has hence been removed
6389 from the documentation, even though it stays supported.
6391 * A tmpfiles snippet to recreate the most basic structure in
6392 /var has been added. This is enough to create the /var/run →
6393 /run symlink and create a couple of structural
6394 directories. This allows systems to boot up with an empty or
6395 volatile /var. Of course, while with this change, the core OS
6396 now is capable with dealing with a volatile /var, not all
6397 user services are ready for it. However, we hope that sooner
6398 or later, many service daemons will be changed upstream so
6399 that they are able to automatically create their necessary
6400 directories in /var at boot, should they be missing. This is
6401 the first step to allow state-less systems that only require
6402 the vendor image for /usr to boot.
6404 * systemd-nspawn has gained a new --tmpfs= switch to mount an
6405 empty tmpfs instance to a specific directory. This is
6406 particularly useful for making use of the automatic
6407 reconstruction of /var (see above), by passing --tmpfs=/var.
6409 * Access modes specified in tmpfiles snippets may now be
6410 prefixed with "~", which indicates that they shall be masked
6411 by whether the existing file or directory is currently
6412 writable, readable or executable at all. Also, if specified,
6413 the sgid/suid/sticky bits will be masked for all
6416 * A new passive target unit "network-pre.target" has been
6417 added which is useful for services that shall run before any
6418 network is configured, for example firewall scripts.
6420 * The "floppy" group that previously owned the /dev/fd*
6421 devices is no longer used. The "disk" group is now used
6422 instead. Distributions should probably deprecate usage of
6425 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
6426 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
6427 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
6428 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
6429 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
6430 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
6433 — Berlin, 2014-06-11
6437 * A new "systemd-timesyncd" daemon has been added for
6438 synchronizing the system clock across the network. It
6439 implements an SNTP client. In contrast to NTP
6440 implementations such as chrony or the NTP reference server,
6441 this only implements a client side, and does not bother with
6442 the full NTP complexity, focusing only on querying time from
6443 one remote server and synchronizing the local clock to
6444 it. Unless you intend to serve NTP to networked clients or
6445 want to connect to local hardware clocks, this simple NTP
6446 client should be more than appropriate for most
6447 installations. The daemon runs with minimal privileges, and
6448 has been hooked up with networkd to only operate when
6449 network connectivity is available. The daemon saves the
6450 current clock to disk every time a new NTP sync has been
6451 acquired, and uses this to possibly correct the system clock
6452 early at bootup, in order to accommodate for systems that
6453 lack an RTC such as the Raspberry Pi and embedded devices,
6454 and to make sure that time monotonically progresses on these
6455 systems, even if it is not always correct. To make use of
6456 this daemon, a new system user and group "systemd-timesync"
6457 needs to be created on installation of systemd.
6459 * The queue "seqnum" interface of libudev has been disabled, as
6460 it was generally incompatible with device namespacing as
6461 sequence numbers of devices go "missing" if the devices are
6462 part of a different namespace.
6464 * "systemctl list-timers" and "systemctl list-sockets" gained
6465 a --recursive switch for showing units of these types also
6466 for all local containers, similar in style to the already
6467 supported --recursive switch for "systemctl list-units".
6469 * A new RebootArgument= setting has been added for service
6470 units, which may be used to specify a kernel reboot argument
6471 to use when triggering reboots with StartLimitAction=.
6473 * A new FailureAction= setting has been added for service
6474 units which may be used to specify an operation to trigger
6475 when a service fails. This works similarly to
6476 StartLimitAction=, but unlike it, controls what is done
6477 immediately rather than only after several attempts to
6478 restart the service in question.
6480 * hostnamed got updated to also expose the kernel name,
6481 release, and version on the bus. This is useful for
6482 executing commands like hostnamectl with the -H switch.
6483 systemd-analyze makes use of this to properly display
6484 details when running non-locally.
6486 * The bootchart tool can now show cgroup information in the
6487 graphs it generates.
6489 * The CFS CPU quota cgroup attribute is now exposed for
6490 services. The new CPUQuota= switch has been added for this
6491 which takes a percentage value. Setting this will have the
6492 result that a service may never get more CPU time than the
6493 specified percentage, even if the machine is otherwise idle.
6495 * systemd-networkd learned IPIP and SIT tunnel support.
6497 * LSB init scripts exposing a dependency on $network will now
6498 get a dependency on network-online.target rather than simply
6499 network.target. This should bring LSB handling closer to
6500 what it was on SysV systems.
6502 * A new fsck.repair= kernel option has been added to control
6503 how fsck shall deal with unclean file systems at boot.
6505 * The (.ini) configuration file parser will now silently
6506 ignore sections whose name begins with "X-". This may be
6507 used to maintain application-specific extension sections in unit
6510 * machined gained a new API to query the IP addresses of
6511 registered containers. "machinectl status" has been updated
6512 to show these addresses in its output.
6514 * A new call sd_uid_get_display() has been added to the
6515 sd-login APIs for querying the "primary" session of a
6516 user. The "primary" session of the user is elected from the
6517 user's sessions and generally a graphical session is
6518 preferred over a text one.
6520 * A minimal systemd-resolved daemon has been added. It
6521 currently simply acts as a companion to systemd-networkd and
6522 manages resolv.conf based on per-interface DNS
6523 configuration, possibly supplied via DHCP. In the long run
6524 we hope to extend this into a local DNSSEC enabled DNS and
6527 * The systemd-networkd-wait-online tool is now enabled by
6528 default. It will delay network-online.target until a network
6529 connection has been configured. The tool primarily integrates
6530 with networkd, but will also make a best effort to make sense
6531 of network configuration performed in some other way.
6533 * Two new service options StartupCPUShares= and
6534 StartupBlockIOWeight= have been added that work similarly to
6535 CPUShares= and BlockIOWeight= however only apply during
6536 system startup. This is useful to prioritize certain services
6537 differently during bootup than during normal runtime.
6539 * hostnamed has been changed to prefer the statically
6540 configured hostname in /etc/hostname (unless set to
6541 'localhost' or empty) over any dynamic one supplied by
6542 dhcp. With this change, the rules for picking the hostname
6543 match more closely the rules of other configuration settings
6544 where the local administrator's configuration in /etc always
6545 overrides any other settings.
6547 Contributions from: Ali H. Caliskan, Alison Chaiken, Bas van
6548 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
6549 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
6550 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
6551 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
6552 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
6553 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
6554 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
6555 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
6556 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
6557 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
6558 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
6559 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
6560 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
6561 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
6562 Lindskog, WaLyong Cho, Will Woods, Zbigniew
6565 — Beijing, 2014-05-28
6569 * When restoring the screen brightness at boot, stay away from
6570 the darkest setting or from the lowest 5% of the available
6571 range, depending on which is the larger value of both. This
6572 should effectively protect the user from rebooting into a
6573 black screen, should the brightness have been set to minimum
6576 * sd-login gained a new sd_machine_get_class() call to
6577 determine the class ("vm" or "container") of a machine
6578 registered with machined.
6580 * sd-login gained new calls
6581 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
6582 to query the identity of the peer of a local AF_UNIX
6583 connection. They operate similarly to their sd_pid_get_xyz()
6586 * PID 1 will now maintain a system-wide system state engine
6587 with the states "starting", "running", "degraded",
6588 "maintenance", "stopping". These states are bound to system
6589 startup, normal runtime, runtime with at least one failed
6590 service, rescue/emergency mode and system shutdown. This
6591 state is shown in the "systemctl status" output when no unit
6592 name is passed. It is useful to determine system state, in
6593 particularly when doing so for many systems or containers at
6596 * A new command "list-machines" has been added to "systemctl"
6597 that lists all local OS containers and shows their system
6598 state (see above), if systemd runs inside of them.
6600 * systemctl gained a new "-r" switch to recursively enumerate
6601 units on all local containers, when used with the
6602 "list-unit" command (which is the default one that is
6603 executed when no parameters are specified).
6605 * The GPT automatic partition discovery logic will now honour
6606 two GPT partition flags: one may be set on a partition to
6607 cause it to be mounted read-only, and the other may be set
6608 on a partition to ignore it during automatic discovery.
6610 * Two new GPT type UUIDs have been added for automatic root
6611 partition discovery, for 32-bit and 64-bit ARM. This is not
6612 particularly useful for discovering the root directory on
6613 these architectures during bare-metal boots (since UEFI is
6614 not common there), but still very useful to allow booting of
6615 ARM disk images in nspawn with the -i option.
6617 * MAC addresses of interfaces created with nspawn's
6618 --network-interface= switch will now be generated from the
6619 machine name, and thus be stable between multiple invocations
6622 * logind will now automatically remove all IPC objects owned
6623 by a user if she or he fully logs out. This makes sure that
6624 users who are logged out cannot continue to consume IPC
6625 resources. This covers SysV memory, semaphores and message
6626 queues as well as POSIX shared memory and message
6627 queues. Traditionally, SysV and POSIX IPC had no lifecycle
6628 limits. With this functionality, that is corrected. This may
6629 be turned off by using the RemoveIPC= switch of logind.conf.
6631 * The systemd-machine-id-setup and tmpfiles tools gained a
6632 --root= switch to operate on a specific root directory,
6635 * journald can now forward logged messages to the TTYs of all
6636 logged in users ("wall"). This is the default for all
6637 emergency messages now.
6639 * A new tool systemd-journal-remote has been added to stream
6640 journal log messages across the network.
6642 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
6643 controller trees are mounted into it. Note that the
6644 directories mounted beneath it are not read-only. This is a
6645 security measure and is particularly useful because glibc
6646 actually includes a search logic to pick any tmpfs it can
6647 find to implement shm_open() if /dev/shm is not available
6648 (which it might very well be in namespaced setups).
6650 * machinectl gained a new "poweroff" command to cleanly power
6651 down a local OS container.
6653 * The PrivateDevices= unit file setting will now also drop the
6654 CAP_MKNOD capability from the capability bound set, and
6655 imply DevicePolicy=closed.
6657 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
6658 comprehensively on all long-running systemd services where
6659 this is appropriate.
6661 * systemd-udevd will now run in a disassociated mount
6662 namespace. To mount directories from udev rules, make sure to
6663 pull in mount units via SYSTEMD_WANTS properties.
6665 * The kdbus support gained support for uploading policy into
6666 the kernel. sd-bus gained support for creating "monitoring"
6667 connections that can eavesdrop into all bus communication
6668 for debugging purposes.
6670 * Timestamps may now be specified in seconds since the UNIX
6671 epoch Jan 1st, 1970 by specifying "@" followed by the value
6674 * Native tcpwrap support in systemd has been removed. tcpwrap
6675 is old code, not really maintained anymore and has serious
6676 shortcomings, and better options such as firewalls
6677 exist. For setups that require tcpwrap usage, please
6678 consider invoking your socket-activated service via tcpd,
6679 like on traditional inetd.
6681 * A new system.conf configuration option
6682 DefaultTimerAccuracySec= has been added that controls the
6683 default AccuracySec= setting of .timer units.
6685 * Timer units gained a new WakeSystem= switch. If enabled,
6686 timers configured this way will cause the system to resume
6687 from system suspend (if the system supports that, which most
6690 * Timer units gained a new Persistent= switch. If enabled,
6691 timers configured this way will save to disk when they have
6692 been last triggered. This information is then used on next
6693 reboot to possible execute overdue timer events, that
6694 could not take place because the system was powered off.
6695 This enables simple anacron-like behaviour for timer units.
6697 * systemctl's "list-timers" will now also list the time a
6698 timer unit was last triggered in addition to the next time
6699 it will be triggered.
6701 * systemd-networkd will now assign predictable IPv4LL
6702 addresses to its local interfaces.
6704 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
6705 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
6706 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
6707 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
6708 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
6709 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
6710 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
6711 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
6714 — Berlin, 2014-03-25
6718 * A new unit file setting RestrictAddressFamilies= has been
6719 added to restrict which socket address families unit
6720 processes gain access to. This takes address family names
6721 like "AF_INET" or "AF_UNIX", and is useful to minimize the
6722 attack surface of services via exotic protocol stacks. This
6723 is built on seccomp system call filters.
6725 * Two new unit file settings RuntimeDirectory= and
6726 RuntimeDirectoryMode= have been added that may be used to
6727 manage a per-daemon runtime directories below /run. This is
6728 an alternative for setting up directory permissions with
6729 tmpfiles snippets, and has the advantage that the runtime
6730 directory's lifetime is bound to the daemon runtime and that
6731 the daemon starts up with an empty directory each time. This
6732 is particularly useful when writing services that drop
6733 privileges using the User= or Group= setting.
6735 * The DeviceAllow= unit setting now supports globbing for
6736 matching against device group names.
6738 * The systemd configuration file system.conf gained new
6739 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
6740 DefaultMemoryAccounting= to globally turn on/off accounting
6741 for specific resources (cgroups) for all units. These
6742 settings may still be overridden individually in each unit
6745 * systemd-gpt-auto-generator is now able to discover /srv and
6746 root partitions in addition to /home and swap partitions. It
6747 also supports LUKS-encrypted partitions now. With this in
6748 place, automatic discovery of partitions to mount following
6749 the Discoverable Partitions Specification
6750 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
6751 is now a lot more complete. This allows booting without
6752 /etc/fstab and without root= on the kernel command line on
6753 systems prepared appropriately.
6755 * systemd-nspawn gained a new --image= switch which allows
6756 booting up disk images and Linux installations on any block
6757 device that follow the Discoverable Partitions Specification
6758 (see above). This means that installations made with
6759 appropriately updated installers may now be started and
6760 deployed using container managers, completely
6761 unmodified. (We hope that libvirt-lxc will add support for
6762 this feature soon, too.)
6764 * systemd-nspawn gained a new --network-macvlan= setting to
6765 set up a private macvlan interface for the
6766 container. Similarly, systemd-networkd gained a new
6767 Kind=macvlan setting in .netdev files.
6769 * systemd-networkd now supports configuring local addresses
6772 * A new tool systemd-network-wait-online has been added to
6773 synchronously wait for network connectivity using
6776 * The sd-bus.h bus API gained a new sd_bus_track object for
6777 tracking the lifecycle of bus peers. Note that sd-bus.h is
6778 still not a public API though (unless you specify
6779 --enable-kdbus on the configure command line, which however
6780 voids your warranty and you get no API stability guarantee).
