1 systemd System and Service Manager
3 CHANGES WITH 246 in spe:
4 * The fs.suid_dumpable sysctl is set to 2 / "suidsafe". This allows
5 systemd-coredump to save core files for suid processes. When saving
6 the core file, systemd-coredump will use the effective uid and gid of
7 the process that faulted.
9 * "systemctl list-units" and "systemctl list-machines" no longer hide
10 their first output column with --no-legend. To hide the first column,
13 * The service manager gained basic support for cgroup v2 freezer. Units
14 can now be suspended or resumed either using new systemctl verbs,
15 freeze and thaw respectively, or via D-Bus.
19 * A new tool "systemd-repart" has been added, that operates as an
20 idempotent declarative repartitioner for GPT partition tables.
21 Specifically, a set of partitions that must or may exist can be
22 configured via drop-in files, and during every boot the partition
23 table on disk is compared with these files, creating missing
24 partitions or growing existing ones based on configurable relative
25 and absolute size constraints. The tool is strictly incremental,
26 i.e. does not delete, shrink or move partitions, but only adds and
27 grows them. The primary use-case is OS images that ship in minimized
28 form, that on first boot are grown to the size of the underlying
29 block device or augmented with additional partitions. For example,
30 the root partition could be extended to cover the whole disk, or a
31 swap or /home partitions could be added on first boot. It can also be
32 used for systems that use an A/B update scheme but ship images with
33 just the A partition, with B added on first boot. The tool is
34 primarily intended to be run in the initrd, shortly before
35 transitioning into the host OS, but can also be run after the
36 transition took place. It automatically discovers the disk backing
37 the root file system, and should hence not require any additional
38 configuration besides the partition definition drop-ins. If no
39 configuration drop-ins are present, no action is taken.
41 * A new component "userdb" has been added, along with a small daemon
42 "systemd-userdb.service" and a client tool "userdbctl". The framework
43 allows defining rich user and group records in a JSON format,
44 extending on the classic "struct passwd" and "struct group"
45 structures. Various components in systemd have been updated to
46 process records in this format, including systemd-logind and
47 pam-systemd. The user records are intended to be extensible, and
48 allow setting various resource management, security and runtime
49 parameters that shall be applied to processes and sessions of the
50 user as they log in. This facility is intended to allow associating
51 such metadata directly with user/group records so that they can be
52 produced, extended and consumed in unified form. We hope that
53 eventually frameworks such as sssd will generate records this way, so
54 that for the first time resource management and various other
55 per-user settings can be configured in LDAP directories and then
56 provided to systemd (specifically to systemd-logind and pam-system)
57 to apply on login. For further details see:
59 https://systemd.io/USER_RECORD
60 https://systemd.io/GROUP_RECORD
61 https://systemd.io/USER_GROUP_API
63 * A small new service systemd-homed.service has been added, that may be
64 used to securely manage home directories with built-in encryption.
65 The complete user record data is unified with the home directory,
66 thus making home directories naturally migratable. Its primary
67 back-end is based on LUKS volumes, but fscrypt, plain directories,
68 and other storage schemes are also supported. This solves a couple of
69 problems we saw with traditional ways to manage home directories, in
70 particular when it comes to encryption. For further discussion of
71 this, see the video of Lennart's talk at AllSystemsGo! 2019:
73 https://media.ccc.de/v/ASG2019-164-reinventing-home-directories
75 For further details about the format and expectations on home
76 directories this new daemon makes, see:
78 https://systemd.io/HOME_DIRECTORY
80 * systemd-journald is now multi-instantiable. In addition to the main
81 instance systemd-journald.service there's now a template unit
82 systemd-journald@.service, with each instance defining a new named
83 log 'namespace' (whose name is specified via the instance part of the
84 unit name). A new unit file setting LogNamespace= has been added,
85 taking such a namespace name, that assigns services to the specified
86 log namespaces. As each log namespace is serviced by its own
87 independent journal daemon, this functionality may be used to improve
88 performance and increase isolation of applications, at the price of
89 losing global message ordering. Each instance of journald has a
90 separate set of configuration files, with possibly different disk
91 usage limitations and other settings.
93 journalctl now takes a new option --namespace= to show logs from a
94 specific log namespace. The sd-journal.h API gained
95 sd_journal_open_namespace() for opening the log stream of a specific
96 log namespace. systemd-journald also gained the ability to exit on
97 idle, which is useful in the context of log namespaces, as this means
98 log daemons for log namespaces can be activated automatically on
99 demand and will stop automatically when no longer used, minimizing
102 * When systemd-tmpfiles copies a file tree using the 'C' line type it
103 will now label every copied file according to the SELinux database.
105 * When systemd/PID 1 detects it is used in the initrd it will now boot
106 into initrd.target rather than default.target by default. This should
107 make it simpler to build initrds with systemd as for many cases the
108 only difference between a host OS image and an initrd image now is
109 the presence of the /etc/initrd-release file.
111 * A new kernel command line option systemd.cpu_affinity= is now
112 understood. It's equivalent to the CPUAffinity= option in
113 /etc/systemd/system.conf and allows setting the CPU mask for PID 1
114 itself and the default for all other processes.
116 * When systemd/PID 1 is reloaded (with systemctl daemon-reload or
117 equivalent), the SELinux database is now reloaded, ensuring that
118 sockets and other file system objects are generated taking the new
119 database into account.
121 * systemd/PID 1 accepts a new "systemd.show-status=error" setting, and
122 "quiet" has been changed to imply that instead of
123 "systemd.show-status=auto". In this mode, only messages about errors
124 and significant delays in boot are shown on the console.
126 * The sd-event.h API gained native support for the new Linux "pidfd"
127 concept. This permits watching processes using file descriptors
128 instead of PID numbers, which fixes a number of races and makes
129 process supervision more robust and efficient. All of systemd's
130 components will now use pidfds if the kernel supports it for process
131 watching, with the exception of PID 1 itself, unfortunately. We hope
132 to move PID 1 to exclusively using pidfds too eventually, but this
133 requires some more kernel work first. (Background: PID 1 watches
134 processes using waitid() with the P_ALL flag, and that does not play
135 together nicely with pidfds yet.)
137 * Closely related to this, the sd-event.h API gained two new calls
138 sd_event_source_send_child_signal() (for sending a signal to a
139 watched process) and sd_event_source_get_child_process_own() (for
140 marking a process so that it is killed automatically whenever the
141 event source watching it is freed).
143 * systemd-networkd gained support for configuring Token Bucket Filter
144 (TBF) parameters in its qdisc configuration support. Similarly,
145 support for Stochastic Fairness Queuing (SFQ), Controlled-Delay
146 Active Queue Management (CoDel), and Fair Queue (FQ) has been added.
148 * systemd-networkd gained support for Intermediate Functional Block
149 (IFB) network devices.
151 * systemd-networkd gained support for configuring multi-path IP routes,
152 using the new MultiPathRoute= setting in the [Route] section.
154 * systemd-networkd's DHCPv4 client has been updated to support a new
155 SendDecline= option. If enabled, duplicate address detection is done
156 after a DHCP offer is received from the server. If a conflict is
157 detected, the address is declined. The DHCPv4 client also gained
158 support for a new RouteMTUBytes= setting that allows to configure the
159 MTU size to be used for routes generated from DHCPv4 leases.
161 * The PrefixRoute= setting in systemd-networkd's [Address] section of
162 .network files has been deprecated, and replaced by AddPrefixRoute=,
163 with its sense inverted.
165 * The Gateway= setting of [Route] sections of .network files gained
166 support for a special new value "_dhcp". If set, the configured
167 static route uses the gateway host configured via DHCP.
169 * New User= and SuppressPrefixLength= settings have been implemented
170 for the [RoutingPolicyRule] section of .network files to configure
171 source routing based on UID ranges and prefix length, respectively.
173 * sd-bus gained a new API call sd_bus_message_sensitive() that marks a
174 D-Bus message object as "sensitive". Those objects are erased from
175 memory when they are freed. This concept is intended to be used for
176 messages that contain security sensitive data. A new flag
177 SD_BUS_VTABLE_SENSITIVE has been introduced as well to mark methods
178 in sd-bus vtables, causing any incoming and outgoing messages of
179 those methods to be implicitly marked as "sensitive".
181 * sd-bus gained a new API call sd_bus_message_dump() for dumping the
182 contents of a message (or parts thereof) to standard output for
185 * systemd-sysusers gained support for creating users with the primary
186 group named differently than the user.
188 * systemd-resolved's DNS-over-TLS support gained SNI validation.
190 * systemd-growfs (i.e. the x-systemd.growfs mount option in /etc/fstab)
191 gained support for growing XFS partitions. Previously it supported
192 only ext4 and btrfs partitions.
194 * The support for /etc/crypttab gained a new x-initrd.attach option. If
195 set, the specified encrypted volume is unlocked already in the
196 initrd. This concept corresponds to the x-initrd.mount option in
199 * systemd-cryptsetup gained native support for unlocking encrypted
200 volumes utilizing PKCS#11 smartcards, i.e. for example to bind
201 encryption of volumes to YubiKeys. This is exposed in the new
202 pkcs11-uri= option in /etc/crypttab.
204 * The /etc/fstab support in systemd now supports two new mount options
205 x-systemd.{required,wanted}-by=, for explicitly configuring the units
206 that the specified mount shall be pulled in by, in place of
207 the usual local-fs.target/remote-fs.target.
209 * The https://systemd.io/ web site has been relaunched, directly
210 populated with most of the documentation included in the systemd
211 repository. systemd also acquired a new logo, thanks to Tobias
214 * systemd-udevd gained support for managing "alternative" network
215 interface names, as supported by new Linux kernels. For the first
216 time this permits assigning multiple (and longer!) names to a network
217 interface. systemd-udevd will now by default assign the names
218 generated via all supported naming schemes to each interface. This
219 may be further tweaked with .link files and the AlternativeName= and
220 AlternativeNamesPolicy= settings. Other components of systemd have
221 been updated to support the new alternative names wherever
222 appropriate. For example, systemd-nspawn will now generate
223 alternative interface names for the host-facing side of container
224 veth links based on the full container name without truncation.
226 * systemd-nspawn interface naming logic has been updated in another way
227 too: if the main interface name (i.e. as opposed to new-style
228 "alternative" names) based on the container name is truncated, a
229 simple hashing scheme is used to give different interface names to
230 multiple containers whose names all begin with the same prefix. Since
231 this changes the primary interface names pointing to containers if
232 truncation happens, the old scheme may still be requested by
233 selecting an older naming scheme, via the net.naming-scheme= kernel
236 * PrivateUsers= in service files now works in services run by the
237 systemd --user per-user instance of the service manager.
239 * A new per-service sandboxing option ProtectClock= has been added that
240 locks down write access to the system clock. It takes away device
241 node access to /dev/rtc as well as the system calls that set the
242 system clock and the CAP_SYS_TIME and CAP_WAKE_ALARM capabilities.
243 Note that this option does not affect access to auxiliary services
244 that allow changing the clock, for example access to
247 * The systemd-id128 tool gained a new "show" verb for listing or
248 resolving a number of well-known UUIDs/128bit IDs, currently mostly
249 GPT partition table types.
251 * The Discoverable Partitions Specification has been updated to support
252 /var and /var/tmp partition discovery. Support for this has been
253 added to systemd-gpt-auto-generator. For details see:
255 https://systemd.io/DISCOVERABLE_PARTITIONS
257 * "systemctl list-unit-files" has been updated to show a new column
258 with the suggested enablement state based on the vendor preset files
259 for the respective units.
261 * "systemctl" gained a new option "--with-dependencies". If specified
262 commands such as "systemctl status" or "systemctl cat" will now show
263 all specified units along with all units they depend on.
265 * networkctl gained support for showing per-interface logs in its
268 * systemd-networkd-wait-online gained support for specifying the maximum
269 operational state to wait for, and to wait for interfaces to
272 * The [Match] section of .link and .network files now supports a new
273 option PermanentMACAddress= which may be used to check against the
274 permanent MAC address of a network device even if a randomized MAC
277 * The [TrafficControlQueueingDiscipline] section in .network files has
278 been renamed to [NetworkEmulator] with the "NetworkEmulator" prefix
279 dropped from the individual setting names.
281 * Any .link and .network files that have an empty [Match] section (this
282 also includes empty and commented-out files) will now be
283 rejected. systemd-udev and systemd-networkd started warning about
284 such files in version 243.
286 * systemd-logind will now validate access to the operation of changing
287 the virtual terminal via a polkit action. By default, only users
288 with at least one session on a local VT are granted permission.
290 * When systemd sets up PAM sessions that invoked service processes
291 shall run in, the pam_setcred() API is now invoked, thus permitting
292 PAM modules to set additional credentials for the processes.
294 * portablectl attach/detach verbs now accept --now and --enable options
295 to combine attachment with enablement and invocation, or detachment
296 with stopping and disablement.
298 * UPGRADE ISSUE: a bug where some jobs were trimmed as redundant was
299 fixed, which in turn exposed bugs in unit configuration of services
300 which have Type=oneshot and should only run once, but do not have
301 RemainAfterExit=yes set. Without RemainAfterExit=yes, a one-shot
302 service may be started again after exiting successfully, for example
303 as a dependency in another transaction. Affected services included
304 some internal systemd services (most notably
305 systemd-vconsole-setup.service, which was updated to have
306 RemainAfterExit=yes), and plymouth-start.service. Please ensure that
307 plymouth has been suitably updated or patched before upgrading to
308 this systemd release. See
309 https://bugzilla.redhat.com/show_bug.cgi?id=1807771 for some
310 additional discussion.
312 Contributions from: AJ Bagwell, Alin Popa, Andreas Rammhold, Anita
313 Zhang, Ansgar Burchardt, Antonio Russo, Arian van Putten, Ashley Davis,
314 Balint Reczey, Bart Willems, Bastien Nocera, Benjamin Dahlhoff, Charles
315 (Chas) Williams, cheese1, Chris Down, Chris Murphy, Christian Ehrhardt,
316 Christian Göttsche, cvoinf, Daan De Meyer, Daniele Medri, Daniel Rusek,
317 Daniel Shahaf, Dann Frazier, Dan Streetman, Dariusz Gadomski, David
318 Michael, Dimitri John Ledkov, Emmanuel Bourg, Evgeny Vereshchagin,
319 ezst036, Felipe Sateler, Filipe Brandenburger, Florian Klink, Franck
320 Bui, Fran Dieguez, Frantisek Sumsal, Greg "GothAck" Miell, Guilhem
321 Lettron, Guillaume Douézan-Grard, Hans de Goede, HATAYAMA Daisuke, Iain
322 Lane, James Buren, Jan Alexander Steffens (heftig), Jérémy Rosen, Jin
323 Park, Jun'ichi Nomura, Kai Krakow, Kevin Kuehler, Kevin P. Fleming,
324 Lennart Poettering, Leonid Bloch, Leonid Evdokimov, lothrond, Luca
325 Boccassi, Lukas K, Lynn Kirby, Mario Limonciello, Mark Deneen, Matthew
326 Leeds, Michael Biebl, Michal Koutný, Michal Sekletár, Mike Auty, Mike
327 Gilbert, mtron, nabijaczleweli, Naïm Favier, Nate Jones, Norbert Lange,
328 Oliver Giles, Paul Davey, Paul Menzel, Peter Hutterer, Piotr Drąg, Rafa
329 Couto, Raphael, rhn, Robert Scheck, Rocka, Romain Naour, Ryan Attard,
330 Sascha Dewald, Shengjing Zhu, Slava Kardakov, Spencer Michaels, Sylvain
331 Plantefeve, Stanislav Angelovič, Susant Sahani, Thomas Haller, Thomas
332 Schmitt, Timo Schlüßler, Timo Wilken, Tobias Bernard, Tobias Klauser,
333 Tobias Stoeckmann, Topi Miettinen, tsia, WataruMatsuoka, Wieland
334 Hoffmann, Wilhelm Schuster, Will Fleming, xduugu, Yong Cong Sin, Yuri
335 Chornoivan, Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek, Zeyu
342 * Support for the cpuset cgroups v2 controller has been added.
343 Processes may be restricted to specific CPUs using the new
344 AllowedCPUs= setting, and to specific memory NUMA nodes using the new
345 AllowedMemoryNodes= setting.
347 * The signal used in restart jobs (as opposed to e.g. stop jobs) may
348 now be configured using a new RestartKillSignal= setting. This
349 allows units which signals to request termination to implement
350 different behaviour when stopping in preparation for a restart.
352 * "systemctl clean" may now be used also for socket, mount, and swap
355 * systemd will also read configuration options from the EFI variable
356 SystemdOptions. This may be used to configure systemd behaviour when
357 modifying the kernel command line is inconvenient, but configuration
358 on disk is read too late, for example for the options related to
359 cgroup hierarchy setup. 'bootctl systemd-efi-options' may be used to
360 set the EFI variable.
362 * systemd will now disable printk ratelimits in early boot. This should
363 allow us to capture more logs from the early boot phase where normal
364 storage is not available and the kernel ring buffer is used for
365 logging. Configuration on the kernel command line has higher priority
366 and overrides the systemd setting.
368 systemd programs which log to /dev/kmsg directly use internal
369 ratelimits to prevent runaway logging. (Normally this is only used
370 during early boot, so in practice this change has very little
373 * Unit files now support top level dropin directories of the form
374 <unit_type>.d/ (e.g. service.d/) that may be used to add configuration
375 that affects all corresponding unit files.
377 * systemctl gained support for 'stop --job-mode=triggering' which will
378 stop the specified unit and any units which could trigger it.
380 * Unit status display now includes units triggering and triggered by
381 the unit being shown.
383 * The RuntimeMaxSec= setting is now supported by scopes, not just
384 .service units. This is particularly useful for PAM sessions which
385 create a scope unit for the user login. systemd.runtime_max_sec=
386 setting may used with the pam_systemd module to limit the duration
387 of the PAM session, for example for time-limited logins.
389 * A new @pkey system call group is now defined to make it easier to
390 whitelist memory protection syscalls for containers and services
391 which need to use them.
393 * systemd-udevd: removed the 30s timeout for killing stale workers on
394 exit. systemd-udevd now waits for workers to finish. The hard-coded
395 exit timeout of 30s was too short for some large installations, where
396 driver initialization could be prematurely interrupted during initrd
397 processing if the root file system had been mounted and init was
398 preparing to switch root. If udevd is run without systemd and workers
399 are hanging while udevd receives an exit signal, udevd will now exit
400 when udev.event_timeout is reached for the last hanging worker. With
401 systemd, the exit timeout can additionally be configured using
402 TimeoutStopSec= in systemd-udevd.service.
404 * udev now provides a program (fido_id) that identifies FIDO CTAP1
405 ("U2F")/CTAP2 security tokens based on the usage declared in their
406 report and descriptor and outputs suitable environment variables.
407 This replaces the externally maintained whitelists of all known
408 security tokens that were used previously.
410 * Automatically generated autosuspend udev rules for whitelisted
411 devices have been imported from the Chromium OS project. This should
412 improve power saving with many more devices.
414 * udev gained a new "CONST{key}=value" setting that allows matching
415 against system-wide constants without forking a helper binary.
416 Currently "arch" and "virt" keys are supported.
418 * udev now opens CDROMs in non-exclusive mode when querying their
419 capabilities. This should fix issues where other programs trying to
420 use the CDROM cannot gain access to it, but carries a risk of
421 interfering with programs writing to the disk, if they did not open
422 the device in exclusive mode as they should.
424 * systemd-networkd does not create a default route for IPv4 link local
425 addressing anymore. The creation of the route was unexpected and was
426 breaking routing in various cases, but people who rely on it being
427 created implicitly will need to adjust. Such a route may be requested
428 with DefaultRouteOnDevice=yes.
430 Similarly, systemd-networkd will not assign a link-local IPv6 address
431 when IPv6 link-local routing is not enabled.
433 * Receive and transmit buffers may now be configured on links with
434 the new RxBufferSize= and TxBufferSize= settings.
436 * systemd-networkd may now advertise additional IPv6 routes. A new
437 [IPv6RoutePrefix] section with Route= and LifetimeSec= options is
440 * systemd-networkd may now configure "next hop" routes using the
441 [NextHop] section and Gateway= and Id= settings.
443 * systemd-networkd will now retain DHCP config on restarts by default
444 (but this may be overridden using the KeepConfiguration= setting).
445 The default for SendRelease= has been changed to true.
447 * The DHCPv4 client now uses the OPTION_INFORMATION_REFRESH_TIME option
448 received from the server.
450 The client will use the received SIP server list if UseSIP=yes is
453 The client may be configured to request specific options from the
454 server using a new RequestOptions= setting.
456 The client may be configured to send arbitrary options to the server
457 using a new SendOption= setting.
459 A new IPServiceType= setting has been added to configure the "IP
460 service type" value used by the client.
462 * The DHCPv6 client learnt a new PrefixDelegationHint= option to
463 request prefix hints in the DHCPv6 solicitation.
465 * The DHCPv4 server may be configured to send arbitrary options using
466 a new SendOption= setting.
468 * The DHCPv4 server may now be configured to emit SIP server list using
469 the new EmitSIP= and SIP= settings.
471 * systemd-networkd and networkctl may now renew DHCP leases on demand.
472 networkctl has a new 'networkctl renew' verb.
474 * systemd-networkd may now reconfigure links on demand. networkctl
475 gained two new verbs: "reload" will reload the configuration, and
476 "reconfigure DEVICE…" will reconfigure one or more devices.
478 * .network files may now match on SSID and BSSID of a wireless network,
479 i.e. the access point name and hardware address using the new SSID=
480 and BSSID= options. networkctl will display the current SSID and
481 BSSID for wireless links.
483 .network files may also match on the wireless network type using the
484 new WLANInterfaceType= option.
486 * systemd-networkd now includes default configuration that enables
487 link-local addressing when connected to an ad-hoc wireless network.
489 * systemd-networkd may configure the Traffic Control queueing
490 disciplines in the kernel using the new
491 [TrafficControlQueueingDiscipline] section and Parent=,
492 NetworkEmulatorDelaySec=, NetworkEmulatorDelayJitterSec=,
493 NetworkEmulatorPacketLimit=, NetworkEmulatorLossRate=,
494 NetworkEmulatorDuplicateRate= settings.
496 * systemd-tmpfiles gained a new w+ setting to append to files.
498 * systemd-analyze dump will now report when the memory configuration in
499 the kernel does not match what systemd has configured (usually,
500 because some external program has modified the kernel configuration
503 * systemd-analyze gained a new --base-time= switch instructs the
504 'calendar' verb to resolve times relative to that timestamp instead
507 * journalctl --update-catalog now produces deterministic output (making
508 reproducible image builds easier).
510 * A new devicetree-overlay setting is now documented in the Boot Loader
513 * The default value of the WatchdogSec= setting used in systemd
514 services (the ones bundled with the project itself) may be set at
515 configuration time using the -Dservice-watchdog= setting. If set to
516 empty, the watchdogs will be disabled.
518 * systemd-resolved validates IP addresses in certificates now when GnuTLS
521 * libcryptsetup >= 2.0.1 is now required.
523 * A configuration option -Duser-path= may be used to override the $PATH
524 used by the user service manager. The default is again to use the same
525 path as the system manager.
527 * The systemd-id128 tool gained a new switch "-u" (or "--uuid") for
528 outputting the 128bit IDs in UUID format (i.e. in the "canonical
531 * Service units gained a new sandboxing option ProtectKernelLogs= which
532 makes sure the program cannot get direct access to the kernel log
533 buffer anymore, i.e. the syslog() system call (not to be confused
534 with the API of the same name in libc, which is not affected), the
535 /proc/kmsg and /dev/kmsg nodes and the CAP_SYSLOG capability are made
536 inaccessible to the service. It's recommended to enable this setting
537 for all services that should not be able to read from or write to the
538 kernel log buffer, which are probably almost all.