6782 * The $XDG_RUNTIME_DIR runtime directories for each user are
6783 now individual tmpfs instances, which has the benefit of
6784 introducing separate pools for each user, with individual
6785 size limits, and thus making sure that unprivileged clients
6786 can no longer negatively impact the system or other users by
6787 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
6788 RuntimeDirectorySize= has been introduced that allows
6789 controlling the default size limit for all users. It
6790 defaults to 10% of the available physical memory. This is no
6791 replacement for quotas on tmpfs though (which the kernel
6792 still does not support), as /dev/shm and /tmp are still
6793 shared resources used by both the system and unprivileged
6796 * logind will now automatically turn off automatic suspending
6797 on laptop lid close when more than one display is
6798 connected. This was previously expected to be implemented
6799 individually in desktop environments (such as GNOME),
6800 however has been added to logind now, in order to fix a
6801 boot-time race where a desktop environment might not have
6802 been started yet and thus not been able to take an inhibitor
6803 lock at the time where logind already suspends the system
6804 due to a closed lid.
6806 * logind will now wait at least 30s after each system
6807 suspend/resume cycle, and 3min after system boot before
6808 suspending the system due to a closed laptop lid. This
6809 should give USB docking stations and similar enough time to
6810 be probed and configured after system resume and boot in
6811 order to then act as suspend blocker.
6813 * systemd-run gained a new --property= setting which allows
6814 initialization of resource control properties (and others)
6815 for the created scope or service unit. Example: "systemd-run
6816 --property=BlockIOWeight=10 updatedb" may be used to run
6817 updatedb at a low block IO scheduling weight.
6819 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
6820 now also work in --scope mode.
6822 * When systemd is compiled with kdbus support, basic support
6823 for enforced policies is now in place. (Note that enabling
6824 kdbus still voids your warranty and no API compatibility
6827 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
6828 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
6829 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
6830 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
6831 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
6832 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
6833 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
6834 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
6835 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
6836 Zbigniew Jędrzejewski-Szmek
6838 — Berlin, 2014-03-12
6842 * systemd will now relabel /dev after loading the SMACK policy
6843 according to SMACK rules.
6845 * A new unit file option AppArmorProfile= has been added to
6846 set the AppArmor profile for the processes of a unit.
6848 * A new condition check ConditionArchitecture= has been added
6849 to conditionalize units based on the system architecture, as
6850 reported by uname()'s "machine" field.
6852 * systemd-networkd now supports matching on the system
6853 virtualization, architecture, kernel command line, host name
6856 * logind is now a lot more aggressive when suspending the
6857 machine due to a closed laptop lid. Instead of acting only
6858 on the lid close action, it will continuously watch the lid
6859 status and act on it. This is useful for laptops where the
6860 power button is on the outside of the chassis so that it can
6861 be reached without opening the lid (such as the Lenovo
6862 Yoga). On those machines, logind will now immediately
6863 re-suspend the machine if the power button has been
6864 accidentally pressed while the laptop was suspended and in a
6865 backpack or similar.
6867 * logind will now watch SW_DOCK switches and inhibit reaction
6868 to the lid switch if it is pressed. This means that logind
6869 will not suspend the machine anymore if the lid is closed
6870 and the system is docked, if the laptop supports SW_DOCK
6871 notifications via the input layer. Note that ACPI docking
6872 stations do not generate this currently. Also note that this
6873 logic is usually not fully sufficient and Desktop
6874 Environments should take a lid switch inhibitor lock when an
6875 external display is connected, as systemd will not watch
6878 * nspawn will now make use of the devices cgroup controller by
6879 default, and only permit creation of and access to the usual
6880 API device nodes like /dev/null or /dev/random, as well as
6881 access to (but not creation of) the pty devices.
6883 * We will now ship a default .network file for
6884 systemd-networkd that automatically configures DHCP for
6885 network interfaces created by nspawn's --network-veth or
6886 --network-bridge= switches.
6888 * systemd will now understand the usual M, K, G, T suffixes
6889 according to SI conventions (i.e. to the base 1000) when
6890 referring to throughput and hardware metrics. It will stay
6891 with IEC conventions (i.e. to the base 1024) for software
6892 metrics, according to what is customary according to
6893 Wikipedia. We explicitly document which base applies for
6894 each configuration option.
6896 * The DeviceAllow= setting in unit files now supports a syntax
6897 to whitelist an entire group of devices node majors at once,
6898 based on the /proc/devices listing. For example, with the
6899 string "char-pts", it is now possible to whitelist all
6900 current and future pseudo-TTYs at once.
6902 * sd-event learned a new "post" event source. Event sources of
6903 this type are triggered by the dispatching of any event
6904 source of a type that is not "post". This is useful for
6905 implementing clean-up and check event sources that are
6906 triggered by other work being done in the program.
6908 * systemd-networkd is no longer statically enabled, but uses
6909 the usual [Install] sections so that it can be
6910 enabled/disabled using systemctl. It still is enabled by
6913 * When creating a veth interface pair with systemd-nspawn, the
6914 host side will now be prefixed with "vb-" if
6915 --network-bridge= is used, and with "ve-" if --network-veth
6916 is used. This way, it is easy to distinguish these cases on
6917 the host, for example to apply different configuration to
6918 them with systemd-networkd.
6920 * The compatibility libraries for libsystemd-journal.so,
6921 libsystem-id128.so, libsystemd-login.so and
6922 libsystemd-daemon.so do not make use of IFUNC
6923 anymore. Instead, we now build libsystemd.so multiple times
6924 under these alternative names. This means that the footprint
6925 is drastically increased, but given that these are
6926 transitional compatibility libraries, this should not matter
6927 much. This change has been made necessary to support the ARM
6928 platform for these compatibility libraries, as the ARM
6929 toolchain is not really at the same level as the toolchain
6930 for other architectures like x86 and does not support
6931 IFUNC. Please make sure to use --enable-compat-libs only
6932 during a transitional period!
6934 * The .include syntax has been deprecated and is not documented
6935 anymore. Drop-in files in .d directories should be used instead.
6937 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
6938 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
6939 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
6940 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
6941 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
6942 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
6943 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
6944 Zbigniew Jędrzejewski-Szmek
6946 — Berlin, 2014-02-24
6950 * A new component "systemd-networkd" has been added that can
6951 be used to configure local network interfaces statically or
6952 via DHCP. It is capable of bringing up bridges, VLANs, and
6953 bonding. Currently, no hook-ups for interactive network
6954 configuration are provided. Use this for your initrd,
6955 container, embedded, or server setup if you need a simple,
6956 yet powerful, network configuration solution. This
6957 configuration subsystem is quite nifty, as it allows wildcard
6958 hotplug matching in interfaces. For example, with a single
6959 configuration snippet, you can configure that all Ethernet
6960 interfaces showing up are automatically added to a bridge,
6961 or similar. It supports link-sensing and more.
6963 * A new tool "systemd-socket-proxyd" has been added which can
6964 act as a bidirectional proxy for TCP sockets. This is
6965 useful for adding socket activation support to services that
6966 do not actually support socket activation, including virtual
6967 machines and the like.
6969 * Add a new tool to save/restore rfkill state on
6972 * Save/restore state of keyboard backlights in addition to
6973 display backlights on shutdown/boot.
6975 * udev learned a new SECLABEL{} construct to label device
6976 nodes with a specific security label when they appear. For
6977 now, only SECLABEL{selinux} is supported, but the syntax is
6978 prepared for additional security frameworks.
6980 * udev gained a new scheme to configure link-level attributes
6981 from files in /etc/systemd/network/*.link. These files can
6982 match against MAC address, device path, driver name and type,
6983 and will apply attributes like the naming policy, link speed,
6984 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
6985 address assignment policy (randomized, ...).
6987 * The configuration of network interface naming rules for
6988 "permanent interface names" has changed: a new NamePolicy=
6989 setting in the [Link] section of .link files determines the
6990 priority of possible naming schemes (onboard, slot, MAC,
6991 path). The default value of this setting is determined by
6992 /usr/lib/net/links/99-default.link. Old
6993 80-net-name-slot.rules udev configuration file has been
6994 removed, so local configuration overriding this file should
6995 be adapted to override 99-default.link instead.
6997 * When the User= switch is used in a unit file, also
6998 initialize $SHELL= based on the user database entry.
7000 * systemd no longer depends on libdbus. All communication is
7001 now done with sd-bus, systemd's low-level bus library
7004 * kdbus support has been added to PID 1 itself. When kdbus is
7005 enabled, this causes PID 1 to set up the system bus and
7006 enable support for a new ".busname" unit type that
7007 encapsulates bus name activation on kdbus. It works a little
7008 bit like ".socket" units, except for bus names. A new
7009 generator has been added that converts classic dbus1 service
7010 activation files automatically into native systemd .busname
7013 * sd-bus: add a light-weight vtable implementation that allows
7014 defining objects on the bus with a simple static const
7015 vtable array of its methods, signals and properties.
7017 * systemd will not generate or install static dbus
7018 introspection data anymore to /usr/share/dbus-1/interfaces,
7019 as the precise format of these files is unclear, and
7020 nothing makes use of it.
7022 * A proxy daemon is now provided to proxy clients connecting
7023 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
7024 compatibility with classic D-Bus.
7026 * A bus driver implementation has been added that supports the
7027 classic D-Bus bus driver calls on kdbus, also for
7028 compatibility purposes.
7030 * A new API "sd-event.h" has been added that implements a
7031 minimal event loop API built around epoll. It provides a
7032 couple of features that direct epoll usage is lacking:
7033 prioritization of events, scales to large numbers of timer
7034 events, per-event timer slack (accuracy), system-wide
7035 coalescing of timer events, exit handlers, watchdog
7036 supervision support using systemd's sd_notify() API, child
7039 * A new API "sd-rntl.h" has been added that provides an API
7040 around the route netlink interface of the kernel, similar in
7041 style to "sd-bus.h".
7043 * A new API "sd-dhcp-client.h" has been added that provides a
7044 small DHCPv4 client-side implementation. This is used by
7047 * There is a new kernel command line option
7048 "systemd.restore_state=0|1". When set to "0", none of the
7049 systemd tools will restore saved runtime state to hardware
7050 devices. More specifically, the rfkill and backlight states
7053 * The FsckPassNo= compatibility option in mount/service units
7054 has been removed. The fstab generator will now add the
7055 necessary dependencies automatically, and does not require
7056 PID1's support for that anymore.
7058 * journalctl gained a new switch, --list-boots, that lists
7059 recent boots with their times and boot IDs.
7061 * The various tools like systemctl, loginctl, timedatectl,
7062 busctl, systemd-run, ... have gained a new switch "-M" to
7063 connect to a specific, local OS container (as direct
7064 connection, without requiring SSH). This works on any
7065 container that is registered with machined, such as those
7066 created by libvirt-lxc or nspawn.
7068 * systemd-run and systemd-analyze also gained support for "-H"
7069 to connect to remote hosts via SSH. This is particularly
7070 useful for systemd-run because it enables queuing of jobs
7071 onto remote systems.
7073 * machinectl gained a new command "login" to open a getty
7074 login in any local container. This works with any container
7075 that is registered with machined (such as those created by
7076 libvirt-lxc or nspawn), and which runs systemd inside.
7078 * machinectl gained a new "reboot" command that may be used to
7079 trigger a reboot on a specific container that is registered
7080 with machined. This works on any container that runs an init
7081 system of some kind.
7083 * systemctl gained a new "list-timers" command to print a nice
7084 listing of installed timer units with the times they elapse
7087 * Alternative reboot() parameters may now be specified on the
7088 "systemctl reboot" command line and are passed to the
7089 reboot() system call.
7091 * systemctl gained a new --job-mode= switch to configure the
7092 mode to queue a job with. This is a more generic version of
7093 --fail, --irreversible, and --ignore-dependencies, which are
7094 still available but not advertised anymore.
7096 * /etc/systemd/system.conf gained new settings to configure
7097 various default timeouts of units, as well as the default
7098 start limit interval and burst. These may still be overridden
7101 * PID1 will now export on the bus profile data of the security
7102 policy upload process (such as the SELinux policy upload to
7105 * journald: when forwarding logs to the console, include
7106 timestamps (following the setting in
7107 /sys/module/printk/parameters/time).
7109 * OnCalendar= in timer units now understands the special
7110 strings "yearly" and "annually". (Both are equivalent)
7112 * The accuracy of timer units is now configurable with the new
7113 AccuracySec= setting. It defaults to 1min.
7115 * A new dependency type JoinsNamespaceOf= has been added that
7116 allows running two services within the same /tmp and network
7117 namespace, if PrivateNetwork= or PrivateTmp= are used.
7119 * A new command "cat" has been added to systemctl. It outputs
7120 the original unit file of a unit, and concatenates the
7121 contents of additional "drop-in" unit file snippets, so that
7122 the full configuration is shown.
7124 * systemctl now supports globbing on the various "list-xyz"
7125 commands, like "list-units" or "list-sockets", as well as on
7126 those commands which take multiple unit names.
7128 * journalctl's --unit= switch gained support for globbing.
7130 * All systemd daemons now make use of the watchdog logic so
7131 that systemd automatically notices when they hang.
7133 * If the $container_ttys environment variable is set,
7134 getty-generator will automatically spawn a getty for each
7135 listed tty. This is useful for container managers to request
7136 login gettys to be spawned on as many ttys as needed.
7138 * %h, %s, %U specifier support is not available anymore when
7139 used in unit files for PID 1. This is because NSS calls are
7140 not safe from PID 1. They stay available for --user
7141 instances of systemd, and as special case for the root user.
7143 * loginctl gained a new "--no-legend" switch to turn off output
7146 * The "sd-login.h" API gained three new calls:
7147 sd_session_is_remote(), sd_session_get_remote_user(),
7148 sd_session_get_remote_host() to query information about
7151 * The udev hardware database now also carries vendor/product
7152 information of SDIO devices.
7154 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
7155 determine whether watchdog notifications are requested by
7158 * Socket-activated per-connection services now include a
7159 short description of the connection parameters in the
7162 * tmpfiles gained a new "--boot" option. When this is not used,
7163 only lines where the command character is not suffixed with
7164 "!" are executed. When this option is specified, those
7165 options are executed too. This partitions tmpfiles
7166 directives into those that can be safely executed at any
7167 time, and those which should be run only at boot (for
7168 example, a line that creates /run/nologin).
7170 * A new API "sd-resolve.h" has been added which provides a simple
7171 asynchronous wrapper around glibc NSS host name resolution
7172 calls, such as getaddrinfo(). In contrast to glibc's
7173 getaddrinfo_a(), it does not use signals. In contrast to most
7174 other asynchronous name resolution libraries, this one does
7175 not reimplement DNS, but reuses NSS, so that alternate
7176 host name resolution systems continue to work, such as mDNS,
7177 LDAP, etc. This API is based on libasyncns, but it has been
7178 cleaned up for inclusion in systemd.