540 Contributions from: Aaron Plattner, Alcaro, Anita Zhang, Balint Reczey,
541 Bastien Nocera, Baybal Ni, Benjamin Bouvier, Benjamin Gilbert, Carlo
542 Teubner, cbzxt, Chen Qi, Chris Down, Christian Rebischke, Claudio
543 Zumbo, ClydeByrdIII, crashfistfight, Cyprien Laplace, Daniel Edgecumbe,
544 Daniel Gorbea, Daniel Rusek, Daniel Stuart, Dan Streetman, David
545 Pedersen, David Tardon, Dimitri John Ledkov, Dominique Martinet, Donald
546 A. Cupp Jr, Evgeny Vereshchagin, Fabian Henneke, Filipe Brandenburger,
547 Franck Bui, Frantisek Sumsal, Georg Müller, Hans de Goede, Haochen
548 Tong, HATAYAMA Daisuke, Iwan Timmer, Jan Janssen, Jan Kundrát, Jan
549 Synacek, Jan Tojnar, Jay Strict, Jérémy Rosen, Jóhann B. Guðmundsson,
550 Jonas Jelten, Jonas Thelemann, Justin Trudell, J. Xing, Kai-Heng Feng,
551 Kenneth D'souza, Kevin Becker, Kevin Kuehler, Lennart Poettering,
552 Léonard Gérard, Lorenz Bauer, Luca Boccassi, Maciej Stanczew, Mario
553 Limonciello, Marko Myllynen, Mark Stosberg, Martin Wilck, matthiasroos,
554 Michael Biebl, Michael Olbrich, Michael Tretter, Michal Sekletar,
555 Michal Sekletár, Michal Suchanek, Mike Gilbert, Mike Kazantsev, Nicolas
556 Douma, nikolas, Norbert Lange, pan93412, Pascal de Bruijn, Paul Menzel,
557 Pavel Hrdina, Peter Wu, Philip Withnall, Piotr Drąg, Rafael Fontenelle,
558 Renaud Métrich, Riccardo Schirone, RoadrunnerWMC, Ronan Pigott, Ryan
559 Attard, Sebastian Wick, Serge, Siddharth Chandrasekara, Steve Ramage,
560 Steve Traylen, Susant Sahani, Thibault Nélis, Tim Teichmann, Tom
561 Fitzhenry, Tommy J, Torsten Hilbrich, Vito Caputo, ypf791, Yu Watanabe,
562 Zach Smith, Zbigniew Jędrzejewski-Szmek
568 * This release enables unprivileged programs (i.e. requiring neither
569 setuid nor file capabilities) to send ICMP Echo (i.e. ping) requests
570 by turning on the "net.ipv4.ping_group_range" sysctl of the Linux
571 kernel for the whole UNIX group range, i.e. all processes. This
572 change should be reasonably safe, as the kernel support for it was
573 specifically implemented to allow safe access to ICMP Echo for
574 processes lacking any privileges. If this is not desirable, it can be
575 disabled again by setting the parameter to "1 0".
577 * Previously, filters defined with SystemCallFilter= would have the
578 effect that any calling of an offending system call would terminate
579 the calling thread. This behaviour never made much sense, since
580 killing individual threads of unsuspecting processes is likely to
581 create more problems than it solves. With this release the default
582 action changed from killing the thread to killing the whole
583 process. For this to work correctly both a kernel version (>= 4.14)
584 and a libseccomp version (>= 2.4.0) supporting this new seccomp
585 action is required. If an older kernel or libseccomp is used the old
586 behaviour continues to be used. This change does not affect any
587 services that have no system call filters defined, or that use
588 SystemCallErrorNumber= (and thus see EPERM or another error instead
589 of being killed when calling an offending system call). Note that
590 systemd documentation always claimed that the whole process is
591 killed. With this change behaviour is thus adjusted to match the
594 * On 64 bit systems, the "kernel.pid_max" sysctl is now bumped to
595 4194304 by default, i.e. the full 22bit range the kernel allows, up
596 from the old 16bit range. This should improve security and
597 robustness, as PID collisions are made less likely (though certainly
598 still possible). There are rumours this might create compatibility
599 problems, though at this moment no practical ones are known to
600 us. Downstream distributions are hence advised to undo this change in
601 their builds if they are concerned about maximum compatibility, but
602 for everybody else we recommend leaving the value bumped. Besides
603 improving security and robustness this should also simplify things as
604 the maximum number of allowed concurrent tasks was previously bounded
605 by both "kernel.pid_max" and "kernel.threads-max" and now effectively
606 only a single knob is left ("kernel.threads-max"). There have been
607 concerns that usability is affected by this change because larger PID
608 numbers are harder to type, but we believe the change from 5 digits
609 to 7 digits doesn't hamper usability.
611 * MemoryLow= and MemoryMin= gained hierarchy-aware counterparts,
612 DefaultMemoryLow= and DefaultMemoryMin=, which can be used to
613 hierarchically set default memory protection values for a particular
614 subtree of the unit hierarchy.
616 * Memory protection directives can now take a value of zero, allowing
617 explicit opting out of a default value propagated by an ancestor.
619 * systemd now defaults to the "unified" cgroup hierarchy setup during
620 build-time, i.e. -Ddefault-hierarchy=unified is now the build-time
621 default. Previously, -Ddefault-hierarchy=hybrid was the default. This
622 change reflects the fact that cgroupsv2 support has matured
623 substantially in both systemd and in the kernel, and is clearly the
624 way forward. Downstream production distributions might want to
625 continue to use -Ddefault-hierarchy=hybrid (or even =legacy) for
626 their builds as unfortunately the popular container managers have not
627 caught up with the kernel API changes.
629 * Man pages are not built by default anymore (html pages were already
630 disabled by default), to make development builds quicker. When
631 building systemd for a full installation with documentation, meson
632 should be called with -Dman=true and/or -Dhtml=true as appropriate.
633 The default was changed based on the assumption that quick one-off or
634 repeated development builds are much more common than full optimized
635 builds for installation, and people need to pass various other
636 options to when doing "proper" builds anyway, so the gain from making
637 development builds quicker is bigger than the one time disruption for
640 Two scripts are created in the *build* directory to generate and
641 preview man and html pages on demand, e.g.:
643 build/man/man systemctl
644 build/man/html systemd.index
646 * libidn2 is used by default if both libidn2 and libidn are installed.
647 Please use -Dlibidn=true if libidn is preferred.
649 * The D-Bus "wire format" of the CPUAffinity= attribute is changed on
650 big-endian machines. Before, bytes were written and read in native
651 machine order as exposed by the native libc __cpu_mask interface.
652 Now, little-endian order is always used (CPUs 0–7 are described by
653 bits 0–7 in byte 0, CPUs 8–15 are described by byte 1, and so on).
654 This change fixes D-Bus calls that cross endianness boundary.
656 The presentation format used for CPUAffinity= by "systemctl show" and
657 "systemd-analyze dump" is changed to present CPU indices instead of
658 the raw __cpu_mask bitmask. For example, CPUAffinity=0-1 would be
659 shown as CPUAffinity=03000000000000000000000000000… (on
660 little-endian) or CPUAffinity=00000000000000300000000000000… (on
661 64-bit big-endian), and is now shown as CPUAffinity=0-1, matching the
662 input format. The maximum integer that will be printed in the new
663 format is 8191 (four digits), while the old format always used a very
664 long number (with the length varying by architecture), so they can be
665 unambiguously distinguished.
667 * /usr/sbin/halt.local is no longer supported. Implementation in
668 distributions was inconsistent and it seems this functionality was
671 To replace this functionality, users should:
672 - either define a new unit and make it a dependency of final.target
673 (systemctl add-wants final.target my-halt-local.service)
674 - or move the shutdown script to /usr/lib/systemd/system-shutdown/
675 and ensure that it accepts "halt", "poweroff", "reboot", and
676 "kexec" as an argument, see the description in systemd-shutdown(8).
678 * When a [Match] section in .link or .network file is empty (contains
679 no match patterns), a warning will be emitted. Please add any "match
680 all" pattern instead, e.g. OriginalName=* or Name=* in case all
681 interfaces should really be matched.
683 * A new setting NUMAPolicy= may be used to set process memory
684 allocation policy. This setting can be specified in
685 /etc/systemd/system.conf and hence will set the default policy for
686 PID1. The default policy can be overridden on a per-service
687 basis. The related setting NUMAMask= is used to specify NUMA node
688 mask that should be associated with the selected policy.
690 * PID 1 will now listen to Out-Of-Memory (OOM) events the kernel
691 generates when processes it manages are reaching their memory limits,
692 and will place their units in a special state, and optionally kill or
695 * The service manager will now expose bus properties for the IO
696 resources used by units. This information is also shown in "systemctl
697 status" now (for services that have IOAccounting=yes set). Moreover,
698 the IO accounting data is included in the resource log message
699 generated whenever a unit stops.
701 * Units may now configure an explicit time-out to wait for when killed
702 with SIGABRT, for example when a service watchdog is hit. Previously,
703 the regular TimeoutStopSec= time-out was applied in this case too —
704 now a separate time-out may be set using TimeoutAbortSec=.
706 * Services may now send a special WATCHDOG=trigger message with
707 sd_notify() to trigger an immediate "watchdog missed" event, and thus
708 trigger service termination. This is useful both for testing watchdog
709 handling, but also for defining error paths in services, that shall
710 be handled the same way as watchdog events.
712 * There are two new per-unit settings IPIngressFilterPath= and
713 IPEgressFilterPath= which allow configuration of a BPF program
714 (usually by specifying a path to a program uploaded to /sys/fs/bpf/)
715 to apply to the IP packet ingress/egress path of all processes of a
716 unit. This is useful to allow running systemd services with BPF
717 programs set up externally.
719 * systemctl gained a new "clean" verb for removing the state, cache,
720 runtime or logs directories of a service while it is terminated. The
721 new verb may also be used to remove the state maintained on disk for
722 timer units that have Persistent= configured.
724 * During the last phase of shutdown systemd will now automatically
725 increase the log level configured in the "kernel.printk" sysctl so
726 that any relevant loggable events happening during late shutdown are
727 made visible. Previously, loggable events happening so late during
728 shutdown were generally lost if the "kernel.printk" sysctl was set to
729 high thresholds, as regular logging daemons are terminated at that
730 time and thus nothing is written to disk.
732 * If processes terminated during the last phase of shutdown do not exit
733 quickly systemd will now show their names after a short time, to make
734 debugging easier. After a longer time-out they are forcibly killed,
737 * journalctl (and the other tools that display logs) will now highlight
738 warnings in yellow (previously, both LOG_NOTICE and LOG_WARNING where
739 shown in bright bold, now only LOG_NOTICE is). Moreover, audit logs
740 are now shown in blue color, to separate them visually from regular
741 logs. References to configuration files are now turned into clickable
742 links on terminals that support that.
744 * systemd-journald will now stop logging to /var/log/journal during
745 shutdown when /var/ is on a separate mount, so that it can be
746 unmounted safely during shutdown.
748 * systemd-resolved gained support for a new 'strict' DNS-over-TLS mode.
750 * systemd-resolved "Cache=" configuration option in resolved.conf has
751 been extended to also accept the 'no-negative' value. Previously,
752 only a boolean option was allowed (yes/no), having yes as the
753 default. If this option is set to 'no-negative', negative answers are
754 not cached while the old cache heuristics are used positive answers.
755 The default remains unchanged.
757 * The predictable naming scheme for network devices now supports
758 generating predictable names for "netdevsim" devices.
760 Moreover, the "en" prefix was dropped from the ID_NET_NAME_ONBOARD
763 Those two changes form a new net.naming-policy-scheme= entry.
764 Distributions which want to preserve naming stability may want to set
765 the -Ddefault-net-naming-scheme= configuration option.
767 * systemd-networkd now supports MACsec, nlmon, IPVTAP and Xfrm
770 * systemd-networkd's bridge FDB support now allows configuration of a
771 destination address for each entry (Destination=), as well as the
772 VXLAN VNI (VNI=), as well as an option to declare what an entry is
773 associated with (AssociatedWith=).
775 * systemd-networkd's DHCPv4 support now understands a new MaxAttempts=
776 option for configuring the maximum number of DHCP lease requests. It
777 also learnt a new BlackList= option for blacklisting DHCP servers (a
778 similar setting has also been added to the IPv6 RA client), as well
779 as a SendRelease= option for configuring whether to send a DHCP
780 RELEASE message when terminating.
782 * systemd-networkd's DHCPv4 and DHCPv6 stacks can now be configured
783 separately in the [DHCPv4] and [DHCPv6] sections.
785 * systemd-networkd's DHCP support will now optionally create an
786 implicit host route to the DNS server specified in the DHCP lease, in
787 addition to the routes listed explicitly in the lease. This should
788 ensure that in multi-homed systems DNS traffic leaves the systems on
789 the interface that acquired the DNS server information even if other
790 routes such as default routes exist. This behaviour may be turned on
791 with the new RoutesToDNS= option.
793 * systemd-networkd's VXLAN support gained a new option
794 GenericProtocolExtension= for enabling VXLAN Generic Protocol
795 Extension support, as well as IPDoNotFragment= for setting the IP
796 "Don't fragment" bit on outgoing packets. A similar option has been
797 added to the GENEVE support.
799 * In systemd-networkd's [Route] section you may now configure
800 FastOpenNoCookie= for configuring per-route TCP fast-open support, as
801 well as TTLPropagate= for configuring Label Switched Path (LSP) TTL
802 propagation. The Type= setting now supports local, broadcast,
803 anycast, multicast, any, xresolve routes, too.
805 * systemd-networkd's [Network] section learnt a new option
806 DefaultRouteOnDevice= for automatically configuring a default route
807 onto the network device.
809 * systemd-networkd's bridging support gained two new options ProxyARP=
810 and ProxyARPWifi= for configuring proxy ARP behaviour as well as
811 MulticastRouter= for configuring multicast routing behaviour. A new
812 option MulticastIGMPVersion= may be used to change bridge's multicast
813 Internet Group Management Protocol (IGMP) version.
815 * systemd-networkd's FooOverUDP support gained the ability to configure
816 local and peer IP addresses via Local= and Peer=. A new option
817 PeerPort= may be used to configure the peer's IP port.
819 * systemd-networkd's TUN support gained a new setting VnetHeader= for
820 tweaking Generic Segment Offload support.
822 * networkctl gained a new "delete" command for removing virtual network
823 devices, as well as a new "--stats" switch for showing device
826 * networkd.conf gained a new setting SpeedMeter= and
827 SpeedMeterIntervalSec=, to measure bitrate of network interfaces. The
828 measured speed may be shown by 'networkctl status'.
830 * "networkctl status" now displays MTU and queue lengths, and more
831 detailed information about VXLAN and bridge devices.
833 * systemd-networkd's .network and .link files gained a new Property=
834 setting in the [Match] section, to match against devices with
835 specific udev properties.
837 * systemd-networkd's tunnel support gained a new option
838 AssignToLoopback= for selecting whether to use the loopback device
839 "lo" as underlying device.
841 * systemd-networkd's MACAddress= setting in the [Neighbor] section has
842 been renamed to LinkLayerAddress=, and it now allows configuration of
845 * systemd-networkd's handling of the kernel's disable_ipv6 sysctl is
846 simplified: systemd-networkd will disable the sysctl (enable IPv6) if
847 IPv6 configuration (static or DHCPv6) was found for a given
848 interface. It will not touch the sysctl otherwise.
850 * The order of entries is $PATH used by the user manager instance was
851 changed to put bin/ entries before the corresponding sbin/ entries.
852 It is recommended to not rely on this order, and only ever have one
853 binary with a given name in the system paths under /usr.
855 * A new tool systemd-network-generator has been added that may generate
856 .network, .netdev and .link files from IP configuration specified on
857 the kernel command line in the format used by Dracut.
859 * The CriticalConnection= setting in .network files is now deprecated,
860 and replaced by a new KeepConfiguration= setting which allows more
861 detailed configuration of the IP configuration to keep in place.
863 * systemd-analyze gained a few new verbs:
865 - "systemd-analyze timestamp" parses and converts timestamps. This is
866 similar to the existing "systemd-analyze calendar" command which
867 does the same for recurring calendar events.
869 - "systemd-analyze timespan" parses and converts timespans (i.e.
870 durations as opposed to points in time).
872 - "systemd-analyze condition" will parse and test ConditionXYZ=
875 - "systemd-analyze exit-status" will parse and convert exit status
876 codes to their names and back.
878 - "systemd-analyze unit-files" will print a list of all unit
879 file paths and unit aliases.
881 * SuccessExitStatus=, RestartPreventExitStatus=, and
882 RestartForceExitStatus= now accept exit status names (e.g. "DATAERR"
883 is equivalent to "65"). Those exit status name mappings may be
884 displayed with the systemd-analyze exit-status verb describe above.
886 * systemd-logind now exposes a per-session SetBrightness() bus call,
887 which may be used to securely change the brightness of a kernel
888 brightness device, if it belongs to the session's seat. By using this
889 call unprivileged clients can make changes to "backlight" and "leds"
890 devices securely with strict requirements on session membership.
891 Desktop environments may use this to generically make brightness
892 changes to such devices without shipping private SUID binaries or
893 udev rules for that purpose.
895 * "udevadm info" gained a --wait-for-initialization switch to wait for
896 a device to be initialized.
898 * systemd-hibernate-resume-generator will now look for resumeflags= on
899 the kernel command line, which is similar to rootflags= and may be
900 used to configure device timeout for the hibernation device.
902 * sd-event learnt a new API call sd_event_source_disable_unref() for
903 disabling and unref'ing an event source in a single function. A
904 related call sd_event_source_disable_unrefp() has been added for use
905 with gcc's cleanup extension.
907 * The sd-id128.h public API gained a new definition
908 SD_ID128_UUID_FORMAT_STR for formatting a 128bit ID in UUID format
911 * "busctl introspect" gained a new switch --xml-interface for dumping
912 XML introspection data unmodified.
914 * PID 1 may now show the unit name instead of the unit description
915 string in its status output during boot. This may be configured in
916 the StatusUnitFormat= setting in /etc/systemd/system.conf or the
917 kernel command line option systemd.status_unit_format=.
919 * PID 1 now understands a new option KExecWatchdogSec= in
920 /etc/systemd/system.conf to set a watchdog timeout for kexec reboots.
921 Previously watchdog functionality was only available for regular
922 reboots. The new setting defaults to off, because we don't know in
923 the general case if the watchdog will be reset after kexec (some
924 drivers do reset it, but not all), and the new userspace might not be
925 configured to handle the watchdog.
927 Moreover, the old ShutdownWatchdogSec= setting has been renamed to
928 RebootWatchdogSec= to more clearly communicate what it is about. The
929 old name is still accepted for compatibility.
931 * The systemd.debug_shell kernel command line option now optionally
932 takes a tty name to spawn the debug shell on, which allows a
933 different tty to be selected than the built-in default.
935 * Service units gained a new ExecCondition= setting which will run
936 before ExecStartPre= and either continue execution of the unit (for
937 clean exit codes), stop execution without marking the unit failed
938 (for exit codes 1 through 254), or stop execution and fail the unit
939 (for exit code 255 or abnormal termination).
941 * A new service systemd-pstore.service has been added that pulls data
942 from /sys/fs/pstore/ and saves it to /var/lib/pstore for later
945 * timedatectl gained new verbs for configuring per-interface NTP
946 service configuration for systemd-timesyncd.
948 * "localectl list-locales" won't list non-UTF-8 locales anymore. It's
949 2019. (You can set non-UTF-8 locales though, if you know their name.)
951 * If variable assignments in sysctl.d/ files are prefixed with "-" any
952 failures to apply them are now ignored.
954 * systemd-random-seed.service now optionally credits entropy when
955 applying the seed to the system. Set $SYSTEMD_RANDOM_SEED_CREDIT to
956 true for the service to enable this behaviour, but please consult the
957 documentation first, since this comes with a couple of caveats.
959 * systemd-random-seed.service is now a synchronization point for full
960 initialization of the kernel's entropy pool. Services that require
961 /dev/urandom to be correctly initialized should be ordered after this
964 * The systemd-boot boot loader has been updated to optionally maintain
965 a random seed file in the EFI System Partition (ESP). During the boot
966 phase, this random seed is read and updated with a new seed
967 cryptographically derived from it. Another derived seed is passed to
968 the OS. The latter seed is then credited to the kernel's entropy pool
969 very early during userspace initialization (from PID 1). This allows
970 systems to boot up with a fully initialized kernel entropy pool from
971 earliest boot on, and thus entirely removes all entropy pool
972 initialization delays from systems using systemd-boot. Special care
973 is taken to ensure different seeds are derived on system images
974 replicated to multiple systems. "bootctl status" will show whether
975 a seed was received from the boot loader.
977 * bootctl gained two new verbs:
979 - "bootctl random-seed" will generate the file in ESP and an EFI
980 variable to allow a random seed to be passed to the OS as described
983 - "bootctl is-installed" checks whether systemd-boot is currently
986 * bootctl will warn if it detects that boot entries are misconfigured
987 (for example if the kernel image was removed without purging the
990 * A new document has been added describing systemd's use and support
991 for the kernel's entropy pool subsystem:
993 https://systemd.io/RANDOM_SEEDS
995 * When the system is hibernated the swap device to write the
996 hibernation image to is now automatically picked from all available
997 swap devices, preferring the swap device with the highest configured
998 priority over all others, and picking the device with the most free
999 space if there are multiple devices with the highest priority.
1001 * /etc/crypttab support has learnt a new keyfile-timeout= per-device
1002 option that permits selecting the timeout how long to wait for a
1003 device with an encryption key before asking for the password.
1005 * IOWeight= has learnt to properly set the IO weight when using the
1006 BFQ scheduler officially found in kernels 5.0+.
1008 * A new mailing list has been created for reporting of security issues:
1009 systemd-security@redhat.com. For mode details, see
1010 https://systemd.io/CONTRIBUTING#security-vulnerability-reports.
1012 Contributions from: Aaron Barany, Adrian Bunk, Alan Jenkins, Albrecht
1013 Lohofener, Andrej Valek, Anita Zhang, Arian van Putten, Balint Reczey,
1014 Bastien Nocera, Ben Boeckel, Benjamin Robin, camoz, Chen Qi, Chris
1015 Chiu, Chris Down, Christian Göttsche, Christian Kellner, Clinton Roy,
1016 Connor Reeder, Daniel Black, Daniel Lublin, Daniele Medri, Dan
1017 Streetman, Dave Reisner, Dave Ross, David Art, David Tardon, Debarshi
1018 Ray, Dimitri John Ledkov, Dominick Grift, Donald Buczek, Douglas
1019 Christman, Eric DeVolder, EtherGraf, Evgeny Vereshchagin, Feldwor,
1020 Felix Riemann, Florian Dollinger, Francesco Pennica, Franck Bui,
1021 Frantisek Sumsal, Franz Pletz, frederik, Hans de Goede, Iago López
1022 Galeiras, Insun Pyo, Ivan Shapovalov, Iwan Timmer, Jack, Jakob
1023 Unterwurzacher, Jan Chren, Jan Klötzke, Jan Losinski, Jan Pokorný, Jan
1024 Synacek, Jan-Michael Brummer, Jeka Pats, Jeremy Soller, Jérémy Rosen,
1025 Jiri Pirko, Joe Lin, Joerg Behrmann, Joe Richey, Jóhann B. Guðmundsson,
1026 Johannes Christ, Johannes Schmitz, Jonathan Rouleau, Jorge Niedbalski,
1027 Jörg Thalheim, Kai Krakow, Kai Lüke, Karel Zak, Kashyap Chamarthy,
1028 Krayushkin Konstantin, Lennart Poettering, Lubomir Rintel, Luca
1029 Boccassi, Luís Ferreira, Marc-André Lureau, Markus Felten, Martin Pitt,
1030 Matthew Leeds, Mattias Jernberg, Michael Biebl, Michael Olbrich,
1031 Michael Prokop, Michael Stapelberg, Michael Zhivich, Michal Koutný,
1032 Michal Sekletar, Mike Gilbert, Milan Broz, Miroslav Lichvar, mpe85,
1033 Mr-Foo, Network Silence, Oliver Harley, pan93412, Paul Menzel, pEJipE,
1034 Peter A. Bigot, Philip Withnall, Piotr Drąg, Rafael Fontenelle, Robert
1035 Scheck, Roberto Santalla, Ronan Pigott, root, RussianNeuroMancer,
1036 Sebastian Jennen, shinygold, Shreyas Behera, Simon Schricker, Susant
1037 Sahani, Thadeu Lima de Souza Cascardo, Theo Ouzhinski, Thiebaud
1038 Weksteen, Thomas Haller, Thomas Weißschuh, Tomas Mraz, Tommi Rantala,
1039 Topi Miettinen, VD-Lycos, ven, Vladimir Yerilov, Wieland Hoffmann,
1040 William A. Kennington III, William Wold, Xi Ruoyao, Yuri Chornoivan,
1041 Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek, Zhang Xianwei
1043 – Camerino, 2019-09-03
1047 * In .link files, MACAddressPolicy=persistent (the default) is changed
1048 to cover more devices. For devices like bridges, tun, tap, bond, and
1049 similar interfaces that do not have other identifying information,
1050 the interface name is used as the basis for persistent seed for MAC
1051 and IPv4LL addresses. The way that devices that were handled
1052 previously is not changed, and this change is about covering more
1053 devices then previously by the "persistent" policy.