7180 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
7181 "sd-daemon.h" are no longer found in individual libraries
7182 libsystemd-journal.so, libsystemd-login.so,
7183 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
7184 merged them into a single library, libsystemd.so, which
7185 provides all symbols. The reason for this is cyclic
7186 dependencies, as these libraries tend to use each other's
7187 symbols. So far, we have managed to workaround that by linking
7188 a copy of a good part of our code into each of these
7189 libraries again and again, which, however, makes certain
7190 things hard to do, like sharing static variables. Also, it
7191 substantially increases footprint. With this change, there
7192 is only one library for the basic APIs systemd
7193 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
7194 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
7195 library as well, however are subject to the --enable-kdbus
7196 switch (see below). Note that "sd-dhcp-client.h" is not part
7197 of this library (this is because it only consumes, never
7198 provides, services of/to other APIs). To make the transition
7199 easy from the separate libraries to the unified one, we
7200 provide the --enable-compat-libs compile-time switch which
7201 will generate stub libraries that are compatible with the
7202 old ones but redirect all calls to the new one.
7204 * All of the kdbus logic and the new APIs "sd-bus.h",
7205 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
7206 and "sd-utf8.h" are compile-time optional via the
7207 "--enable-kdbus" switch, and they are not compiled in by
7208 default. To make use of kdbus, you have to explicitly enable
7209 the switch. Note however, that neither the kernel nor the
7210 userspace API for all of this is considered stable yet. We
7211 want to maintain the freedom to still change the APIs for
7212 now. By specifying this build-time switch, you acknowledge
7213 that you are aware of the instability of the current
7216 * Also, note that while kdbus is pretty much complete,
7217 it lacks one thing: proper policy support. This means you
7218 can build a fully working system with all features; however,
7219 it will be highly insecure. Policy support will be added in
7220 one of the next releases, at the same time that we will
7221 declare the APIs stable.
7223 * When the kernel command line argument "kdbus" is specified,
7224 systemd will automatically load the kdbus.ko kernel module. At
7225 this stage of development, it is only useful for testing kdbus
7226 and should not be used in production. Note: if "--enable-kdbus"
7227 is specified, and the kdbus.ko kernel module is available, and
7228 "kdbus" is added to the kernel command line, the entire system
7229 runs with kdbus instead of dbus-daemon, with the above mentioned
7230 problem of missing the system policy enforcement. Also a future
7231 version of kdbus.ko or a newer systemd will not be compatible with
7232 each other, and will unlikely be able to boot the machine if only
7233 one of them is updated.
7235 * systemctl gained a new "import-environment" command which
7236 uploads the caller's environment (or parts thereof) into the
7237 service manager so that it is inherited by services started
7238 by the manager. This is useful to upload variables like
7239 $DISPLAY into the user service manager.
7241 * A new PrivateDevices= switch has been added to service units
7242 which allows running a service with a namespaced /dev
7243 directory that does not contain any device nodes for
7244 physical devices. More specifically, it only includes devices
7245 such as /dev/null, /dev/urandom, and /dev/zero which are API
7248 * logind has been extended to support behaviour like VT
7249 switching on seats that do not support a VT. This makes
7250 multi-session available on seats that are not the first seat
7251 (seat0), and on systems where kernel support for VTs has
7252 been disabled at compile-time.
7254 * If a process holds a delay lock for system sleep or shutdown
7255 and fails to release it in time, we will now log its
7256 identity. This makes it easier to identify processes that
7257 cause slow suspends or power-offs.
7259 * When parsing /etc/crypttab, support for a new key-slot=
7260 option as supported by Debian is added. It allows indicating
7261 which LUKS slot to use on disk, speeding up key loading.
7263 * The sd_journal_sendv() API call has been checked and
7264 officially declared to be async-signal-safe so that it may
7265 be invoked from signal handlers for logging purposes.
7267 * Boot-time status output is now enabled automatically after a
7268 short timeout if boot does not progress, in order to give
7269 the user an indication what she or he is waiting for.
7271 * The boot-time output has been improved to show how much time
7272 remains until jobs expire.
7274 * The KillMode= switch in service units gained a new possible
7275 value "mixed". If set, and the unit is shut down, then the
7276 initial SIGTERM signal is sent only to the main daemon
7277 process, while the following SIGKILL signal is sent to
7278 all remaining processes of the service.
7280 * When a scope unit is registered, a new property "Controller"
7281 may be set. If set to a valid bus name, systemd will send a
7282 RequestStop() signal to this name when it would like to shut
7283 down the scope. This may be used to hook manager logic into
7284 the shutdown logic of scope units. Also, scope units may now
7285 be put in a special "abandoned" state, in which case the
7286 manager process which created them takes no further
7287 responsibilities for it.
7289 * When reading unit files, systemd will now verify
7290 the access mode of these files, and warn about certain
7291 suspicious combinations. This has been added to make it
7292 easier to track down packaging bugs where unit files are
7293 marked executable or world-writable.
7295 * systemd-nspawn gained a new "--setenv=" switch to set
7296 container-wide environment variables. The similar option in
7297 systemd-activate was renamed from "--environment=" to
7298 "--setenv=" for consistency.
7300 * systemd-nspawn has been updated to create a new kdbus domain
7301 for each container that is invoked, thus allowing each
7302 container to have its own set of system and user buses,
7303 independent of the host.
7305 * systemd-nspawn gained a new --drop-capability= switch to run
7306 the container with less capabilities than the default. Both
7307 --drop-capability= and --capability= now take the special
7308 string "all" for dropping or keeping all capabilities.
7310 * systemd-nspawn gained new switches for executing containers
7311 with specific SELinux labels set.
7313 * systemd-nspawn gained a new --quiet switch to not generate
7314 any additional output but the container's own console
7317 * systemd-nspawn gained a new --share-system switch to run a
7318 container without PID namespacing enabled.
7320 * systemd-nspawn gained a new --register= switch to control
7321 whether the container is registered with systemd-machined or
7322 not. This is useful for containers that do not run full
7323 OS images, but only specific apps.
7325 * systemd-nspawn gained a new --keep-unit which may be used
7326 when invoked as the only program from a service unit, and
7327 results in registration of the unit service itself in
7328 systemd-machined, instead of a newly opened scope unit.
7330 * systemd-nspawn gained a new --network-interface= switch for
7331 moving arbitrary interfaces to the container. The new
7332 --network-veth switch creates a virtual Ethernet connection
7333 between host and container. The new --network-bridge=
7334 switch then allows assigning the host side of this virtual
7335 Ethernet connection to a bridge device.
7337 * systemd-nspawn gained a new --personality= switch for
7338 setting the kernel personality for the container. This is
7339 useful when running a 32-bit container on a 64-bit host. A
7340 similar option Personality= is now also available for service
7343 * logind will now also track a "Desktop" identifier for each
7344 session which encodes the desktop environment of it. This is
7345 useful for desktop environments that want to identify
7346 multiple running sessions of itself easily.
7348 * A new SELinuxContext= setting for service units has been
7349 added that allows setting a specific SELinux execution
7350 context for a service.
7352 * Most systemd client tools will now honour $SYSTEMD_LESS for
7353 settings of the "less" pager. By default, these tools will
7354 override $LESS to allow certain operations to work, such as
7355 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
7356 influence this logic.
7358 * systemd's "seccomp" hook-up has been changed to make use of
7359 the libseccomp library instead of using its own
7360 implementation. This has benefits for portability among
7363 * For usage together with SystemCallFilter=, a new
7364 SystemCallErrorNumber= setting has been introduced that
7365 allows configuration of a system error number to be returned
7366 on filtered system calls, instead of immediately killing the
7367 process. Also, SystemCallArchitectures= has been added to
7368 limit access to system calls of a particular architecture
7369 (in order to turn off support for unused secondary
7370 architectures). There is also a global
7371 SystemCallArchitectures= setting in system.conf now to turn
7372 off support for non-native system calls system-wide.
7374 * systemd requires a kernel with a working name_to_handle_at(),
7375 please see the kernel config requirements in the README file.
7377 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
7378 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
7379 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
7380 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
7381 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
7382 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
7383 Elia Pinto, Florian Weimer, George McCollister, Goffredo
7384 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
7385 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
7386 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
7387 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
7388 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
7389 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
7390 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
7391 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
7392 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
7393 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
7394 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
7395 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
7396 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
7397 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
7398 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
7399 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
7400 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
7402 — Berlin, 2014-02-20
7406 * logind has gained support for facilitating privileged input
7407 and drm device access for unprivileged clients. This work is
7408 useful to allow Wayland display servers (and similar
7409 programs, such as kmscon) to run under the user's ID and
7410 access input and drm devices which are normally
7411 protected. When this is used (and the kernel is new enough)
7412 logind will "mute" IO on the file descriptors passed to
7413 Wayland as long as it is in the background and "unmute" it
7414 if it returns into the foreground. This allows secure
7415 session switching without allowing background sessions to
7416 eavesdrop on input and display data. This also introduces
7417 session switching support if VT support is turned off in the
7418 kernel, and on seats that are not seat0.
7420 * A new kernel command line option luks.options= is understood
7421 now which allows specifying LUKS options for usage for LUKS
7422 encrypted partitions specified with luks.uuid=.
7424 * tmpfiles.d(5) snippets may now use specifier expansion in
7425 path names. More specifically %m, %b, %H, %v, are now
7426 replaced by the local machine id, boot id, hostname, and
7427 kernel version number.
7429 * A new tmpfiles.d(5) command "m" has been introduced which
7430 may be used to change the owner/group/access mode of a file
7431 or directory if it exists, but do nothing if it does not.
7433 * This release removes high-level support for the
7434 MemorySoftLimit= cgroup setting. The underlying kernel
7435 cgroup attribute memory.soft_limit= is currently badly
7436 designed and likely to be removed from the kernel API in its
7437 current form, hence we should not expose it for now.
7439 * The memory.use_hierarchy cgroup attribute is now enabled for
7440 all cgroups systemd creates in the memory cgroup
7441 hierarchy. This option is likely to be come the built-in
7442 default in the kernel anyway, and the non-hierarchical mode
7443 never made much sense in the intrinsically hierarchical
7446 * A new field _SYSTEMD_SLICE= is logged along with all journal
7447 messages containing the slice a message was generated
7448 from. This is useful to allow easy per-customer filtering of
7449 logs among other things.
7451 * systemd-journald will no longer adjust the group of journal
7452 files it creates to the "systemd-journal" group. Instead we
7453 rely on the journal directory to be owned by the
7454 "systemd-journal" group, and its setgid bit set, so that the
7455 kernel file system layer will automatically enforce that
7456 journal files inherit this group assignment. The reason for
7457 this change is that we cannot allow NSS look-ups from
7458 journald which would be necessary to resolve
7459 "systemd-journal" to a numeric GID, because this might
7460 create deadlocks if NSS involves synchronous queries to
7461 other daemons (such as nscd, or sssd) which in turn are
7462 logging clients of journald and might block on it, which
7463 would then dead lock. A tmpfiles.d(5) snippet included in
7464 systemd will make sure the setgid bit and group are
7465 properly set on the journal directory if it exists on every
7466 boot. However, we recommend adjusting it manually after
7467 upgrades too (or from RPM scriptlets), so that the change is
7468 not delayed until next reboot.
7470 * Backlight and random seed files in /var/lib/ have moved into
7471 the /var/lib/systemd/ directory, in order to centralize all
7472 systemd generated files in one directory.
7474 * Boot time performance measurements (as displayed by
7475 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
7476 performance information if that's available to determine how
7477 much time BIOS and boot loader initialization required. With
7478 a sufficiently new BIOS you hence no longer need to boot
7479 with Gummiboot to get access to such information.
7481 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
7482 Cristian Rodríguez, Dave Reisner, David Herrmann, David
7483 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
7484 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
7485 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
7486 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
7487 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
7489 — Berlin, 2013-10-02
7493 * The Restart= option for services now understands a new
7494 on-watchdog setting, which will restart the service
7495 automatically if the service stops sending out watchdog keep
7496 alive messages (as configured with WatchdogSec=).
7498 * The getty generator (which is responsible for bringing up a
7499 getty on configured serial consoles) will no longer only
7500 start a getty on the primary kernel console but on all
7501 others, too. This makes the order in which console= is
7502 specified on the kernel command line less important.
7504 * libsystemd-logind gained a new sd_session_get_vt() call to
7505 retrieve the VT number of a session.
7507 * If the option "tries=0" is set for an entry of /etc/crypttab
7508 its passphrase is queried indefinitely instead of any
7509 maximum number of tries.
7511 * If a service with a configure PID file terminates its PID
7512 file will now be removed automatically if it still exists
7513 afterwards. This should put an end to stale PID files.
7515 * systemd-run will now also take relative binary path names
7516 for execution and no longer insists on absolute paths.
7518 * InaccessibleDirectories= and ReadOnlyDirectories= now take
7519 paths that are optionally prefixed with "-" to indicate that
7520 it should not be considered a failure if they do not exist.
7522 * journalctl -o (and similar commands) now understands a new
7523 output mode "short-precise", it is similar to "short" but
7524 shows timestamps with usec accuracy.
7526 * The option "discard" (as known from Debian) is now
7527 synonymous to "allow-discards" in /etc/crypttab. In fact,
7528 "discard" is preferred now (since it is easier to remember
7531 * Some licensing clean-ups were made, so that more code is now
7532 LGPL-2.1 licensed than before.
7534 * A minimal tool to save/restore the display backlight
7535 brightness across reboots has been added. It will store the
7536 backlight setting as late as possible at shutdown, and
7537 restore it as early as possible during reboot.
7539 * A logic to automatically discover and enable home and swap
7540 partitions on GPT disks has been added. With this in place
7541 /etc/fstab becomes optional for many setups as systemd can
7542 discover certain partitions located on the root disk
7543 automatically. Home partitions are recognized under their
7544 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
7545 partitions are recognized under their GPT type ID
7546 0657fd6da4ab43c484e50933c84b4f4f.
7548 * systemd will no longer pass any environment from the kernel
7549 or initrd to system services. If you want to set an
7550 environment for all services, do so via the kernel command
7551 line systemd.setenv= assignment.
7553 * The systemd-sysctl tool no longer natively reads the file
7554 /etc/sysctl.conf. If desired, the file should be symlinked
7555 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
7556 legacy support by a symlink rather than built-in code, it
7557 also makes the otherwise hidden order of application of the
7558 different files visible. (Note that this partly reverts to a
7559 pre-198 application order of sysctl knobs!)
7561 * The "systemctl set-log-level" and "systemctl dump" commands
7562 have been moved to systemd-analyze.
7564 * systemd-run learned the new --remain-after-exit switch,
7565 which causes the scope unit not to be cleaned up
7566 automatically after the process terminated.