1055 MACAddressPolicy=random may be used to force randomized MACs and
1056 IPv4LL addresses for a device if desired.
1058 Hint: the log output from udev (at debug level) was enhanced to
1059 clarify what policy is followed and which attributes are used.
1060 `SYSTEMD_LOG_LEVEL=debug udevadm test-builtin net_setup_link /sys/class/net/<name>`
1061 may be used to view this.
1063 Hint: if a bridge interface is created without any slaves, and gains
1064 a slave later, then now the bridge does not inherit slave's MAC.
1065 To inherit slave's MAC, for example, create the following file:
1067 # /etc/systemd/network/98-bridge-inherit-mac.link
1072 MACAddressPolicy=none
1075 * The .device units generated by systemd-fstab-generator and other
1076 generators do not automatically pull in the corresponding .mount unit
1077 as a Wants= dependency. This means that simply plugging in the device
1078 will not cause the mount unit to be started automatically. But please
1079 note that the mount unit may be started for other reasons, in
1080 particular if it is part of local-fs.target, and any unit which
1081 (transitively) depends on local-fs.target is started.
1083 * networkctl list/status/lldp now accept globbing wildcards for network
1084 interface names to match against all existing interfaces.
1086 * The $PIDFILE environment variable is set to point the absolute path
1087 configured with PIDFile= for processes of that service.
1089 * The fallback DNS server list was augmented with Cloudflare public DNS
1090 servers. Use `-Ddns-servers=` to set a different fallback.
1092 * A new special target usb-gadget.target will be started automatically
1093 when a USB Device Controller is detected (which means that the system
1094 is a USB peripheral).
1096 * A new unit setting CPUQuotaPeriodSec= assigns the time period
1097 relatively to which the CPU time quota specified by CPUQuota= is
1100 * A new unit setting ProtectHostname= may be used to prevent services
1101 from modifying hostname information (even if they otherwise would
1102 have privileges to do so).
1104 * A new unit setting NetworkNamespacePath= may be used to specify a
1105 namespace for service or socket units through a path referring to a
1106 Linux network namespace pseudo-file.
1108 * The PrivateNetwork= setting and JoinsNamespaceOf= dependencies now
1109 have an effect on .socket units: when used the listening socket is
1110 created within the configured network namespace instead of the host
1113 * ExecStart= command lines in unit files may now be prefixed with ':'
1114 in which case environment variable substitution is
1115 disabled. (Supported for the other ExecXYZ= settings, too.)
1117 * .timer units gained two new boolean settings OnClockChange= and
1118 OnTimezoneChange= which may be used to also trigger a unit when the
1119 system clock is changed or the local timezone is
1120 modified. systemd-run has been updated to make these options easily
1121 accessible from the command line for transient timers.
1123 * Two new conditions for units have been added: ConditionMemory= may be
1124 used to conditionalize a unit based on installed system
1125 RAM. ConditionCPUs= may be used to conditionalize a unit based on
1126 installed CPU cores.
1128 * The @default system call filter group understood by SystemCallFilter=
1129 has been updated to include the new rseq() system call introduced in
1132 * A new time-set.target has been added that indicates that the system
1133 time has been set from a local source (possibly imprecise). The
1134 existing time-sync.target is stronger and indicates that the time has
1135 been synchronized with a precise external source. Services where
1136 approximate time is sufficient should use the new target.
1138 * "systemctl start" (and related commands) learnt a new
1139 --show-transaction option. If specified brief information about all
1140 jobs queued because of the requested operation is shown.
1142 * systemd-networkd recognizes a new operation state 'enslaved', used
1143 (instead of 'degraded' or 'carrier') for interfaces which form a
1144 bridge, bond, or similar, and an new 'degraded-carrier' operational
1145 state used for the bond or bridge master interface when one of the
1146 enslaved devices is not operational.
1148 * .network files learnt the new IgnoreCarrierLoss= option for leaving
1149 networks configured even if the carrier is lost.
1151 * The RequiredForOnline= setting in .network files may now specify a
1152 minimum operational state required for the interface to be considered
1153 "online" by systemd-networkd-wait-online. Related to this
1154 systemd-networkd-wait-online gained a new option --operational-state=
1155 to configure the same, and its --interface= option was updated to
1156 optionally also take an operational state specific for an interface.
1158 * systemd-networkd-wait-online gained a new setting --any for waiting
1159 for only one of the requested interfaces instead of all of them.
1161 * systemd-networkd now implements L2TP tunnels.
1163 * Two new .network settings UseAutonomousPrefix= and UseOnLinkPrefix=
1164 may be used to cause autonomous and onlink prefixes received in IPv6
1165 Router Advertisements to be ignored.
1167 * New MulticastFlood=, NeighborSuppression=, and Learning= .network
1168 file settings may be used to tweak bridge behaviour.
1170 * The new TripleSampling= option in .network files may be used to
1171 configure CAN triple sampling.
1173 * A new .netdev settings PrivateKeyFile= and PresharedKeyFile= may be
1174 used to point to private or preshared key for a WireGuard interface.
1176 * /etc/crypttab now supports the same-cpu-crypt and
1177 submit-from-crypt-cpus options to tweak encryption work scheduling
1180 * systemd-tmpfiles will now take a BSD file lock before operating on a
1181 contents of directory. This may be used to temporarily exclude
1182 directories from aging by taking the same lock (useful for example
1183 when extracting a tarball into /tmp or /var/tmp as a privileged user,
1184 which might create files with really old timestamps, which
1185 nevertheless should not be deleted). For further details, see:
1187 https://systemd.io/TEMPORARY_DIRECTORIES
1189 * systemd-tmpfiles' h line type gained support for the
1190 FS_PROJINHERIT_FL ('P') file attribute (introduced in kernel 4.5),
1191 controlling project quota inheritance.
1193 * sd-boot and bootctl now implement support for an Extended Boot Loader
1194 (XBOOTLDR) partition, that is intended to be mounted to /boot, in
1195 addition to the ESP partition mounted to /efi or /boot/efi.
1196 Configuration file fragments, kernels, initrds and other EFI images
1197 to boot will be loaded from both the ESP and XBOOTLDR partitions.
1198 The XBOOTLDR partition was previously described by the Boot Loader
1199 Specification, but implementation was missing in sd-boot. Support for
1200 this concept allows using the sd-boot boot loader in more
1201 conservative scenarios where the boot loader itself is placed in the
1202 ESP but the kernels to boot (and their metadata) in a separate
1205 * A system may now be booted with systemd.volatile=overlay on the
1206 kernel command line, which causes the root file system to be set up
1207 an overlayfs mount combining the root-only root directory with a
1208 writable tmpfs. In this setup, the underlying root device is not
1209 modified, and any changes are lost at reboot.
1211 * Similar, systemd-nspawn can now boot containers with a volatile
1212 overlayfs root with the new --volatile=overlay switch.
1214 * systemd-nspawn can now consume OCI runtime bundles using a new
1215 --oci-bundle= option. This implementation is fully usable, with most
1216 features in the specification implemented, but since this a lot of
1217 new code and functionality, this feature should most likely not
1218 be used in production yet.
1220 * systemd-nspawn now supports various options described by the OCI
1221 runtime specification on the command-line and in .nspawn files:
1222 --inaccessible=/Inaccessible= may be used to mask parts of the file
1223 system tree, --console=/--pipe may be used to configure how standard
1224 input, output, and error are set up.
1226 * busctl learned the `emit` verb to generate D-Bus signals.
1228 * systemd-analyze cat-config may be used to gather and display
1229 configuration spread over multiple files, for example system and user
1230 presets, tmpfiles.d, sysusers.d, udev rules, etc.
1232 * systemd-analyze calendar now takes an optional new parameter
1233 --iterations= which may be used to show a maximum number of iterations
1234 the specified expression will elapse next.
1236 * The sd-bus C API gained support for naming method parameters in the
1239 * systemd-logind gained D-Bus APIs to specify the "reboot parameter"
1240 the reboot() system call expects.
1242 * journalctl learnt a new --cursor-file= option that points to a file
1243 from which a cursor should be loaded in the beginning and to which
1244 the updated cursor should be stored at the end.
1246 * ACRN hypervisor and Windows Subsystem for Linux (WSL) are now
1247 detected by systemd-detect-virt (and may also be used in
1248 ConditionVirtualization=).
1250 * The behaviour of systemd-logind may now be modified with environment
1251 variables $SYSTEMD_REBOOT_TO_FIRMWARE_SETUP,
1252 $SYSTEMD_REBOOT_TO_BOOT_LOADER_MENU, and
1253 $SYSTEMD_REBOOT_TO_BOOT_LOADER_ENTRY. They cause logind to either
1254 skip the relevant operation completely (when set to false), or to
1255 create a flag file in /run/systemd (when set to true), instead of
1256 actually commencing the real operation when requested. The presence
1257 of /run/systemd/reboot-to-firmware-setup,
1258 /run/systemd/reboot-to-boot-loader-menu, and
1259 /run/systemd/reboot-to-boot-loader-entry, may be used by alternative
1260 boot loader implementations to replace some steps logind performs
1261 during reboot with their own operations.
1263 * systemctl can be used to request a reboot into the boot loader menu
1264 or a specific boot loader entry with the new --boot-load-menu= and
1265 --boot-loader-entry= options to a reboot command. (This requires a
1266 boot loader that supports this, for example sd-boot.)
1268 * kernel-install will no longer unconditionally create the output
1269 directory (e.g. /efi/<machine-id>/<kernel-version>) for boot loader
1270 snippets, but will do only if the machine-specific parent directory
1271 (i.e. /efi/<machine-id>/) already exists. bootctl has been modified
1272 to create this parent directory during sd-boot installation.
1274 This makes it easier to use kernel-install with plugins which support
1275 a different layout of the bootloader partitions (for example grub2).
1277 * During package installation (with `ninja install`), we would create
1278 symlinks for getty@tty1.service, systemd-networkd.service,
1279 systemd-networkd.socket, systemd-resolved.service,
1280 remote-cryptsetup.target, remote-fs.target,
1281 systemd-networkd-wait-online.service, and systemd-timesyncd.service
1282 in /etc, as if `systemctl enable` was called for those units, to make
1283 the system usable immediately after installation. Now this is not
1284 done anymore, and instead calling `systemctl preset-all` is
1285 recommended after the first installation of systemd.
1287 * A new boolean sandboxing option RestrictSUIDSGID= has been added that
1288 is built on seccomp. When turned on creation of SUID/SGID files is
1291 * The NoNewPrivileges= and the new RestrictSUIDSGID= options are now
1292 implied if DynamicUser= is turned on for a service. This hardens
1293 these services, so that they neither can benefit from nor create
1294 SUID/SGID executables. This is a minor compatibility breakage, given
1295 that when DynamicUser= was first introduced SUID/SGID behaviour was
1296 unaffected. However, the security benefit of these two options is
1297 substantial, and the setting is still relatively new, hence we opted
1298 to make it mandatory for services with dynamic users.
1300 Contributions from: Adam Jackson, Alexander Tsoy, Andrey Yashkin,
1301 Andrzej Pietrasiewicz, Anita Zhang, Balint Reczey, Beniamino Galvani,
1302 Ben Iofel, Benjamin Berg, Benjamin Dahlhoff, Chris, Chris Morin,
1303 Christopher Wong, Claudius Ellsel, Clemens Gruber, dana, Daniel Black,
1304 Davide Cavalca, David Michael, David Rheinsberg, emersion, Evgeny
1305 Vereshchagin, Filipe Brandenburger, Franck Bui, Frantisek Sumsal,
1306 Giacinto Cifelli, Hans de Goede, Hugo Kindel, Ignat Korchagin, Insun
1307 Pyo, Jan Engelhardt, Jonas Dorel, Jonathan Lebon, Jonathon Kowalski,
1308 Jörg Sommer, Jörg Thalheim, Jussi Pakkanen, Kai-Heng Feng, Lennart
1309 Poettering, Lubomir Rintel, Luís Ferreira, Martin Pitt, Matthias
1310 Klumpp, Michael Biebl, Michael Niewöhner, Michael Olbrich, Michal
1311 Sekletar, Mike Lothian, Paul Menzel, Piotr Drąg, Riccardo Schirone,
1312 Robin Elvedi, Roman Kulikov, Ronald Tschalär, Ross Burton, Ryan
1313 Gonzalez, Sebastian Krzyszkowiak, Stephane Chazelas, StKob, Susant
1314 Sahani, Sylvain Plantefève, Szabolcs Fruhwald, Taro Yamada, Theo
1315 Ouzhinski, Thomas Haller, Tobias Jungel, Tom Yan, Tony Asleson, Topi
1316 Miettinen, unixsysadmin, Van Laser, Vesa Jääskeläinen, Yu, Li-Yu,
1317 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
1319 — Warsaw, 2019-04-11
1323 * The default locale can now be configured at compile time. Otherwise,
1324 a suitable default will be selected automatically (one of C.UTF-8,
1325 en_US.UTF-8, and C).
1327 * The version string shown by systemd and other tools now includes the
1328 git commit hash when built from git. An override may be specified
1329 during compilation, which is intended to be used by distributions to
1330 include the package release information.
1332 * systemd-cat can now filter standard input and standard error streams
1333 for different syslog priorities using the new --stderr-priority=
1336 * systemd-journald and systemd-journal-remote reject entries which
1337 contain too many fields (CVE-2018-16865) and set limits on the
1338 process' command line length (CVE-2018-16864).
1340 * $DBUS_SESSION_BUS_ADDRESS environment variable is set by pam_systemd
1343 * A new network device NamePolicy "keep" is implemented for link files,
1344 and used by default in 99-default.link (the fallback configuration
1345 provided by systemd). With this policy, if the network device name
1346 was already set by userspace, the device will not be renamed again.
1347 This matches the naming scheme that was implemented before
1348 systemd-240. If naming-scheme < 240 is specified, the "keep" policy
1349 is also enabled by default, even if not specified. Effectively, this
1350 means that if naming-scheme >= 240 is specified, network devices will
1351 be renamed according to the configuration, even if they have been
1352 renamed already, if "keep" is not specified as the naming policy in
1353 the .link file. The 99-default.link file provided by systemd includes
1354 "keep" for backwards compatibility, but it is recommended for user
1355 installed .link files to *not* include it.
1357 The "kernel" policy, which keeps kernel names declared to be
1358 "persistent", now works again as documented.
1360 * kernel-install script now optionally takes the paths to one or more
1361 initrd files, and passes them to all plugins.
1363 * The mincore() system call has been dropped from the @system-service
1364 system call filter group, as it is pretty exotic and may potentially
1365 used for side-channel attacks.
1367 * -fPIE is dropped from compiler and linker options. Please specify
1368 -Db_pie=true option to meson to build position-independent
1369 executables. Note that the meson option is supported since meson-0.49.
1371 * The fs.protected_regular and fs.protected_fifos sysctls, which were
1372 added in Linux 4.19 to make some data spoofing attacks harder, are
1373 now enabled by default. While this will hopefully improve the
1374 security of most installations, it is technically a backwards
1375 incompatible change; to disable these sysctls again, place the
1376 following lines in /etc/sysctl.d/60-protected.conf or a similar file:
1378 fs.protected_regular = 0
1379 fs.protected_fifos = 0
1381 Note that the similar hardlink and symlink protection has been
1382 enabled since v199, and may be disabled likewise.
1384 * The files read from the EnvironmentFile= setting in unit files now
1385 parse backslashes inside quotes literally, matching the behaviour of
1388 * udevadm trigger, udevadm control, udevadm settle and udevadm monitor
1389 now automatically become NOPs when run in a chroot() environment.
1391 * The tmpfiles.d/ "C" line type will now copy directory trees not only
1392 when the destination is so far missing, but also if it already exists
1393 as a directory and is empty. This is useful to cater for systems
1394 where directory trees are put together from multiple separate mount
1395 points but otherwise empty.
1397 * A new function sd_bus_close_unref() (and the associated
1398 sd_bus_close_unrefp()) has been added to libsystemd, that combines
1399 sd_bus_close() and sd_bus_unref() in one.
1401 * udevadm control learnt a new option for --ping for testing whether a
1402 systemd-udevd instance is running and reacting.
1404 * udevadm trigger learnt a new option for --wait-daemon for waiting
1405 systemd-udevd daemon to be initialized.
1407 Contributions from: Aaron Plattner, Alberts Muktupāvels, Alex Mayer,
1408 Ayman Bagabas, Beniamino Galvani, Burt P, Chris Down, Chris Lamb, Chris
1409 Morin, Christian Hesse, Claudius Ellsel, dana, Daniel Axtens, Daniele
1410 Medri, Dave Reisner, David Santamaría Rogado, Diego Canuhe, Dimitri
1411 John Ledkov, Evgeny Vereshchagin, Fabrice Fontaine, Filipe
1412 Brandenburger, Franck Bui, Frantisek Sumsal, govwin, Hans de Goede,
1413 James Hilliard, Jan Engelhardt, Jani Uusitalo, Jan Janssen, Jan
1414 Synacek, Jonathan McDowell, Jonathan Roemer, Jonathon Kowalski, Joost
1415 Heitbrink, Jörg Thalheim, Lance, Lennart Poettering, Louis Taylor,
1416 Lucas Werkmeister, Mantas Mikulėnas, Marc-Antoine Perennou,
1417 marvelousblack, Michael Biebl, Michael Sloan, Michal Sekletar, Mike
1418 Auty, Mike Gilbert, Mikhail Kasimov, Neil Brown, Niklas Hambüchen,
1419 Patrick Williams, Paul Seyfert, Peter Hutterer, Philip Withnall, Roger
1420 James, Ronnie P. Thomas, Ryan Gonzalez, Sam Morris, Stephan Edel,
1421 Stephan Gerhold, Susant Sahani, Taro Yamada, Thomas Haller, Topi
1422 Miettinen, YiFei Zhu, YmrDtnJu, YunQiang Su, Yu Watanabe, Zbigniew
1423 Jędrzejewski-Szmek, zsergeant77, Дамјан Георгиевски
1425 — Berlin, 2019-02-14
1429 * NoNewPrivileges=yes has been set for all long-running services
1430 implemented by systemd. Previously, this was problematic due to
1431 SELinux (as this would also prohibit the transition from PID1's label
1432 to the service's label). This restriction has since been lifted, but
1433 an SELinux policy update is required.
1434 (See e.g. https://github.com/fedora-selinux/selinux-policy/pull/234.)
1436 * DynamicUser=yes is dropped from systemd-networkd.service,
1437 systemd-resolved.service and systemd-timesyncd.service, which was
1438 enabled in v239 for systemd-networkd.service and systemd-resolved.service,
1439 and since v236 for systemd-timesyncd.service. The users and groups
1440 systemd-network, systemd-resolve and systemd-timesync are created
1441 by systemd-sysusers again. Distributors or system administrators
1442 may need to create these users and groups if they not exist (or need
1443 to re-enable DynamicUser= for those units) while upgrading systemd.
1444 Also, the clock file for systemd-timesyncd may need to move from
1445 /var/lib/private/systemd/timesync/clock to /var/lib/systemd/timesync/clock.
1447 * When unit files are loaded from disk, previously systemd would
1448 sometimes (depending on the unit loading order) load units from the
1449 target path of symlinks in .wants/ or .requires/ directories of other
1450 units. This meant that unit could be loaded from different paths
1451 depending on whether the unit was requested explicitly or as a
1452 dependency of another unit, not honouring the priority of directories
1453 in search path. It also meant that it was possible to successfully
1454 load and start units which are not found in the unit search path, as
1455 long as they were requested as a dependency and linked to from
1456 .wants/ or .requires/. The target paths of those symlinks are not
1457 used for loading units anymore and the unit file must be found in
1460 * A new service type has been added: Type=exec. It's very similar to
1461 Type=simple but ensures the service manager will wait for both fork()
1462 and execve() of the main service binary to complete before proceeding
1463 with follow-up units. This is primarily useful so that the manager
1464 propagates any errors in the preparation phase of service execution
1465 back to the job that requested the unit to be started. For example,
1466 consider a service that has ExecStart= set to a file system binary
1467 that doesn't exist. With Type=simple starting the unit would be
1468 considered instantly successful, as only fork() has to complete
1469 successfully and the manager does not wait for execve(), and hence
1470 its failure is seen "too late". With the new Type=exec service type
1471 starting the unit will fail, as the manager will wait for the
1472 execve() and notice its failure, which is then propagated back to the
1475 NOTE: with the next release 241 of systemd we intend to change the
1476 systemd-run tool to default to Type=exec for transient services
1477 started by it. This should be mostly safe, but in specific corner
1478 cases might result in problems, as the systemd-run tool will then
1479 block on NSS calls (such as user name look-ups due to User=) done
1480 between the fork() and execve(), which under specific circumstances
1481 might cause problems. It is recommended to specify "-p Type=simple"
1482 explicitly in the few cases where this applies. For regular,
1483 non-transient services (i.e. those defined with unit files on disk)
1484 we will continue to default to Type=simple.
1486 * The Linux kernel's current default RLIMIT_NOFILE resource limit for
1487 userspace processes is set to 1024 (soft) and 4096
1488 (hard). Previously, systemd passed this on unmodified to all
1489 processes it forked off. With this systemd release the hard limit
1490 systemd passes on is increased to 512K, overriding the kernel's
1491 defaults and substantially increasing the number of simultaneous file
1492 descriptors unprivileged userspace processes can allocate. Note that
1493 the soft limit remains at 1024 for compatibility reasons: the
1494 traditional UNIX select() call cannot deal with file descriptors >=
1495 1024 and increasing the soft limit globally might thus result in
1496 programs unexpectedly allocating a high file descriptor and thus
1497 failing abnormally when attempting to use it with select() (of
1498 course, programs shouldn't use select() anymore, and prefer
1499 poll()/epoll, but the call unfortunately remains undeservedly popular
1500 at this time). This change reflects the fact that file descriptor
1501 handling in the Linux kernel has been optimized in more recent
1502 kernels and allocating large numbers of them should be much cheaper
1503 both in memory and in performance than it used to be. Programs that
1504 want to take benefit of the increased limit have to "opt-in" into
1505 high file descriptors explicitly by raising their soft limit. Of
1506 course, when they do that they must acknowledge that they cannot use
1507 select() anymore (and neither can any shared library they use — or
1508 any shared library used by any shared library they use and so on).