7568 * tmpfiles learned a new --exclude-prefix= switch to exclude
7569 certain paths from operation.
7571 * journald will now automatically flush all messages to disk
7572 as soon as a message at the log level CRIT, ALERT or EMERG
7575 Contributions from: Andrew Cook, Brandon Philips, Christian
7576 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
7577 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
7578 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
7579 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
7580 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
7581 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
7582 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
7583 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
7584 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
7585 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
7586 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
7587 William Giokas, Zbigniew Jędrzejewski-Szmek
7589 — Berlin, 2013-09-13
7593 * The documentation has been updated to cover the various new
7594 concepts introduced with 205.
7596 * Unit files now understand the new %v specifier which
7597 resolves to the kernel version string as returned by "uname
7600 * systemctl now supports filtering the unit list output by
7601 load state, active state and sub state, using the new
7604 * "systemctl status" will now show the results of the
7605 condition checks (like ConditionPathExists= and similar) of
7606 the last start attempts of the unit. They are also logged to
7609 * "journalctl -b" may now be used to look for boot output of a
7610 specific boot. Try "journalctl -b -1" for the previous boot,
7611 but the syntax is substantially more powerful.
7613 * "journalctl --show-cursor" has been added which prints the
7614 cursor string the last shown log line. This may then be used
7615 with the new "journalctl --after-cursor=" switch to continue
7616 browsing logs from that point on.
7618 * "journalctl --force" may now be used to force regeneration
7621 * Creation of "dead" device nodes has been moved from udev
7622 into kmod and tmpfiles. Previously, udev would read the kmod
7623 databases to pre-generate dead device nodes based on meta
7624 information contained in kernel modules, so that these would
7625 be auto-loaded on access rather then at boot. As this
7626 does not really have much to do with the exposing actual
7627 kernel devices to userspace this has always been slightly
7628 alien in the udev codebase. Following the new scheme kmod
7629 will now generate a runtime snippet for tmpfiles from the
7630 module meta information and it now is tmpfiles' job to the
7631 create the nodes. This also allows overriding access and
7632 other parameters for the nodes using the usual tmpfiles
7633 facilities. As side effect this allows us to remove the
7634 CAP_SYS_MKNOD capability bit from udevd entirely.
7636 * logind's device ACLs may now be applied to these "dead"
7637 devices nodes too, thus finally allowing managed access to
7638 devices such as /dev/snd/sequencer without loading the
7639 backing module right-away.
7641 * A new RPM macro has been added that may be used to apply
7642 tmpfiles configuration during package installation.
7644 * systemd-detect-virt and ConditionVirtualization= now can
7645 detect User-Mode-Linux machines (UML).
7647 * journald will now implicitly log the effective capabilities
7648 set of processes in the message metadata.
7650 * systemd-cryptsetup has gained support for TrueCrypt volumes.
7652 * The initrd interface has been simplified (more specifically,
7653 support for passing performance data via environment
7654 variables and fsck results via files in /run has been
7655 removed). These features were non-essential, and are
7656 nowadays available in a much nicer way by having systemd in
7657 the initrd serialize its state and have the hosts systemd
7658 deserialize it again.
7660 * The udev "keymap" data files and tools to apply keyboard
7661 specific mappings of scan to key codes, and force-release
7662 scan code lists have been entirely replaced by a udev
7663 "keyboard" builtin and a hwdb data file.
7665 * systemd will now honour the kernel's "quiet" command line
7666 argument also during late shutdown, resulting in a
7667 completely silent shutdown when used.
7669 * There's now an option to control the SO_REUSEPORT socket
7670 option in .socket units.
7672 * Instance units will now automatically get a per-template
7673 subslice of system.slice unless something else is explicitly
7674 configured. For example, instances of sshd@.service will now
7675 implicitly be placed in system-sshd.slice rather than
7676 system.slice as before.
7678 * Test coverage support may now be enabled at build time.
7680 Contributions from: Dave Reisner, Frederic Crozat, Harald
7681 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
7682 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
7683 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
7684 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
7685 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
7686 Giokas, Zbigniew Jędrzejewski-Szmek
7688 — Berlin, 2013-07-23
7692 * Two new unit types have been introduced:
7694 Scope units are very similar to service units, however, are
7695 created out of pre-existing processes — instead of PID 1
7696 forking off the processes. By using scope units it is
7697 possible for system services and applications to group their
7698 own child processes (worker processes) in a powerful way
7699 which then maybe used to organize them, or kill them
7700 together, or apply resource limits on them.
7702 Slice units may be used to partition system resources in an
7703 hierarchical fashion and then assign other units to them. By
7704 default there are now three slices: system.slice (for all
7705 system services), user.slice (for all user sessions),
7706 machine.slice (for VMs and containers).
7708 Slices and scopes have been introduced primarily in
7709 context of the work to move cgroup handling to a
7710 single-writer scheme, where only PID 1
7711 creates/removes/manages cgroups.
7713 * There's a new concept of "transient" units. In contrast to
7714 normal units these units are created via an API at runtime,
7715 not from configuration from disk. More specifically this
7716 means it is now possible to run arbitrary programs as
7717 independent services, with all execution parameters passed
7718 in via bus APIs rather than read from disk. Transient units
7719 make systemd substantially more dynamic then it ever was,
7720 and useful as a general batch manager.
7722 * logind has been updated to make use of scope and slice units
7723 for managing user sessions. As a user logs in he will get
7724 his own private slice unit, to which all sessions are added
7725 as scope units. We also added support for automatically
7726 adding an instance of user@.service for the user into the
7727 slice. Effectively logind will no longer create cgroup
7728 hierarchies on its own now, it will defer entirely to PID 1
7729 for this by means of scope, service and slice units. Since
7730 user sessions this way become entities managed by PID 1
7731 the output of "systemctl" is now a lot more comprehensive.
7733 * A new mini-daemon "systemd-machined" has been added which
7734 may be used by virtualization managers to register local
7735 VMs/containers. nspawn has been updated accordingly, and
7736 libvirt will be updated shortly. machined will collect a bit
7737 of meta information about the VMs/containers, and assign
7738 them their own scope unit (see above). The collected
7739 meta-data is then made available via the "machinectl" tool,
7740 and exposed in "ps" and similar tools. machined/machinectl
7741 is compile-time optional.
7743 * As discussed earlier, the low-level cgroup configuration
7744 options ControlGroup=, ControlGroupModify=,
7745 ControlGroupPersistent=, ControlGroupAttribute= have been
7746 removed. Please use high-level attribute settings instead as
7747 well as slice units.
7749 * A new bus call SetUnitProperties() has been added to alter
7750 various runtime parameters of a unit. This is primarily
7751 useful to alter cgroup parameters dynamically in a nice way,
7752 but will be extended later on to make more properties
7753 modifiable at runtime. systemctl gained a new set-properties
7754 command that wraps this call.
7756 * A new tool "systemd-run" has been added which can be used to
7757 run arbitrary command lines as transient services or scopes,
7758 while configuring a number of settings via the command
7759 line. This tool is currently very basic, however already
7760 very useful. We plan to extend this tool to even allow
7761 queuing of execution jobs with time triggers from the
7762 command line, similar in fashion to "at".
7764 * nspawn will now inform the user explicitly that kernels with
7765 audit enabled break containers, and suggest the user to turn
7768 * Support for detecting the IMA and AppArmor security
7769 frameworks with ConditionSecurity= has been added.
7771 * journalctl gained a new "-k" switch for showing only kernel
7772 messages, mimicking dmesg output; in addition to "--user"
7773 and "--system" switches for showing only user's own logs
7776 * systemd-delta can now show information about drop-in
7777 snippets extending unit files.
7779 * libsystemd-bus has been substantially updated but is still
7780 not available as public API.
7782 * systemd will now look for the "debug" argument on the kernel
7783 command line and enable debug logging, similar to what
7784 "systemd.log_level=debug" already did before.
7786 * "systemctl set-default", "systemctl get-default" has been
7787 added to configure the default.target symlink, which
7788 controls what to boot into by default.
7790 * "systemctl set-log-level" has been added as a convenient
7791 way to raise and lower systemd logging threshold.
7793 * "systemd-analyze plot" will now show the time the various
7794 generators needed for execution, as well as information
7795 about the unit file loading.
7797 * libsystemd-journal gained a new sd_journal_open_files() call
7798 for opening specific journal files. journactl also gained a
7799 new switch to expose this new functionality. Previously we
7800 only supported opening all files from a directory, or all
7801 files from the system, as opening individual files only is
7802 racy due to journal file rotation.
7804 * systemd gained the new DefaultEnvironment= setting in
7805 /etc/systemd/system.conf to set environment variables for
7808 * If a privileged process logs a journal message with the
7809 OBJECT_PID= field set, then journald will automatically
7810 augment this with additional OBJECT_UID=, OBJECT_GID=,
7811 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
7812 system services want to log events about specific client
7813 processes. journactl/systemctl has been updated to make use
7814 of this information if all log messages regarding a specific
7817 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
7818 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
7819 Reisner, David Coppa, David King, David Strauss, Eelco
7820 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
7821 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
7822 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
7823 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
7824 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
7825 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
7826 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
7827 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
7828 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
7829 Łukasz Stelmach, 장동준
7833 * The Python bindings gained some minimal support for the APIs
7834 exposed by libsystemd-logind.
7836 * ConditionSecurity= gained support for detecting SMACK. Since
7837 this condition already supports SELinux and AppArmor we only
7838 miss IMA for this. Patches welcome!
7840 Contributions from: Karol Lewandowski, Lennart Poettering,
7841 Zbigniew Jędrzejewski-Szmek
7845 * systemd-nspawn will now create /etc/resolv.conf if
7846 necessary, before bind-mounting the host's file onto it.
7848 * systemd-nspawn will now store meta information about a
7849 container on the container's cgroup as extended attribute
7850 fields, including the root directory.
7852 * The cgroup hierarchy has been reworked in many ways. All
7853 objects any of the components systemd creates in the cgroup
7854 tree are now suffixed. More specifically, user sessions are
7855 now placed in cgroups suffixed with ".session", users in
7856 cgroups suffixed with ".user", and nspawn containers in
7857 cgroups suffixed with ".nspawn". Furthermore, all cgroup
7858 names are now escaped in a simple scheme to avoid collision
7859 of userspace object names with kernel filenames. This work
7860 is preparation for making these objects relocatable in the
7861 cgroup tree, in order to allow easy resource partitioning of
7862 these objects without causing naming conflicts.
7864 * systemctl list-dependencies gained the new switches
7865 --plain, --reverse, --after and --before.
7867 * systemd-inhibit now shows the process name of processes that
7868 have taken an inhibitor lock.
7870 * nss-myhostname will now also resolve "localhost"
7871 implicitly. This makes /etc/hosts an optional file and
7872 nicely handles that on IPv6 ::1 maps to both "localhost" and
7875 * libsystemd-logind.so gained a new call
7876 sd_get_machine_names() to enumerate running containers and
7877 VMs (currently only supported by very new libvirt and
7878 nspawn). sd_login_monitor can now be used to watch
7879 VMs/containers coming and going.
7881 * .include is not allowed recursively anymore, and only in
7882 unit files. Usually it is better to use drop-in snippets in
7883 .d/*.conf anyway, as introduced with systemd 198.
7885 * systemd-analyze gained a new "critical-chain" command that
7886 determines the slowest chain of units run during system
7887 boot-up. It is very useful for tracking down where
7888 optimizing boot time is the most beneficial.
7890 * systemd will no longer allow manipulating service paths in
7891 the name=systemd:/system cgroup tree using ControlGroup= in
7892 units. (But is still fine with it in all other dirs.)
7894 * There's a new systemd-nspawn@.service service file that may
7895 be used to easily run nspawn containers as system
7896 services. With the container's root directory in
7897 /var/lib/container/foobar it is now sufficient to run
7898 "systemctl start systemd-nspawn@foobar.service" to boot it.
7900 * systemd-cgls gained a new parameter "--machine" to list only
7901 the processes within a certain container.
7903 * ConditionSecurity= now can check for "apparmor". We still
7904 are lacking checks for SMACK and IMA for this condition
7905 check though. Patches welcome!
7907 * A new configuration file /etc/systemd/sleep.conf has been
7908 added that may be used to configure which kernel operation
7909 systemd is supposed to execute when "suspend", "hibernate"
7910 or "hybrid-sleep" is requested. This makes the new kernel
7911 "freeze" state accessible to the user.
7913 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
7914 the passed argument if applicable.
7916 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
7917 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
7918 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
7919 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
7920 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
7921 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
7922 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
7927 * The output of 'systemctl list-jobs' got some polishing. The
7928 '--type=' argument may now be passed more than once. A new
7929 command 'systemctl list-sockets' has been added which shows
7930 a list of kernel sockets systemd is listening on with the
7931 socket units they belong to, plus the units these socket
7934 * The experimental libsystemd-bus library got substantial
7935 updates to work in conjunction with the (also experimental)
7936 kdbus kernel project. It works well enough to exchange
7937 messages with some sophistication. Note that kdbus is not
7938 ready yet, and the library is mostly an elaborate test case
7939 for now, and not installable.
7941 * systemd gained a new unit 'systemd-static-nodes.service'
7942 that generates static device nodes earlier during boot, and
7943 can run in conjunction with udev.
7945 * libsystemd-login gained a new call sd_pid_get_user_unit()
7946 to retrieve the user systemd unit a process is running
7947 in. This is useful for systems where systemd is used as
7950 * systemd-nspawn now places all containers in the new /machine
7951 top-level cgroup directory in the name=systemd
7952 hierarchy. libvirt will soon do the same, so that we get a
7953 uniform separation of /system, /user and /machine for system
7954 services, user processes and containers/virtual
7955 machines. This new cgroup hierarchy is also useful to stick
7956 stable names to specific container instances, which can be
7957 recognized later this way (this name may be controlled
7958 via systemd-nspawn's new -M switch). libsystemd-login also
7959 gained a new call sd_pid_get_machine_name() to retrieve the
7960 name of the container/VM a specific process belongs to.
7962 * bootchart can now store its data in the journal.
7964 * libsystemd-journal gained a new call
7965 sd_journal_add_conjunction() for AND expressions to the
7966 matching logic. This can be used to express more complex
7967 logical expressions.
7969 * journactl can now take multiple --unit= and --user-unit=
7972 * The cryptsetup logic now understands the "luks.key=" kernel
7973 command line switch for specifying a file to read the
7974 decryption key from. Also, if a configured key file is not
7975 found the tool will now automatically fall back to prompting
7978 * Python systemd.journal module was updated to wrap recently
7979 added functions from libsystemd-journal. The interface was
7980 changed to bring the low level interface in s.j._Reader
7981 closer to the C API, and the high level interface in
7982 s.j.Reader was updated to wrap and convert all data about
7985 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
7986 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
7987 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
7988 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
7989 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
7990 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
7994 * journalctl --update-catalog now understands a new --root=
7995 option to operate on catalogs found in a different root
7998 * During shutdown after systemd has terminated all running
7999 services a final killing loop kills all remaining left-over
8000 processes. We will now print the name of these processes
8001 when we send SIGKILL to them, since this usually indicates a
8004 * If /etc/crypttab refers to password files stored on
8005 configured mount points automatic dependencies will now be
8006 generated to ensure the specific mount is established first
8007 before the key file is attempted to be read.
8009 * 'systemctl status' will now show information about the
8010 network sockets a socket unit is listening on.
8012 * 'systemctl status' will also shown information about any
8013 drop-in configuration file for units. (Drop-In configuration
8014 files in this context are files such as
8015 /etc/systemd/systemd/foobar.service.d/*.conf)
8017 * systemd-cgtop now optionally shows summed up CPU times of
8018 cgroups. Press '%' while running cgtop to switch between
8019 percentage and absolute mode. This is useful to determine
8020 which cgroups use up the most CPU time over the entire
8021 runtime of the system. systemd-cgtop has also been updated
8022 to be 'pipeable' for processing with further shell tools.
8024 * 'hostnamectl set-hostname' will now allow setting of FQDN
8027 * The formatting and parsing of time span values has been
8028 changed. The parser now understands fractional expressions
8029 such as "5.5h". The formatter will now output fractional
8030 expressions for all time spans under 1min, i.e. "5.123456s"
8031 rather than "5s 123ms 456us". For time spans under 1s
8032 millisecond values are shown, for those under 1ms
8033 microsecond values are shown. This should greatly improve
8034 all time-related output of systemd.
8036 * libsystemd-login and libsystemd-journal gained new
8037 functions for querying the poll() events mask and poll()
8038 timeout value for integration into arbitrary event
8041 * localectl gained the ability to list available X11 keymaps
8042 (models, layouts, variants, options).
8044 * 'systemd-analyze dot' gained the ability to filter for
8045 specific units via shell-style globs, to create smaller,
8046 more useful graphs. I.e. it is now possible to create simple
8047 graphs of all the dependencies between only target units, or
8048 of all units that Avahi has dependencies with.
8050 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
8051 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
8052 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
8053 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
8054 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
8055 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
8056 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
8060 * The boot-time readahead implementation for rotating media
8061 will now read the read-ahead data in multiple passes which
8062 consist of all read requests made in equidistant time
8063 intervals. This means instead of strictly reading read-ahead
8064 data in its physical order on disk we now try to find a
8065 middle ground between physical and access time order.
8067 * /etc/os-release files gained a new BUILD_ID= field for usage
8068 on operating systems that provide continuous builds of OS
8071 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
8072 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
8073 William Douglas, Zbigniew Jędrzejewski-Szmek
8077 * systemd-python gained an API exposing libsystemd-daemon.
8079 * The SMACK setup logic gained support for uploading CIPSO
8082 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
8083 ReadOnlyDirectories= and InaccessibleDirectories= has
8084 changed. The private /tmp and /var/tmp directories are now
8085 shared by all processes of a service (which means
8086 ExecStartPre= may now leave data in /tmp that ExecStart= of
8087 the same service can still access). When a service is
8088 stopped its temporary directories are immediately deleted
8089 (normal clean-up with tmpfiles is still done in addition to
8092 * By default, systemd will now set a couple of sysctl
8093 variables in the kernel: the safe sysrq options are turned
8094 on, IP route verification is turned on, and source routing
8095 disabled. The recently added hardlink and softlink
8096 protection of the kernel is turned on. These settings should
8097 be reasonably safe, and good defaults for all new systems.
8099 * The predictable network naming logic may now be turned off
8100 with a new kernel command line switch: net.ifnames=0.
8102 * A new libsystemd-bus module has been added that implements a
8103 pretty complete D-Bus client library. For details see:
8105 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
8107 * journald will now explicitly flush the journal files to disk
8108 at the latest 5min after each write. The file will then also
8109 be marked offline until the next write. This should increase
8110 reliability in case of a crash. The synchronization delay
8111 can be configured via SyncIntervalSec= in journald.conf.
8113 * There's a new remote-fs-setup.target unit that can be used
8114 to pull in specific services when at least one remote file
8115 system is to be mounted.
8117 * There are new targets timers.target and paths.target as
8118 canonical targets to pull user timer and path units in
8119 from. This complements sockets.target with a similar
8120 purpose for socket units.
8122 * libudev gained a new call udev_device_set_attribute_value()
8123 to set sysfs attributes of a device.
8125 * The udev daemon now sets the default number of worker
8126 processes executed in parallel based on the number of available
8127 CPUs instead of the amount of available RAM. This is supposed
8128 to provide a more reliable default and limit a too aggressive
8129 parallelism for setups with 1000s of devices connected.
8131 Contributions from: Auke Kok, Colin Walters, Cristian
8132 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
8133 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
8134 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
8135 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
8136 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
8137 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
8138 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
8139 Zbigniew Jędrzejewski-Szmek
8143 * Configuration of unit files may now be extended via drop-in
8144 files without having to edit/override the unit files
8145 themselves. More specifically, if the administrator wants to
8146 change one value for a service file foobar.service he can
8147 now do so by dropping in a configuration snippet into
8148 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
8149 will load all these snippets and apply them on top of the
8150 main unit configuration file, possibly extending or
8151 overriding its settings. Using these drop-in snippets is
8152 generally nicer than the two earlier options for changing
8153 unit files locally: copying the files from
8154 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
8155 them there; or creating a new file in /etc/systemd/system/
8156 that incorporates the original one via ".include". Drop-in
8157 snippets into these .d/ directories can be placed in any
8158 directory systemd looks for units in, and the usual
8159 overriding semantics between /usr/lib, /etc and /run apply
8162 * Most unit file settings which take lists of items can now be
8163 reset by assigning the empty string to them. For example,
8164 normally, settings such as Environment=FOO=BAR append a new
8165 environment variable assignment to the environment block,
8166 each time they are used. By assigning Environment= the empty
8167 string the environment block can be reset to empty. This is
8168 particularly useful with the .d/*.conf drop-in snippets
8169 mentioned above, since this adds the ability to reset list
8170 settings from vendor unit files via these drop-ins.
8172 * systemctl gained a new "list-dependencies" command for
8173 listing the dependencies of a unit recursively.
8175 * Inhibitors are now honored and listed by "systemctl
8176 suspend", "systemctl poweroff" (and similar) too, not only
8177 GNOME. These commands will also list active sessions by
8180 * Resource limits (as exposed by the various control group
8181 controllers) can now be controlled dynamically at runtime
8182 for all units. More specifically, you can now use a command
8183 like "systemctl set-cgroup-attr foobar.service cpu.shares
8184 2000" to alter the CPU shares a specific service gets. These
8185 settings are stored persistently on disk, and thus allow the
8186 administrator to easily adjust the resource usage of
8187 services with a few simple commands. This dynamic resource
8188 management logic is also available to other programs via the
8189 bus. Almost any kernel cgroup attribute and controller is
8192 * systemd-vconsole-setup will now copy all font settings to
8193 all allocated VTs, where it previously applied them only to
8196 * libsystemd-login gained the new sd_session_get_tty() API
8199 * This release drops support for a few legacy or
8200 distribution-specific LSB facility names when parsing init
8201 scripts: $x-display-manager, $mail-transfer-agent,
8202 $mail-transport-agent, $mail-transfer-agent, $smtp,
8203 $null. Also, the mail-transfer-agent.target unit backing
8204 this has been removed. Distributions which want to retain
8205 compatibility with this should carry the burden for
8206 supporting this themselves and patch support for these back
8207 in, if they really need to. Also, the facilities $syslog and
8208 $local_fs are now ignored, since systemd does not support
8209 early-boot LSB init scripts anymore, and these facilities
8210 are implied anyway for normal services. syslog.target has
8213 * There are new bus calls on PID1's Manager object for
8214 cancelling jobs, and removing snapshot units. Previously,
8215 both calls were only available on the Job and Snapshot
8218 * systemd-journal-gatewayd gained SSL support.
8220 * The various "environment" files, such as /etc/locale.conf
8221 now support continuation lines with a backslash ("\") as
8222 last character in the line, similarly in style (but different)
8223 to how this is supported in shells.
8225 * For normal user processes the _SYSTEMD_USER_UNIT= field is
8226 now implicitly appended to every log entry logged. systemctl
8227 has been updated to filter by this field when operating on a
8228 user systemd instance.
8230 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
8231 CAP_AUDIT_CONTROL capabilities to the capabilities set for
8232 the container. This makes it easier to boot unmodified
8233 Fedora systems in a container, which however still requires
8234 audit=0 to be passed on the kernel command line. Auditing in
8235 kernel and userspace is unfortunately still too broken in
8236 context of containers, hence we recommend compiling it out
8237 of the kernel or using audit=0. Hopefully this will be fixed
8238 one day for good in the kernel.
8240 * nspawn gained the new --bind= and --bind-ro= parameters to
8241 bind mount specific directories from the host into the
8244 * nspawn will now mount its own devpts file system instance
8245 into the container, in order not to leak pty devices from
8246 the host into the container.
8248 * systemd will now read the firmware boot time performance
8249 information from the EFI variables, if the used boot loader
8250 supports this, and takes it into account for boot performance
8251 analysis via "systemd-analyze". This is currently supported
8252 only in conjunction with Gummiboot, but could be supported
8253 by other boot loaders too. For details see:
8255 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
8257 * A new generator has been added that automatically mounts the
8258 EFI System Partition (ESP) to /boot, if that directory
8259 exists, is empty, and no other file system has been
8260 configured to be mounted there.
8262 * logind will now send out PrepareForSleep(false) out
8263 unconditionally, after coming back from suspend. This may be
8264 used by applications as asynchronous notification for
8265 system resume events.
8267 * "systemctl unlock-sessions" has been added, that allows
8268 unlocking the screens of all user sessions at once, similar
8269 to how "systemctl lock-sessions" already locked all users
8270 sessions. This is backed by a new D-Bus call UnlockSessions().
8272 * "loginctl seat-status" will now show the master device of a
8273 seat. (i.e. the device of a seat that needs to be around for
8274 the seat to be considered available, usually the graphics
8277 * tmpfiles gained a new "X" line type, that allows
8278 configuration of files and directories (with wildcards) that
8279 shall be excluded from automatic cleanup ("aging").
8281 * udev default rules set the device node permissions now only
8282 at "add" events, and do not change them any longer with a
8283 later "change" event.
8285 * The log messages for lid events and power/sleep keypresses
8286 now carry a message ID.
8288 * We now have a substantially larger unit test suite, but this
8289 continues to be work in progress.
8291 * udevadm hwdb gained a new --root= parameter to change the
8292 root directory to operate relative to.
8294 * logind will now issue a background sync() request to the kernel
8295 early at shutdown, so that dirty buffers are flushed to disk early
8296 instead of at the last moment, in order to optimize shutdown
8299 * A new bootctl tool has been added that is an interface for
8300 certain boot loader operations. This is currently a preview
8301 and is likely to be extended into a small mechanism daemon
8302 like timedated, localed, hostnamed, and can be used by
8303 graphical UIs to enumerate available boot options, and
8304 request boot into firmware operations.
8306 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
8307 the rest of the package. It also has been updated to work
8308 correctly in initrds.
8310 * polkit previously has been runtime optional, and is now also
8311 compile time optional via a configure switch.
8313 * systemd-analyze has been reimplemented in C. Also "systemctl
8314 dot" has moved into systemd-analyze.
8316 * "systemctl status" with no further parameters will now print
8317 the status of all active or failed units.
8319 * Operations such as "systemctl start" can now be executed
8320 with a new mode "--irreversible" which may be used to queue
8321 operations that cannot accidentally be reversed by a later
8322 job queuing. This is by default used to make shutdown
8323 requests more robust.
8325 * The Python API of systemd now gained a new module for
8326 reading journal files.
8328 * A new tool kernel-install has been added that can install
8329 kernel images according to the Boot Loader Specification:
8331 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
8333 * Boot time console output has been improved to provide
8334 animated boot time output for hanging jobs.
8336 * A new tool systemd-activate has been added which can be used
8337 to test socket activation with, directly from the command
8338 line. This should make it much easier to test and debug
8339 socket activation in daemons.
8341 * journalctl gained a new "--reverse" (or -r) option to show
8342 journal output in reverse order (i.e. newest line first).
8344 * journalctl gained a new "--pager-end" (or -e) option to jump
8345 to immediately jump to the end of the journal in the
8346 pager. This is only supported in conjunction with "less".
8348 * journalctl gained a new "--user-unit=" option, that works
8349 similarly to "--unit=" but filters for user units rather than
8352 * A number of unit files to ease adoption of systemd in
8353 initrds has been added. This moves some minimal logic from
8354 the various initrd implementations into systemd proper.
8356 * The journal files are now owned by a new group
8357 "systemd-journal", which exists specifically to allow access
8358 to the journal, and nothing else. Previously, we used the
8359 "adm" group for that, which however possibly covers more
8360 than just journal/log file access. This new group is now
8361 already used by systemd-journal-gatewayd to ensure this
8362 daemon gets access to the journal files and as little else
8363 as possible. Note that "make install" will also set FS ACLs
8364 up for /var/log/journal to give "adm" and "wheel" read
8365 access to it, in addition to "systemd-journal" which owns
8366 the journal files. We recommend that packaging scripts also
8367 add read access to "adm" + "wheel" to /var/log/journal, and
8368 all existing/future journal files. To normal users and
8369 administrators little changes, however packagers need to
8370 ensure to create the "systemd-journal" system group at
8371 package installation time.
8373 * The systemd-journal-gatewayd now runs as unprivileged user
8374 systemd-journal-gateway:systemd-journal-gateway. Packaging
8375 scripts need to create these system user/group at
8378 * timedated now exposes a new boolean property CanNTP that
8379 indicates whether a local NTP service is available or not.
8381 * systemd-detect-virt will now also detect xen PVs
8383 * The pstore file system is now mounted by default, if it is
8386 * In addition to the SELinux and IMA policies we will now also
8387 load SMACK policies at early boot.