1509 Which default hard limit is most appropriate is of course hard to
1510 decide. However, given reports that ~300K file descriptors are used
1511 in real-life applications we believe 512K is sufficiently high as new
1512 default for now. Note that there are also reports that using very
1513 high hard limits (e.g. 1G) is problematic: some software allocates
1514 large arrays with one element for each potential file descriptor
1515 (Java, …) — a high hard limit thus triggers excessively large memory
1516 allocations in these applications. Hopefully, the new default of 512K
1517 is a good middle ground: higher than what real-life applications
1518 currently need, and low enough for avoid triggering excessively large
1519 allocations in problematic software. (And yes, somebody should fix
1522 * The fs.nr_open and fs.file-max sysctls are now automatically bumped
1523 to the highest possible values, as separate accounting of file
1524 descriptors is no longer necessary, as memcg tracks them correctly as
1525 part of the memory accounting anyway. Thus, from the four limits on
1526 file descriptors currently enforced (fs.file-max, fs.nr_open,
1527 RLIMIT_NOFILE hard, RLIMIT_NOFILE soft) we turn off the first two,
1528 and keep only the latter two. A set of build-time options
1529 (-Dbump-proc-sys-fs-file-max=false and -Dbump-proc-sys-fs-nr-open=false)
1530 has been added to revert this change in behaviour, which might be
1531 an option for systems that turn off memcg in the kernel.
1533 * When no /etc/locale.conf file exists (and hence no locale settings
1534 are in place), systemd will now use the "C.UTF-8" locale by default,
1535 and set LANG= to it. This locale is supported by various
1536 distributions including Fedora, with clear indications that upstream
1537 glibc is going to make it available too. This locale enables UTF-8
1538 mode by default, which appears appropriate for 2018.
1540 * The "net.ipv4.conf.all.rp_filter" sysctl will now be set to 2 by
1541 default. This effectively switches the RFC3704 Reverse Path filtering
1542 from Strict mode to Loose mode. This is more appropriate for hosts
1543 that have multiple links with routes to the same networks (e.g.
1544 a client with a Wi-Fi and Ethernet both connected to the internet).
1546 Consult the kernel documentation for details on this sysctl:
1547 https://www.kernel.org/doc/Documentation/networking/ip-sysctl.txt
1549 * CPUAccounting=yes no longer enables the CPU controller when using
1550 kernel 4.15+ and the unified cgroup hierarchy, as required accounting
1551 statistics are now provided independently from the CPU controller.
1553 * Support for disabling a particular cgroup controller within a sub-tree
1554 has been added through the DisableControllers= directive.
1556 * cgroup_no_v1=all on the kernel command line now also implies
1557 using the unified cgroup hierarchy, unless one explicitly passes
1558 systemd.unified_cgroup_hierarchy=0 on the kernel command line.
1560 * The new "MemoryMin=" unit file property may now be used to set the
1561 memory usage protection limit of processes invoked by the unit. This
1562 controls the cgroup v2 memory.min attribute. Similarly, the new
1563 "IODeviceLatencyTargetSec=" property has been added, wrapping the new
1564 cgroup v2 io.latency cgroup property for configuring per-service I/O
1567 * systemd now supports the cgroup v2 devices BPF logic, as counterpart
1568 to the cgroup v1 "devices" cgroup controller.
1570 * systemd-escape now is able to combine --unescape with --template. It
1571 also learnt a new option --instance for extracting and unescaping the
1572 instance part of a unit name.
1574 * sd-bus now provides the sd_bus_message_readv() which is similar to
1575 sd_bus_message_read() but takes a va_list object. The pair
1576 sd_bus_set_method_call_timeout() and sd_bus_get_method_call_timeout()
1577 has been added for configuring the default method call timeout to
1578 use. sd_bus_error_move() may be used to efficiently move the contents
1579 from one sd_bus_error structure to another, invalidating the
1580 source. sd_bus_set_close_on_exit() and sd_bus_get_close_on_exit() may
1581 be used to control whether a bus connection object is automatically
1582 flushed when an sd-event loop is exited.
1584 * When processing classic BSD syslog log messages, journald will now
1585 save the original time-stamp string supplied in the new
1586 SYSLOG_TIMESTAMP= journal field. This permits consumers to
1587 reconstruct the original BSD syslog message more correctly.
1589 * StandardOutput=/StandardError= in service files gained support for
1590 new "append:…" parameters, for connecting STDOUT/STDERR of a service
1591 to a file, and appending to it.
1593 * The signal to use as last step of killing of unit processes is now
1594 configurable. Previously it was hard-coded to SIGKILL, which may now
1595 be overridden with the new KillSignal= setting. Note that this is the
1596 signal used when regular termination (i.e. SIGTERM) does not suffice.
1597 Similarly, the signal used when aborting a program in case of a
1598 watchdog timeout may now be configured too (WatchdogSignal=).
1600 * The XDG_SESSION_DESKTOP environment variable may now be configured in
1601 the pam_systemd argument line, using the new desktop= switch. This is
1602 useful to initialize it properly from a display manager without
1603 having to touch C code.
1605 * Most configuration options that previously accepted percentage values
1606 now also accept permille values with the '‰' suffix (instead of '%').
1608 * systemd-resolved may now optionally use OpenSSL instead of GnuTLS for
1611 * systemd-resolved's configuration file resolved.conf gained a new
1612 option ReadEtcHosts= which may be used to turn off processing and
1613 honoring /etc/hosts entries.
1615 * The "--wait" switch may now be passed to "systemctl
1616 is-system-running", in which case the tool will synchronously wait
1617 until the system finished start-up.
1619 * hostnamed gained a new bus call to determine the DMI product UUID.
1621 * On x86-64 systemd will now prefer using the RDRAND processor
1622 instruction over /dev/urandom whenever it requires randomness that
1623 neither has to be crypto-grade nor should be reproducible. This
1624 should substantially reduce the amount of entropy systemd requests
1625 from the kernel during initialization on such systems, though not
1626 reduce it to zero. (Why not zero? systemd still needs to allocate
1627 UUIDs and such uniquely, which require high-quality randomness.)
1629 * networkd gained support for Foo-Over-UDP, ERSPAN and ISATAP
1630 tunnels. It also gained a new option ForceDHCPv6PDOtherInformation=
1631 for forcing the "Other Information" bit in IPv6 RA messages. The
1632 bonding logic gained four new options AdActorSystemPriority=,
1633 AdUserPortKey=, AdActorSystem= for configuring various 802.3ad
1634 aspects, and DynamicTransmitLoadBalancing= for enabling dynamic
1635 shuffling of flows. The tunnel logic gained a new
1636 IPv6RapidDeploymentPrefix= option for configuring IPv6 Rapid
1637 Deployment. The policy rule logic gained four new options IPProtocol=,
1638 SourcePort= and DestinationPort=, InvertRule=. The bridge logic gained
1639 support for the MulticastToUnicast= option. networkd also gained
1640 support for configuring static IPv4 ARP or IPv6 neighbor entries.
1642 * .preset files (as read by 'systemctl preset') may now be used to
1643 instantiate services.
1645 * /etc/crypttab now understands the sector-size= option to configure
1646 the sector size for an encrypted partition.
1648 * Key material for encrypted disks may now be placed on a formatted
1649 medium, and referenced from /etc/crypttab by the UUID of the file
1650 system, followed by "=" suffixed by the path to the key file.
1652 * The "collect" udev component has been removed without replacement, as
1653 it is neither used nor maintained.
1655 * When the RuntimeDirectory=, StateDirectory=, CacheDirectory=,
1656 LogsDirectory=, ConfigurationDirectory= settings are used in a
1657 service the executed processes will now receive a set of environment
1658 variables containing the full paths of these directories.
1659 Specifically, RUNTIME_DIRECTORY=, STATE_DIRECTORY, CACHE_DIRECTORY,
1660 LOGS_DIRECTORY, CONFIGURATION_DIRECTORY are now set if these options
1661 are used. Note that these options may be used multiple times per
1662 service in which case the resulting paths will be concatenated and
1663 separated by colons.
1665 * Predictable interface naming has been extended to cover InfiniBand
1666 NICs. They will be exposed with an "ib" prefix.
1668 * tmpfiles.d/ line types may now be suffixed with a '-' character, in
1669 which case the respective line failing is ignored.
1671 * .link files may now be used to configure the equivalent to the
1672 "ethtool advertise" commands.
1674 * The sd-device.h and sd-hwdb.h APIs are now exported, as an
1675 alternative to libudev.h. Previously, the latter was just an internal
1676 wrapper around the former, but now these two APIs are exposed
1679 * sd-id128.h gained a new function sd_id128_get_boot_app_specific()
1680 which calculates an app-specific boot ID similar to how
1681 sd_id128_get_machine_app_specific() generates an app-specific machine
1684 * A new tool systemd-id128 has been added that can be used to determine
1685 and generate various 128bit IDs.
1687 * /etc/os-release gained two new standardized fields DOCUMENTATION_URL=
1690 * systemd-hibernate-resume-generator will now honor the "noresume"
1691 kernel command line option, in which case it will bypass resuming
1692 from any hibernated image.
1694 * The systemd-sleep.conf configuration file gained new options
1695 AllowSuspend=, AllowHibernation=, AllowSuspendThenHibernate=,
1696 AllowHybridSleep= for prohibiting specific sleep modes even if the
1697 kernel exports them.
1699 * portablectl is now officially supported and has thus moved to
1702 * bootctl learnt the two new commands "set-default" and "set-oneshot"
1703 for setting the default boot loader item to boot to (either
1704 persistently or only for the next boot). This is currently only
1705 compatible with sd-boot, but may be implemented on other boot loaders
1706 too, that follow the boot loader interface. The updated interface is
1707 now documented here:
1709 https://systemd.io/BOOT_LOADER_INTERFACE
1711 * A new kernel command line option systemd.early_core_pattern= is now
1712 understood which may be used to influence the core_pattern PID 1
1713 installs during early boot.
1715 * busctl learnt two new options -j and --json= for outputting method
1716 call replies, properties and monitoring output in JSON.
1718 * journalctl's JSON output now supports simple ANSI coloring as well as
1719 a new "json-seq" mode for generating RFC7464 output.
1721 * Unit files now support the %g/%G specifiers that resolve to the UNIX
1722 group/GID of the service manager runs as, similar to the existing
1723 %u/%U specifiers that resolve to the UNIX user/UID.
1725 * systemd-logind learnt a new global configuration option
1726 UserStopDelaySec= that may be set in logind.conf. It specifies how
1727 long the systemd --user instance shall remain started after a user
1728 logs out. This is useful to speed up repetitive re-connections of the
1729 same user, as it means the user's service manager doesn't have to be
1730 stopped/restarted on each iteration, but can be reused between
1731 subsequent options. This setting defaults to 10s. systemd-logind also
1732 exports two new properties on its Manager D-Bus objects indicating
1733 whether the system's lid is currently closed, and whether the system
1736 * systemd gained support for a generic boot counting logic, which
1737 generically permits automatic reverting to older boot loader entries
1738 if newer updated ones don't work. The boot loader side is implemented
1739 in sd-boot, but is kept open for other boot loaders too. For details
1742 https://systemd.io/AUTOMATIC_BOOT_ASSESSMENT
1744 * The SuccessAction=/FailureAction= unit file settings now learnt two
1745 new parameters: "exit" and "exit-force", which result in immediate
1746 exiting of the service manager, and are only useful in systemd --user
1747 and container environments.
1749 * Unit files gained support for a pair of options
1750 FailureActionExitStatus=/SuccessActionExitStatus= for configuring the
1751 exit status to use as service manager exit status when
1752 SuccessAction=/FailureAction= is set to exit or exit-force.
1754 * A pair of LogRateLimitIntervalSec=/LogRateLimitBurst= per-service
1755 options may now be used to configure the log rate limiting applied by
1756 journald per-service.
1758 * systemd-analyze gained a new verb "timespan" for parsing and
1759 normalizing time span values (i.e. strings like "5min 7s 8us").
1761 * systemd-analyze also gained a new verb "security" for analyzing the
1762 security and sand-boxing settings of services in order to determine an
1763 "exposure level" for them, indicating whether a service would benefit
1764 from more sand-boxing options turned on for them.
1766 * "systemd-analyze syscall-filter" will now also show system calls
1767 supported by the local kernel but not included in any of the defined
1770 * .nspawn files now understand the Ephemeral= setting, matching the
1771 --ephemeral command line switch.
1773 * sd-event gained the new APIs sd_event_source_get_floating() and
1774 sd_event_source_set_floating() for controlling whether a specific
1775 event source is "floating", i.e. destroyed along with the even loop
1778 * Unit objects on D-Bus gained a new "Refs" property that lists all
1779 clients that currently have a reference on the unit (to ensure it is
1782 * The JoinControllers= option in system.conf is no longer supported, as
1783 it didn't work correctly, is hard to support properly, is legacy (as
1784 the concept only exists on cgroup v1) and apparently wasn't used.
1786 * Journal messages that are generated whenever a unit enters the failed
1787 state are now tagged with a unique MESSAGE_ID. Similarly, messages
1788 generated whenever a service process exits are now made recognizable,
1789 too. A tagged message is also emitted whenever a unit enters the
1790 "dead" state on success.
1792 * systemd-run gained a new switch --working-directory= for configuring
1793 the working directory of the service to start. A shortcut -d is
1794 equivalent, setting the working directory of the service to the
1795 current working directory of the invoking program. The new --shell
1796 (or just -S) option has been added for invoking the $SHELL of the
1797 caller as a service, and implies --pty --same-dir --wait --collect
1798 --service-type=exec. Or in other words, "systemd-run -S" is now the
1799 quickest way to quickly get an interactive in a fully clean and
1800 well-defined system service context.
1802 * machinectl gained a new verb "import-fs" for importing an OS tree
1803 from a directory. Moreover, when a directory or tarball is imported
1804 and single top-level directory found with the OS itself below the OS
1805 tree is automatically mangled and moved one level up.
1807 * systemd-importd will no longer set up an implicit btrfs loop-back
1808 file system on /var/lib/machines. If one is already set up, it will
1809 continue to be used.
1811 * A new generator "systemd-run-generator" has been added. It will
1812 synthesize a unit from one or more program command lines included in
1813 the kernel command line. This is very useful in container managers
1816 # systemd-nspawn -i someimage.raw -b systemd.run='"some command line"'
1818 This will run "systemd-nspawn" on an image, invoke the specified
1819 command line and immediately shut down the container again, returning
1820 the command line's exit code.
1822 * The block device locking logic is now documented:
1824 https://systemd.io/BLOCK_DEVICE_LOCKING
1826 * loginctl and machinectl now optionally output the various tables in
1827 JSON using the --output= switch. It is our intention to add similar
1828 support to systemctl and all other commands.
1830 * udevadm's query and trigger verb now optionally take a .device unit
1833 * systemd-udevd's network naming logic now understands a new
1834 net.naming-scheme= kernel command line switch, which may be used to
1835 pick a specific version of the naming scheme. This helps stabilizing
1836 interface names even as systemd/udev are updated and the naming logic
1839 * sd-id128.h learnt two new auxiliary helpers: sd_id128_is_allf() and
1840 SD_ID128_ALLF to test if a 128bit ID is set to all 0xFF bytes, and to
1841 initialize one to all 0xFF.
1843 * After loading the SELinux policy systemd will now recursively relabel
1844 all files and directories listed in
1845 /run/systemd/relabel-extra.d/*.relabel (which should be simple
1846 newline separated lists of paths) in addition to the ones it already
1847 implicitly relabels in /run, /dev and /sys. After the relabelling is
1848 completed the *.relabel files (and /run/systemd/relabel-extra.d/) are
1849 removed. This is useful to permit initrds (i.e. code running before
1850 the SELinux policy is in effect) to generate files in the host
1851 filesystem safely and ensure that the correct label is applied during
1852 the transition to the host OS.
1854 * KERNEL API BREAKAGE: Linux kernel 4.18 changed behaviour regarding
1855 mknod() handling in user namespaces. Previously mknod() would always
1856 fail with EPERM in user namespaces. Since 4.18 mknod() will succeed
1857 but device nodes generated that way cannot be opened, and attempts to
1858 open them result in EPERM. This breaks the "graceful fallback" logic
1859 in systemd's PrivateDevices= sand-boxing option. This option is
1860 implemented defensively, so that when systemd detects it runs in a
1861 restricted environment (such as a user namespace, or an environment
1862 where mknod() is blocked through seccomp or absence of CAP_SYS_MKNOD)
1863 where device nodes cannot be created the effect of PrivateDevices= is
1864 bypassed (following the logic that 2nd-level sand-boxing is not
1865 essential if the system systemd runs in is itself already sand-boxed
1866 as a whole). This logic breaks with 4.18 in container managers where
1867 user namespacing is used: suddenly PrivateDevices= succeeds setting
1868 up a private /dev/ file system containing devices nodes — but when
1869 these are opened they don't work.
1871 At this point it is recommended that container managers utilizing
1872 user namespaces that intend to run systemd in the payload explicitly
1873 block mknod() with seccomp or similar, so that the graceful fallback
1876 We are very sorry for the breakage and the requirement to change
1877 container configurations for newer kernels. It's purely caused by an
1878 incompatible kernel change. The relevant kernel developers have been
1879 notified about this userspace breakage quickly, but they chose to
1882 * PermissionsStartOnly= setting is deprecated (but is still supported
1883 for backwards compatibility). The same functionality is provided by
1884 the more flexible "+", "!", and "!!" prefixes to ExecStart= and other
1887 * $DBUS_SESSION_BUS_ADDRESS environment variable is not set by
1888 pam_systemd anymore.
1890 * The naming scheme for network devices was changed to always rename
1891 devices, even if they were already renamed by userspace. The "kernel"
1892 policy was changed to only apply as a fallback, if no other naming
1895 * The requirements to build systemd is bumped to meson-0.46 and
1898 Contributions from: afg, Alan Jenkins, Aleksei Timofeyev, Alexander
1899 Filippov, Alexander Kurtz, Alexey Bogdanenko, Andreas Henriksson,
1900 Andrew Jorgensen, Anita Zhang, apnix-uk, Arkan49, Arseny Maslennikov,
1901 asavah, Asbjørn Apeland, aszlig, Bastien Nocera, Ben Boeckel, Benedikt
1902 Morbach, Benjamin Berg, Bruce Zhang, Carlo Caione, Cedric Viou, Chen
1903 Qi, Chris Chiu, Chris Down, Chris Morin, Christian Rebischke, Claudius
1904 Ellsel, Colin Guthrie, dana, Daniel, Daniele Medri, Daniel Kahn
1905 Gillmor, Daniel Rusek, Daniel van Vugt, Dariusz Gadomski, Dave Reisner,
1906 David Anderson, Davide Cavalca, David Leeds, David Malcolm, David
1907 Strauss, David Tardon, Dimitri John Ledkov, Dmitry Torokhov, dj-kaktus,
1908 Dongsu Park, Elias Probst, Emil Soleyman, Erik Kooistra, Ervin Peters,
1909 Evgeni Golov, Evgeny Vereshchagin, Fabrice Fontaine, Faheel Ahmad,
1910 Faizal Luthfi, Felix Yan, Filipe Brandenburger, Franck Bui, Frank
1911 Schaefer, Frantisek Sumsal, Gautier Husson, Gianluca Boiano, Giuseppe
1912 Scrivano, glitsj16, Hans de Goede, Harald Hoyer, Harry Mallon, Harshit
1913 Jain, Helmut Grohne, Henry Tung, Hui Yiqun, imayoda, Insun Pyo, Iwan
1914 Timmer, Jan Janssen, Jan Pokorný, Jan Synacek, Jason A. Donenfeld,
1915 javitoom, Jérémy Nouhaud, Jeremy Su, Jiuyang Liu, João Paulo Rechi
1916 Vita, Joe Hershberger, Joe Rayhawk, Joerg Behrmann, Joerg Steffens,
1917 Jonas Dorel, Jon Ringle, Josh Soref, Julian Andres Klode, Jun Bo Bi,
1918 Jürg Billeter, Keith Busch, Khem Raj, Kirill Marinushkin, Larry
1919 Bernstone, Lennart Poettering, Lion Yang, Li Song, Lorenz
1920 Hübschle-Schneider, Lubomir Rintel, Lucas Werkmeister, Ludwin Janvier,
1921 Lukáš Nykrýn, Luke Shumaker, mal, Marc-Antoine Perennou, Marcin
1922 Skarbek, Marco Trevisan (Treviño), Marian Cepok, Mario Hros, Marko
1923 Myllynen, Markus Grimm, Martin Pitt, Martin Sobotka, Martin Wilck,
1924 Mathieu Trudel-Lapierre, Matthew Leeds, Michael Biebl, Michael Olbrich,
1925 Michael 'pbone' Pobega, Michael Scherer, Michal Koutný, Michal
1926 Sekletar, Michal Soltys, Mike Gilbert, Mike Palmer, Muhammet Kara, Neal
1927 Gompa, Neil Brown, Network Silence, Niklas Tibbling, Nikolas Nyby,
1928 Nogisaka Sadata, Oliver Smith, Patrik Flykt, Pavel Hrdina, Paweł
1929 Szewczyk, Peter Hutterer, Piotr Drąg, Ray Strode, Reinhold Mueller,
1930 Renaud Métrich, Roman Gushchin, Ronny Chevalier, Rubén Suárez Alvarez,
1931 Ruixin Bao, RussianNeuroMancer, Ryutaroh Matsumoto, Saleem Rashid, Sam
1932 Morris, Samuel Morris, Sandy Carter, scootergrisen, Sébastien Bacher,
1933 Sergey Ptashnick, Shawn Landden, Shengyao Xue, Shih-Yuan Lee
1934 (FourDollars), Silvio Knizek, Sjoerd Simons, Stasiek Michalski, Stephen
1935 Gallagher, Steven Allen, Steve Ramage, Susant Sahani, Sven Joachim,
1936 Sylvain Plantefève, Tanu Kaskinen, Tejun Heo, Thiago Macieira, Thomas
1937 Blume, Thomas Haller, Thomas H. P. Andersen, Tim Ruffing, TJ, Tobias
1938 Jungel, Todd Walton, Tommi Rantala, Tomsod M, Tony Novak, Tore
1939 Anderson, Trevonn, Victor Laskurain, Victor Tapia, Violet Halo, Vojtech
1940 Trefny, welaq, William A. Kennington III, William Douglas, Wyatt Ward,
1941 Xiang Fan, Xi Ruoyao, Xuanwo, Yann E. Morin, YmrDtnJu, Yu Watanabe,
1942 Zbigniew Jędrzejewski-Szmek, Zhang Xianwei, Zsolt Dollenstein
1944 — Warsaw, 2018-12-21
1948 * NETWORK INTERFACE DEVICE NAMING CHANGES: systemd-udevd's "net_id"
1949 builtin will name network interfaces differently than in previous
1950 versions for virtual network interfaces created with SR-IOV and NPAR
1951 and for devices where the PCI network controller device does not have
1952 a slot number associated.
1954 SR-IOV virtual devices are now named based on the name of the parent
1955 interface, with a suffix of "v<N>", where <N> is the virtual device
1956 number. Previously those virtual devices were named as if completely
1959 The ninth and later NPAR virtual devices will be named following the
1960 scheme used for the first eight NPAR partitions. Previously those
1961 devices were not renamed and the kernel default (eth<n>) was used.
1963 "net_id" will also generate names for PCI devices where the PCI
1964 network controller device does not have an associated slot number
1965 itself, but one of its parents does. Previously those devices were
1966 not renamed and the kernel default (eth<n>) was used.
1968 * AF_INET and AF_INET6 are dropped from RestrictAddressFamilies= in
1969 systemd-logind.service. Since v235, IPAddressDeny=any has been set to
1970 the unit. So, it is expected that the default behavior of
1971 systemd-logind is not changed. However, if distribution packagers or
1972 administrators disabled or modified IPAddressDeny= setting by a
1973 drop-in config file, then it may be necessary to update the file to
1974 re-enable AF_INET and AF_INET6 to support network user name services,
1977 * When the RestrictNamespaces= unit property is specified multiple
1978 times, then the specified types are merged now. Previously, only the
1979 last assignment was used. So, if distribution packagers or
1980 administrators modified the setting by a drop-in config file, then it
1981 may be necessary to update the file.
1983 * When OnFailure= is used in combination with Restart= on a service
1984 unit, then the specified units will no longer be triggered