8389 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
8390 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
8391 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
8392 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
8393 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
8394 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
8395 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
8396 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
8397 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
8398 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
8399 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
8400 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
8401 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
8402 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
8406 * Timer units now support calendar time events in addition to
8407 monotonic time events. That means you can now trigger a unit
8408 based on a calendar time specification such as "Thu,Fri
8409 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
8410 or fifth day of any month of the year 2013, given that it is
8411 a thursday or friday. This brings timer event support
8412 considerably closer to cron's capabilities. For details on
8413 the supported calendar time specification language see
8416 * udev now supports a number of different naming policies for
8417 network interfaces for predictable names, and a combination
8418 of these policies is now the default. Please see this wiki
8419 document for details:
8421 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
8423 * Auke Kok's bootchart implementation has been added to the
8424 systemd tree. It is an optional component that can graph the
8425 boot in quite some detail. It is one of the best bootchart
8426 implementations around and minimal in its code and
8429 * nss-myhostname has been integrated into the systemd source
8430 tree. nss-myhostname guarantees that the local hostname
8431 always stays resolvable via NSS. It has been a weak
8432 requirement of systemd-hostnamed since a long time, and
8433 since its code is actually trivial we decided to just
8434 include it in systemd's source tree. It can be turned off
8435 with a configure switch.
8437 * The read-ahead logic is now capable of properly detecting
8438 whether a btrfs file system is on SSD or rotating media, in
8439 order to optimize the read-ahead scheme. Previously, it was
8440 only capable of detecting this on traditional file systems
8443 * In udev, additional device properties are now read from the
8444 IAB in addition to the OUI database. Also, Bluetooth company
8445 identities are attached to the devices as well.
8447 * In service files %U may be used as specifier that is
8448 replaced by the configured user name of the service.
8450 * nspawn may now be invoked without a controlling TTY. This
8451 makes it suitable for invocation as its own service. This
8452 may be used to set up a simple containerized server system
8453 using only core OS tools.
8455 * systemd and nspawn can now accept socket file descriptors
8456 when they are started for socket activation. This enables
8457 implementation of socket activated nspawn
8458 containers. i.e. think about autospawning an entire OS image
8459 when the first SSH or HTTP connection is received. We expect
8460 that similar functionality will also be added to libvirt-lxc
8463 * journalctl will now suppress ANSI color codes when
8464 presenting log data.
8466 * systemctl will no longer show control group information for
8467 a unit if the control group is empty anyway.
8469 * logind can now automatically suspend/hibernate/shutdown the
8472 * /etc/machine-info and hostnamed now also expose the chassis
8473 type of the system. This can be used to determine whether
8474 the local system is a laptop, desktop, handset or
8475 tablet. This information may either be configured by the
8476 user/vendor or is automatically determined from ACPI and DMI
8477 information if possible.
8479 * A number of polkit actions are now bound together with "imply"
8480 rules. This should simplify creating UIs because many actions
8481 will now authenticate similar ones as well.
8483 * Unit files learnt a new condition ConditionACPower= which
8484 may be used to conditionalize a unit depending on whether an
8485 AC power source is connected or not, of whether the system
8486 is running on battery power.
8488 * systemctl gained a new "is-failed" verb that may be used in
8489 shell scripts and suchlike to check whether a specific unit
8490 is in the "failed" state.
8492 * The EnvironmentFile= setting in unit files now supports file
8493 globbing, and can hence be used to easily read a number of
8494 environment files at once.
8496 * systemd will no longer detect and recognize specific
8497 distributions. All distribution-specific #ifdeffery has been
8498 removed, systemd is now fully generic and
8499 distribution-agnostic. Effectively, not too much is lost as
8500 a lot of the code is still accessible via explicit configure
8501 switches. However, support for some distribution specific
8502 legacy configuration file formats has been dropped. We
8503 recommend distributions to simply adopt the configuration
8504 files everybody else uses now and convert the old
8505 configuration from packaging scripts. Most distributions
8506 already did that. If that's not possible or desirable,
8507 distributions are welcome to forward port the specific
8508 pieces of code locally from the git history.
8510 * When logging a message about a unit systemd will now always
8511 log the unit name in the message meta data.
8513 * localectl will now also discover system locale data that is
8514 not stored in locale archives, but directly unpacked.
8516 * logind will no longer unconditionally use framebuffer
8517 devices as seat masters, i.e. as devices that are required
8518 to be existing before a seat is considered preset. Instead,
8519 it will now look for all devices that are tagged as
8520 "seat-master" in udev. By default, framebuffer devices will
8521 be marked as such, but depending on local systems, other
8522 devices might be marked as well. This may be used to
8523 integrate graphics cards using closed source drivers (such
8524 as NVidia ones) more nicely into logind. Note however, that
8525 we recommend using the open source NVidia drivers instead,
8526 and no udev rules for the closed-source drivers will be
8527 shipped from us upstream.
8529 Contributions from: Adam Williamson, Alessandro Crismani, Auke
8530 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
8531 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
8532 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
8533 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
8534 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
8535 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
8536 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
8537 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
8538 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
8539 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
8540 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
8545 * udev gained support for loading additional device properties
8546 from an indexed database that is keyed by vendor/product IDs
8547 and similar device identifiers. For the beginning this
8548 "hwdb" is populated with data from the well-known PCI and
8549 USB database, but also includes PNP, ACPI and OID data. In
8550 the longer run this indexed database shall grow into
8551 becoming the one central database for non-essential
8552 userspace device metadata. Previously, data from the PCI/USB
8553 database was only attached to select devices, since the
8554 lookup was a relatively expensive operation due to O(n) time
8555 complexity (with n being the number of entries in the
8556 database). Since this is now O(1), we decided to add in this
8557 data for all devices where this is available, by
8558 default. Note that the indexed database needs to be rebuilt
8559 when new data files are installed. To achieve this you need
8560 to update your packaging scripts to invoke "udevadm hwdb
8561 --update" after installation of hwdb data files. For
8562 RPM-based distributions we introduced the new
8563 %udev_hwdb_update macro for this purpose.
8565 * The Journal gained support for the "Message Catalog", an
8566 indexed database to link up additional information with
8567 journal entries. For further details please check:
8569 https://www.freedesktop.org/wiki/Software/systemd/catalog
8571 The indexed message catalog database also needs to be
8572 rebuilt after installation of message catalog files. Use
8573 "journalctl --update-catalog" for this. For RPM-based
8574 distributions we introduced the %journal_catalog_update
8575 macro for this purpose.
8577 * The Python Journal bindings gained support for the standard
8578 Python logging framework.
8580 * The Journal API gained new functions for checking whether
8581 the underlying file system of a journal file is capable of
8582 properly reporting file change notifications, or whether
8583 applications that want to reflect journal changes "live"
8584 need to recheck journal files continuously in appropriate
8587 * It is now possible to set the "age" field for tmpfiles
8588 entries to 0, indicating that files matching this entry
8589 shall always be removed when the directories are cleaned up.
8591 * coredumpctl gained a new "gdb" verb which invokes gdb
8592 right-away on the selected coredump.
8594 * There's now support for "hybrid sleep" on kernels that
8595 support this, in addition to "suspend" and "hibernate". Use
8596 "systemctl hybrid-sleep" to make use of this.
8598 * logind's HandleSuspendKey= setting (and related settings)
8599 now gained support for a new "lock" setting to simply
8600 request the screen lock on all local sessions, instead of
8601 actually executing a suspend or hibernation.
8603 * systemd will now mount the EFI variables file system by
8606 * Socket units now gained support for configuration of the
8607 SMACK security label.
8609 * timedatectl will now output the time of the last and next
8610 daylight saving change.
8612 * We dropped support for various legacy and distro-specific
8613 concepts, such as insserv, early-boot SysV services
8614 (i.e. those for non-standard runlevels such as 'b' or 'S')
8615 or ArchLinux /etc/rc.conf support. We recommend the
8616 distributions who still need support this to either continue
8617 to maintain the necessary patches downstream, or find a
8618 different solution. (Talk to us if you have questions!)
8620 * Various systemd components will now bypass polkit checks for
8621 root and otherwise handle properly if polkit is not found to
8622 be around. This should fix most issues for polkit-less
8623 systems. Quite frankly this should have been this way since
8624 day one. It is absolutely our intention to make systemd work
8625 fine on polkit-less systems, and we consider it a bug if
8626 something does not work as it should if polkit is not around.
8628 * For embedded systems it is now possible to build udev and
8629 systemd without blkid and/or kmod support.
8631 * "systemctl switch-root" is now capable of switching root
8632 more than once. I.e. in addition to transitions from the
8633 initrd to the host OS it is now possible to transition to
8634 further OS images from the host. This is useful to implement
8635 offline updating tools.
8637 * Various other additions have been made to the RPM macros
8638 shipped with systemd. Use %udev_rules_update() after
8639 installing new udev rules files. %_udevhwdbdir,
8640 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
8641 %_sysctldir are now available which resolve to the right
8642 directories for packages to place various data files in.
8644 * journalctl gained the new --full switch (in addition to
8645 --all, to disable ellipsation for long messages.
8647 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
8648 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
8649 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
8650 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
8651 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
8652 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
8653 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
8654 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
8655 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
8659 * journalctl gained new --since= and --until= switches to
8660 filter by time. It also now supports nice filtering for
8661 units via --unit=/-u.
8663 * Type=oneshot services may use ExecReload= and do the
8666 * The journal daemon now supports time-based rotation and
8667 vacuuming, in addition to the usual disk-space based
8670 * The journal will now index the available field values for
8671 each field name. This enables clients to show pretty drop
8672 downs of available match values when filtering. The bash
8673 completion of journalctl has been updated
8674 accordingly. journalctl gained a new switch -F to list all
8675 values a certain field takes in the journal database.
8677 * More service events are now written as structured messages
8678 to the journal, and made recognizable via message IDs.
8680 * The timedated, localed and hostnamed mini-services which
8681 previously only provided support for changing time, locale
8682 and hostname settings from graphical DEs such as GNOME now
8683 also have a minimal (but very useful) text-based client
8684 utility each. This is probably the nicest way to changing
8685 these settings from the command line now, especially since
8686 it lists available options and is fully integrated with bash
8689 * There's now a new tool "systemd-coredumpctl" to list and
8690 extract coredumps from the journal.
8692 * We now install a README each in /var/log/ and
8693 /etc/rc.d/init.d explaining where the system logs and init
8694 scripts went. This hopefully should help folks who go to
8695 that dirs and look into the otherwise now empty void and
8696 scratch their heads.
8698 * When user-services are invoked (by systemd --user) the
8699 $MANAGERPID env var is set to the PID of systemd.
8701 * SIGRTMIN+24 when sent to a --user instance will now result
8702 in immediate termination of systemd.
8704 * gatewayd received numerous feature additions such as a
8705 "follow" mode, for live syncing and filtering.
8707 * browse.html now allows filtering and showing detailed
8708 information on specific entries. Keyboard navigation and
8709 mouse screen support has been added.
8711 * gatewayd/journalctl now supports HTML5/JSON
8712 Server-Sent-Events as output.
8714 * The SysV init script compatibility logic will now
8715 heuristically determine whether a script supports the
8716 "reload" verb, and only then make this available as
8719 * "systemctl status --follow" has been removed, use "journalctl
8722 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
8723 have been removed since they are hardly useful to be
8726 * And I'd like to take the opportunity to specifically mention
8727 Zbigniew for his great contributions. Zbigniew, you rock!
8729 Contributions from: Andrew Eikum, Christian Hesse, Colin
8730 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
8731 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
8732 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
8733 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
8734 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
8735 Jędrzejewski-Szmek, Сковорода Никита Андреевич
8739 * If /etc/vconsole.conf is non-existent or empty we will no
8740 longer load any console font or key map at boot by
8741 default. Instead the kernel defaults will be left
8742 intact. This is definitely the right thing to do, as no
8743 configuration should mean no configuration, and hard-coding
8744 font names that are different on all archs is probably a bad
8745 idea. Also, the kernel default key map and font should be
8746 good enough for most cases anyway, and mostly identical to
8747 the userspace fonts/key maps we previously overloaded them
8748 with. If distributions want to continue to default to a
8749 non-kernel font or key map they should ship a default
8750 /etc/vconsole.conf with the appropriate contents.
8752 Contributions from: Colin Walters, Daniel J Walsh, Dave
8753 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
8754 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
8758 * journalctl gained a new --cursor= switch to show entries
8759 starting from the specified location in the journal.
8761 * We now enforce a size limit on journal entry fields exported
8762 with "-o json" in journalctl. Fields larger than 4K will be
8763 assigned null. This can be turned off with --all.
8765 * An (optional) journal gateway daemon is now available as
8766 "systemd-journal-gatewayd.service". This service provides
8767 access to the journal via HTTP and JSON. This functionality
8768 will be used to implement live log synchronization in both
8769 pull and push modes, but has various other users too, such
8770 as easy log access for debugging of embedded devices. Right
8771 now it is already useful to retrieve the journal via HTTP:
8773 # systemctl start systemd-journal-gatewayd.service
8774 # wget http://localhost:19531/entries
8776 This will download the journal contents in a
8777 /var/log/messages compatible format. The same as JSON:
8779 # curl -H"Accept: application/json" http://localhost:19531/entries
8781 This service is also accessible via a web browser where a
8782 single static HTML5 app is served that uses the JSON logic
8783 to enable the user to do some basic browsing of the
8784 journal. This will be extended later on. Here's an example
8785 screenshot of this app in its current state:
8787 http://0pointer.de/public/journal-gatewayd
8789 Contributions from: Kay Sievers, Lennart Poettering, Robert
8790 Milasan, Tom Gundersen
8794 * The bash completion logic is now available for journalctl
8797 * We do not mount the "cpuset" controller anymore together with
8798 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
8799 started if no parameters are assigned to it. "cpuset" hence
8800 broke code that assumed it could create "cpu" groups and
8803 * journalctl -f will now subscribe to terminal size changes,
8804 and line break accordingly.
8806 Contributions from: Dave Reisner, Kay Sievers, Lennart
8807 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
8811 * nspawn will now create a symlink /etc/localtime in the
8812 container environment, copying the host's timezone
8813 setting. Previously this has been done via a bind mount, but
8814 since symlinks cannot be bind mounted this has now been
8815 changed to create/update the appropriate symlink.
8817 * journalctl -n's line number argument is now optional, and
8818 will default to 10 if omitted.
8820 * journald will now log the maximum size the journal files may
8821 take up on disk. This is particularly useful if the default
8822 built-in logic of determining this parameter from the file
8823 system size is used. Use "systemctl status
8824 systemd-journald.service" to see this information.
8826 * The multi-seat X wrapper tool has been stripped down. As X
8827 is now capable of enumerating graphics devices via udev in a
8828 seat-aware way the wrapper is not strictly necessary
8829 anymore. A stripped down temporary stop-gap is still shipped
8830 until the upstream display managers have been updated to
8831 fully support the new X logic. Expect this wrapper to be
8832 removed entirely in one of the next releases.
8834 * HandleSleepKey= in logind.conf has been split up into
8835 HandleSuspendKey= and HandleHibernateKey=. The old setting
8836 is not available anymore. X11 and the kernel are
8837 distinguishing between these keys and we should too. This
8838 also means the inhibition lock for these keys has been split
8841 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
8842 Poettering, Lukas Nykryn, Václav Pavlín
8846 * Whenever a unit changes state we will now log this to the
8847 journal and show along the unit's own log output in
8850 * ConditionPathIsMountPoint= can now properly detect bind
8851 mount points too. (Previously, a bind mount of one file
8852 system to another place in the same file system could not be
8853 detected as mount, since they shared struct stat's st_dev
8856 * We will now mount the cgroup controllers cpu, cpuacct,
8857 cpuset and the controllers net_cls, net_prio together by
8860 * nspawn containers will now have a virtualized boot
8861 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
8862 over with a randomized ID at container initialization). This
8863 has the effect of making "journalctl -b" do the right thing
8866 * The JSON output journal serialization has been updated not
8867 to generate "endless" list objects anymore, but rather one
8868 JSON object per line. This is more in line how most JSON
8869 parsers expect JSON objects. The new output mode
8870 "json-pretty" has been added to provide similar output, but
8871 neatly aligned for readability by humans.
8873 * We dropped all explicit sync() invocations in the shutdown
8874 code. The kernel does this implicitly anyway in the kernel
8875 reboot() syscall. halt(8)'s -n option is now a compatibility
8878 * We now support virtualized reboot() in containers, as
8879 supported by newer kernels. We will fall back to exit() if
8880 CAP_SYS_REBOOT is not available to the container. Also,
8881 nspawn makes use of this now and will actually reboot the
8882 container if the containerized OS asks for that.
8884 * journalctl will only show local log output by default
8885 now. Use --merge (-m) to show remote log output, too.
8887 * libsystemd-journal gained the new sd_journal_get_usage()
8888 call to determine the current disk usage of all journal
8889 files. This is exposed in the new "journalctl --disk-usage"
8892 * journald gained a new configuration setting SplitMode= in
8893 journald.conf which may be used to control how user journals
8894 are split off. See journald.conf(5) for details.
8896 * A new condition type ConditionFileNotEmpty= has been added.
8898 * tmpfiles' "w" lines now support file globbing, to write
8899 multiple files at once.
8901 * We added Python bindings for the journal submission
8902 APIs. More Python APIs for a number of selected APIs will
8903 likely follow. Note that we intend to add native bindings
8904 only for the Python language, as we consider it common
8905 enough to deserve bindings shipped within systemd. There are
8906 various projects outside of systemd that provide bindings
8907 for languages such as PHP or Lua.
8909 * Many conditions will now resolve specifiers such as %i. In
8910 addition, PathChanged= and related directives of .path units
8911 now support specifiers as well.
8913 * There's now a new RPM macro definition for the system preset
8916 * journald will now warn if it ca not forward a message to the
8917 syslog daemon because its socket is full.
8919 * timedated will no longer write or process /etc/timezone,
8920 except on Debian. As we do not support late mounted /usr
8921 anymore /etc/localtime always being a symlink is now safe,
8922 and hence the information in /etc/timezone is not necessary
8925 * logind will now always reserve one VT for a text getty (VT6
8926 by default). Previously if more than 6 X sessions where
8927 started they took up all the VTs with auto-spawned gettys,
8928 so that no text gettys were available anymore.
8930 * udev will now automatically inform the btrfs kernel logic
8931 about btrfs RAID components showing up. This should make
8932 simple hotplug based btrfs RAID assembly work.
8934 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
8935 (but not for its children which will stay at the kernel
8936 default). This should allow setups with a lot more listening
8939 * systemd will now always pass the configured timezone to the
8940 kernel at boot. timedated will do the same when the timezone
8943 * logind's inhibition logic has been updated. By default,
8944 logind will now handle the lid switch, the power and sleep
8945 keys all the time, even in graphical sessions. If DEs want
8946 to handle these events on their own they should take the new
8947 handle-power-key, handle-sleep-key and handle-lid-switch
8948 inhibitors during their runtime. A simple way to achieve
8949 that is to invoke the DE wrapped in an invocation of:
8951 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
8953 * Access to unit operations is now checked via SELinux taking
8954 the unit file label and client process label into account.
8956 * systemd will now notify the administrator in the journal
8957 when he over-mounts a non-empty directory.
8959 * There are new specifiers that are resolved in unit files,
8960 for the host name (%H), the machine ID (%m) and the boot ID
8963 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
8964 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
8965 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
8966 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
8967 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
8968 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
8969 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
8973 * Support for reading structured kernel messages from
8974 /dev/kmsg has now been added and is enabled by default.
8976 * Support for reading kernel messages from /proc/kmsg has now
8977 been removed. If you want kernel messages in the journal
8978 make sure to run a recent kernel (>= 3.5) that supports
8979 reading structured messages from /dev/kmsg (see
8980 above). /proc/kmsg is now exclusive property of classic
8981 syslog daemons again.
8983 * The libudev API gained the new
8984 udev_device_new_from_device_id() call.
8986 * The logic for file system namespace (ReadOnlyDirectory=,
8987 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
8988 require pivot_root() anymore. This means fewer temporary
8989 directories are created below /tmp for this feature.
8991 * nspawn containers will now see and receive all submounts
8992 made on the host OS below the root file system of the
8995 * Forward Secure Sealing is now supported for Journal files,
8996 which provide cryptographical sealing of journal files so
8997 that attackers cannot alter log history anymore without this
8998 being detectable. Lennart will soon post a blog story about
8999 this explaining it in more detail.
9001 * There are two new service settings RestartPreventExitStatus=
9002 and SuccessExitStatus= which allow configuration of exit
9003 status (exit code or signal) which will be excepted from the
9004 restart logic, resp. consider successful.
9006 * journalctl gained the new --verify switch that can be used
9007 to check the integrity of the structure of journal files and
9008 (if Forward Secure Sealing is enabled) the contents of
9011 * nspawn containers will now be run with /dev/stdin, /dev/fd/
9012 and similar symlinks pre-created. This makes running shells
9013 as container init process a lot more fun.
9015 * The fstab support can now handle PARTUUID= and PARTLABEL=
9018 * A new ConditionHost= condition has been added to match
9019 against the hostname (with globs) and machine ID. This is
9020 useful for clusters where a single OS image is used to
9021 provision a large number of hosts which shall run slightly
9022 different sets of services.
9024 * Services which hit the restart limit will now be placed in a
9027 Contributions from: Bertram Poettering, Dave Reisner, Huang
9028 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
9029 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
9033 * When running in --user mode systemd will now become a
9034 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
9035 tree a lot more organized.
9037 * A new PartOf= unit dependency type has been introduced that
9038 may be used to group services in a natural way.
9040 * "systemctl enable" may now be used to enable instances of
9043 * journalctl now prints error log levels in red, and
9044 warning/notice log levels in bright white. It also supports
9045 filtering by log level now.
9047 * cgtop gained a new -n switch (similar to top), to configure
9048 the maximum number of iterations to run for. It also gained
9049 -b, to run in batch mode (accepting no input).
9051 * The suffix ".service" may now be omitted on most systemctl
9052 command lines involving service unit names.
9054 * There's a new bus call in logind to lock all sessions, as
9055 well as a loginctl verb for it "lock-sessions".
9057 * libsystemd-logind.so gained a new call sd_journal_perror()
9058 that works similar to libc perror() but logs to the journal
9059 and encodes structured information about the error number.
9061 * /etc/crypttab entries now understand the new keyfile-size=
9064 * shutdown(8) now can send a (configurable) wall message when
9065 a shutdown is cancelled.
9067 * The mount propagation mode for the root file system will now
9068 default to "shared", which is useful to make containers work
9069 nicely out-of-the-box so that they receive new mounts from
9070 the host. This can be undone locally by running "mount
9071 --make-rprivate /" if needed.
9073 * The prefdm.service file has been removed. Distributions
9074 should maintain this unit downstream if they intend to keep
9075 it around. However, we recommend writing normal unit files
9076 for display managers instead.
9078 * Since systemd is a crucial part of the OS we will now
9079 default to a number of compiler switches that improve
9080 security (hardening) such as read-only relocations, stack
9081 protection, and suchlike.
9083 * The TimeoutSec= setting for services is now split into
9084 TimeoutStartSec= and TimeoutStopSec= to allow configuration
9085 of individual time outs for the start and the stop phase of
9088 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
9089 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
9090 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
9091 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
9092 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
9093 Gundersen, Zbigniew Jędrzejewski-Szmek
9097 * The journal and id128 C APIs are now fully documented as man
9100 * Extra safety checks have been added when transitioning from
9101 the initial RAM disk to the main system to avoid accidental
9104 * /etc/crypttab entries now understand the new keyfile-offset=
9107 * systemctl -t can now be used to filter by unit load state.
9109 * The journal C API gained the new sd_journal_wait() call to
9110 make writing synchronous journal clients easier.
9112 * journalctl gained the new -D switch to show journals from a
9115 * journalctl now displays a special marker between log
9116 messages of two different boots.
9118 * The journal is now explicitly flushed to /var via a service
9119 systemd-journal-flush.service, rather than implicitly simply
9120 by seeing /var/log/journal to be writable.
9122 * journalctl (and the journal C APIs) can now match for much
9123 more complex expressions, with alternatives and
9126 * When transitioning from the initial RAM disk to the main
9127 system we will now kill all processes in a killing spree to
9128 ensure no processes stay around by accident.
9130 * Three new specifiers may be used in unit files: %u, %h, %s
9131 resolve to the user name, user home directory resp. user
9132 shell. This is useful for running systemd user instances.
9134 * We now automatically rotate journal files if their data
9135 object hash table gets a fill level > 75%. We also size the
9136 hash table based on the configured maximum file size. This
9137 together should lower hash collisions drastically and thus
9138 speed things up a bit.
9140 * journalctl gained the new "--header" switch to introspect
9141 header data of journal files.
9143 * A new setting SystemCallFilters= has been added to services
9144 which may be used to apply blacklists or whitelists to
9145 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
9147 * nspawn gained a new --link-journal= switch (and quicker: -j)
9148 to link the container journal with the host. This makes it
9149 very easy to centralize log viewing on the host for all
9150 guests while still keeping the journal files separated.
9152 * Many bugfixes and optimizations
9154 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
9155 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
9156 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
9161 * Several tools now understand kernel command line arguments,
9162 which are only read when run in an initial RAM disk. They
9163 usually follow closely their normal counterparts, but are
9166 * There's a new tool to analyze the readahead files that are
9167 automatically generated at boot. Use:
9169 /usr/lib/systemd/systemd-readahead analyze /.readahead
9171 * We now provide an early debug shell on tty9 if this enabled. Use:
9173 systemctl enable debug-shell.service
9175 * All plymouth related units have been moved into the Plymouth
9176 package. Please make sure to upgrade your Plymouth version
9179 * systemd-tmpfiles now supports getting passed the basename of
9180 a configuration file only, in which case it will look for it
9181 in all appropriate directories automatically.
9183 * udevadm info now takes a /dev or /sys path as argument, and
9184 does the right thing. Example:
9186 udevadm info /dev/sda
9187 udevadm info /sys/class/block/sda
9189 * systemctl now prints a warning if a unit is stopped but a
9190 unit that might trigger it continues to run. Example: a
9191 service is stopped but the socket that activates it is left
9194 * "systemctl status" will now mention if the log output was
9195 shortened due to rotation since a service has been started.
9197 * The journal API now exposes functions to determine the
9198 "cutoff" times due to rotation.
9200 * journald now understands SIGUSR1 and SIGUSR2 for triggering
9201 immediately flushing of runtime logs to /var if possible,
9202 resp. for triggering immediate rotation of the journal
9205 * It is now considered an error if a service is attempted to
9206 be stopped that is not loaded.
9208 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
9210 * systemd-analyze now supports Python 3
9212 * tmpfiles now supports cleaning up directories via aging
9213 where the first level dirs are always kept around but
9214 directories beneath it automatically aged. This is enabled
9215 by prefixing the age field with '~'.
9217 * Seat objects now expose CanGraphical, CanTTY properties
9218 which is required to deal with very fast bootups where the
9219 display manager might be running before the graphics drivers
9220 completed initialization.
9222 * Seat objects now expose a State property.
9224 * We now include RPM macros for service enabling/disabling
9225 based on the preset logic. We recommend RPM based
9226 distributions to make use of these macros if possible. This
9227 makes it simpler to reuse RPM spec files across
9230 * We now make sure that the collected systemd unit name is
9231 always valid when services log to the journal via
9234 * There's a new man page kernel-command-line(7) detailing all
9235 command line options we understand.
9237 * The fstab generator may now be disabled at boot by passing
9238 fstab=0 on the kernel command line.
9240 * A new kernel command line option modules-load= is now understood
9241 to load a specific kernel module statically, early at boot.
9243 * Unit names specified on the systemctl command line are now
9244 automatically escaped as needed. Also, if file system or
9245 device paths are specified they are automatically turned
9246 into the appropriate mount or device unit names. Example:
9248 systemctl status /home
9249 systemctl status /dev/sda
9251 * The SysVConsole= configuration option has been removed from
9252 system.conf parsing.
9254 * The SysV search path is no longer exported on the D-Bus
9257 * The Names= option has been removed from unit file parsing.
9259 * There's a new man page bootup(7) detailing the boot process.
9261 * Every unit and every generator we ship with systemd now
9262 comes with full documentation. The self-explanatory boot is
9265 * A couple of services gained "systemd-" prefixes in their
9266 name if they wrap systemd code, rather than only external
9267 code. Among them fsck@.service which is now
9268 systemd-fsck@.service.
9270 * The HaveWatchdog property has been removed from the D-Bus
9273 * systemd.confirm_spawn= on the kernel command line should now
9276 * There's a new man page crypttab(5) which details all options
9277 we actually understand.
9279 * systemd-nspawn gained a new --capability= switch to pass
9280 additional capabilities to the container.
9282 * timedated will now read known NTP implementation unit names
9283 from /usr/lib/systemd/ntp-units.d/*.list,
9284 systemd-timedated-ntp.target has been removed.
9286 * journalctl gained a new switch "-b" that lists log data of
9287 the current boot only.
9289 * The notify socket is in the abstract namespace again, in
9290 order to support daemons which chroot() at start-up.
9292 * There is a new Storage= configuration option for journald
9293 which allows configuration of where log data should go. This
9294 also provides a way to disable journal logging entirely, so
9295 that data collected is only forwarded to the console, the
9296 kernel log buffer or another syslog implementation.
9298 * Many bugfixes and optimizations
9300 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
9301 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
9302 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
9303 Shawn Landden, Tom Gundersen
9307 * "systemctl help <unit>" now shows the man page if one is
9310 * Several new man pages have been added.
9312 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
9313 MaxLevelConsole= can now be specified in
9314 journald.conf. These options allow reducing the amount of
9315 data stored on disk or forwarded by the log level.
9317 * TimerSlackNSec= can now be specified in system.conf for
9318 PID1. This allows system-wide power savings.
9320 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
9321 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
9326 * logind is now capable of (optionally) handling power and
9327 sleep keys as well as the lid switch.
9329 * journalctl now understands the syntax "journalctl
9330 /usr/bin/avahi-daemon" to get all log output of a specific
9333 * CapabilityBoundingSet= in system.conf now also influences
9334 the capability bound set of usermode helpers of the kernel.
9336 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
9337 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
9338 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
9339 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
9343 * Note that we skipped 139 releases here in order to set the
9344 new version to something that is greater than both udev's
9345 and systemd's most recent version number.
9347 * udev: all udev sources are merged into the systemd source tree now.
9348 All future udev development will happen in the systemd tree. It
9349 is still fully supported to use the udev daemon and tools without
9350 systemd running, like in initramfs or other init systems. Building
9351 udev though, will require the *build* of the systemd tree, but
9352 udev can be properly *run* without systemd.
9354 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
9355 should be used to create dead device nodes as workarounds for broken
9358 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
9359 no longer supported. udev_monitor_new_from_netlink() needs to be
9360 used to subscribe to events.
9362 * udev: when udevd is started by systemd, processes which are left
9363 behind by forking them off of udev rules, are unconditionally cleaned
9364 up and killed now after the event handling has finished. Services or
9365 daemons must be started as systemd services. Services can be
9366 pulled-in by udev to get started, but they can no longer be directly
9367 forked by udev rules.
9369 * udev: the daemon binary is called systemd-udevd now and installed
9370 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
9371 to adapt to that, create symlink, or rename the binary after building
9374 * libudev no longer provides these symbols:
9375 udev_monitor_from_socket()
9376 udev_queue_get_failed_list_entry()
9377 udev_get_{dev,sys,run}_path()
9378 The versions number was bumped and symbol versioning introduced.
9380 * systemd-loginctl and systemd-journalctl have been renamed
9381 to loginctl and journalctl to match systemctl.
9383 * The config files: /etc/systemd/systemd-logind.conf and
9384 /etc/systemd/systemd-journald.conf have been renamed to
9385 logind.conf and journald.conf. Package updates should rename
9386 the files to the new names on upgrade.
9388 * For almost all files the license is now LGPL2.1+, changed
9389 from the previous GPL2.0+. Exceptions are some minor stuff
9390 of udev (which will be changed to LGPL2.1 eventually, too),
9391 and the MIT licensed sd-daemon.[ch] library that is suitable
9392 to be used as drop-in files.
9394 * systemd and logind now handle system sleep states, in
9395 particular suspending and hibernating.
9397 * logind now implements a sleep/shutdown/idle inhibiting logic
9398 suitable for a variety of uses. Soonishly Lennart will blog
9399 about this in more detail.
9401 * var-run.mount and var-lock.mount are no longer provided
9402 (which previously bind mounted these directories to their new
9403 places). Distributions which have not converted these
9404 directories to symlinks should consider stealing these files
9405 from git history and add them downstream.
9407 * We introduced the Documentation= field for units and added
9408 this to all our shipped units. This is useful to make it
9409 easier to explore the boot and the purpose of the various
9412 * All smaller setup units (such as
9413 systemd-vconsole-setup.service) now detect properly if they
9414 are run in a container and are skipped when
9415 appropriate. This guarantees an entirely noise-free boot in
9416 Linux container environments such as systemd-nspawn.
9418 * A framework for implementing offline system updates is now
9419 integrated, for details see:
9420 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
9422 * A new service type Type=idle is available now which helps us
9423 avoiding ugly interleaving of getty output and boot status
9426 * There's now a system-wide CapabilityBoundingSet= option to
9427 globally reduce the set of capabilities for the
9428 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
9429 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
9430 even CAP_NET_ADMIN system-wide for secure systems.
9432 * There are now system-wide DefaultLimitXXX= options to
9433 globally change the defaults of the various resource limits
9434 for all units started by PID 1.
9436 * Harald Hoyer's systemd test suite has been integrated into
9437 systemd which allows easy testing of systemd builds in qemu
9438 and nspawn. (This is really awesome! Ask us for details!)
9440 * The fstab parser is now implemented as generator, not inside
9443 * systemctl will now warn you if .mount units generated from
9444 /etc/fstab are out of date due to changes in fstab that
9445 have not been read by systemd yet.
9447 * systemd is now suitable for usage in initrds. Dracut has
9448 already been updated to make use of this. With this in place
9449 initrds get a slight bit faster but primarily are much
9450 easier to introspect and debug since "systemctl status" in
9451 the host system can be used to introspect initrd services,
9452 and the journal from the initrd is kept around too.
9454 * systemd-delta has been added, a tool to explore differences
9455 between user/admin configuration and vendor defaults.
9457 * PrivateTmp= now affects both /tmp and /var/tmp.
9459 * Boot time status messages are now much prettier and feature
9460 proper english language. Booting up systemd has never been
9463 * Read-ahead pack files now include the inode number of all
9464 files to pre-cache. When the inode changes the pre-caching
9465 is not attempted. This should be nicer to deal with updated
9466 packages which might result in changes of read-ahead
9469 * We now temporaritly lower the kernel's read_ahead_kb variable
9470 when collecting read-ahead data to ensure the kernel's
9471 built-in read-ahead does not add noise to our measurements
9472 of necessary blocks to pre-cache.
9474 * There's now RequiresMountsFor= to add automatic dependencies
9475 for all mounts necessary for a specific file system path.
9477 * MountAuto= and SwapAuto= have been removed from
9478 system.conf. Mounting file systems at boot has to take place
9481 * nspawn now learned a new switch --uuid= to set the machine
9482 ID on the command line.
9484 * nspawn now learned the -b switch to automatically search
9487 * vt102 is now the default TERM for serial TTYs, upgraded from
9490 * systemd-logind now works on VT-less systems.
9492 * The build tree has been reorganized. The individual
9493 components now have directories of their own.
9495 * A new condition type ConditionPathIsReadWrite= is now available.
9497 * nspawn learned the new -C switch to create cgroups for the
9498 container in other hierarchies.
9500 * We now have support for hardware watchdogs, configurable in
9503 * The scheduled shutdown logic now has a public API.
9505 * We now mount /tmp as tmpfs by default, but this can be
9506 masked and /etc/fstab can override it.
9508 * Since udisks does not make use of /media anymore we are not
9509 mounting a tmpfs on it anymore.
9511 * journalctl gained a new --local switch to only interleave
9512 locally generated journal files.
9514 * We can now load the IMA policy at boot automatically.
9516 * The GTK tools have been split off into a systemd-ui.
9518 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
9519 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
9520 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
9521 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
9522 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
9523 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
9524 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
9525 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
9526 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
9531 * This is mostly a bugfix release
9533 * Support optional initialization of the machine ID from the
9534 KVM or container configured UUID.
9536 * Support immediate reboots with "systemctl reboot -ff"
9538 * Show /etc/os-release data in systemd-analyze output
9540 * Many bugfixes for the journal, including endianness fixes and
9541 ensuring that disk space enforcement works
9543 * sd-login.h is C++ compatible again
9545 * Extend the /etc/os-release format on request of the Debian
9548 * We now refuse non-UTF8 strings used in various configuration
9549 and unit files. This is done to ensure we do not pass invalid
9550 data over D-Bus or expose it elsewhere.
9552 * Register Mimo USB Screens as suitable for automatic seat
9555 * Read SELinux client context from journal clients in a race
9558 * Reorder configuration file lookup order. /etc now always
9559 overrides /run in order to allow the administrator to always
9560 and unconditionally override vendor-supplied or
9561 automatically generated data.
9563 * The various user visible bits of the journal now have man
9564 pages. We still lack man pages for the journal API calls
9567 * We now ship all man pages in HTML format again in the
9570 Contributions from: Dave Reisner, Dirk Eibach, Frederic
9571 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
9572 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
9577 * This is mostly a bugfix release
9579 * systems lacking /etc/os-release are no longer supported.
9581 * Various functionality updates to libsystemd-login.so
9583 * Track class of PAM logins to distinguish greeters from
9586 Contributions from: Kay Sievers, Lennart Poettering, Michael
9591 * This is an important bugfix release for v41.
9593 * Building man pages is now optional which should be useful
9594 for those building systemd from git but unwilling to install
9597 * Watchdog support for supervising services is now usable. In
9598 a future release support for hardware watchdogs
9599 (i.e. /dev/watchdog) will be added building on this.
9601 * Service start rate limiting is now configurable and can be
9602 turned off per service. When a start rate limit is hit a
9603 reboot can automatically be triggered.
9605 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
9607 Contributions from: Benjamin Franzke, Bill Nottingham,
9608 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
9609 Schmidt, Michał Górny, Piotr Drąg
9613 * The systemd binary is installed /usr/lib/systemd/systemd now;
9614 An existing /sbin/init symlink needs to be adapted with the
9617 * The code that loads kernel modules has been ported to invoke
9618 libkmod directly, instead of modprobe. This means we do not
9619 support systems with module-init-tools anymore.
9621 * Watchdog support is now already useful, but still not
9624 * A new kernel command line option systemd.setenv= is
9625 understood to set system wide environment variables
9626 dynamically at boot.
9628 * We now limit the set of capabilities of systemd-journald.
9630 * We now set SIGPIPE to ignore by default, since it only is
9631 useful in shell pipelines, and has little use in general
9632 code. This can be disabled with IgnoreSIPIPE=no in unit
9635 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
9636 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
9641 * This is mostly a bugfix release
9643 * We now expose the reason why a service failed in the
9644 "Result" D-Bus property.
9646 * Rudimentary service watchdog support (will be completed over
9647 the next few releases.)
9649 * When systemd forks off in order execute some service we will
9650 now immediately changes its argv[0] to reflect which process
9651 it will execute. This is useful to minimize the time window
9652 with a generic argv[0], which makes bootcharts more useful
9654 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
9655 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
9656 Mike Kazantsev, Ray Strode
9660 * This is mostly a test release, but incorporates many
9663 * New systemd-cgtop tool to show control groups by their
9666 * Linking against libacl for ACLs is optional again. If
9667 disabled, support tracking device access for active logins
9668 goes becomes unavailable, and so does access to the user
9669 journals by the respective users.
9671 * If a group "adm" exists, journal files are automatically
9672 owned by them, thus allow members of this group full access
9673 to the system journal as well as all user journals.
9675 * The journal now stores the SELinux context of the logging
9676 client for all entries.
9678 * Add C++ inclusion guards to all public headers
9680 * New output mode "cat" in the journal to print only text
9681 messages, without any meta data like date or time.
9683 * Include tiny X server wrapper as a temporary stop-gap to
9684 teach XOrg udev display enumeration. This is used by display
9685 managers such as gdm, and will go away as soon as XOrg
9686 learned native udev hotplugging for display devices.
9688 * Add new systemd-cat tool for executing arbitrary programs
9689 with STDERR/STDOUT connected to the journal. Can also act as
9690 BSD logger replacement, and does so by default.
9692 * Optionally store all locally generated coredumps in the
9693 journal along with meta data.
9695 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
9696 writing short strings to files (for usage for /sys), and for
9697 creating symlinks, character and block device nodes.
9699 * New unit file option ControlGroupPersistent= to make cgroups
9700 persistent, following the mechanisms outlined in
9701 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
9703 * Support multiple local RTCs in a sane way
9705 * No longer monopolize IO when replaying readahead data on
9706 rotating disks, since we might starve non-file-system IO to
9707 death, since fanotify() will not see accesses done by blkid,
9710 * Do not show kernel threads in systemd-cgls anymore, unless
9711 requested with new -k switch.
9713 Contributions from: Dan Horák, Kay Sievers, Lennart
9714 Poettering, Michal Schmidt
9718 * This is mostly a test release, but incorporates many
9721 * The git repository moved to:
9722 git://anongit.freedesktop.org/systemd/systemd
9723 ssh://git.freedesktop.org/git/systemd/systemd
9725 * First release with the journal
9726 http://0pointer.de/blog/projects/the-journal.html
9728 * The journal replaces both systemd-kmsg-syslogd and
9729 systemd-stdout-bridge.
9731 * New sd_pid_get_unit() API call in libsystemd-logind
9733 * Many systemadm clean-ups
9735 * Introduce remote-fs-pre.target which is ordered before all
9736 remote mounts and may be used to start services before all
9739 * Added Mageia support
9741 * Add bash completion for systemd-loginctl
9743 * Actively monitor PID file creation for daemons which exit in
9744 the parent process before having finished writing the PID
9745 file in the daemon process. Daemons which do this need to be
9746 fixed (i.e. PID file creation must have finished before the
9747 parent exits), but we now react a bit more gracefully to them.
9749 * Add colourful boot output, mimicking the well-known output
9750 of existing distributions.
9752 * New option PassCredentials= for socket units, for
9753 compatibility with a recent kernel ABI breakage.
9755 * /etc/rc.local is now hooked in via a generator binary, and
9756 thus will no longer act as synchronization point during
9759 * systemctl list-unit-files now supports --root=.
9761 * systemd-tmpfiles now understands two new commands: z, Z for
9762 relabelling files according to the SELinux database. This is
9763 useful to apply SELinux labels to specific files in /sys,
9766 * Output of SysV services is now forwarded to both the console
9767 and the journal by default, not only just the console.
9769 * New man pages for all APIs from libsystemd-login.
9771 * The build tree got reorganized and the build system is a
9772 lot more modular allowing embedded setups to specifically
9773 select the components of systemd they are interested in.
9775 * Support for Linux systems lacking the kernel VT subsystem is
9778 * configure's --with-rootdir= got renamed to
9779 --with-rootprefix= to follow the naming used by udev and
9782 * Unless specified otherwise we will now install to /usr instead
9783 of /usr/local by default.
9785 * Processes with '@' in argv[0][0] are now excluded from the
9786 final shut-down killing spree, following the logic explained
9788 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
9790 * All processes remaining in a service cgroup when we enter
9791 the START or START_PRE states are now killed with
9792 SIGKILL. That means it is no longer possible to spawn
9793 background processes from ExecStart= lines (which was never
9794 supported anyway, and bad style).
9796 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
9797 reloading of units together.
9799 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
9800 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
9801 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
9802 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
9803 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek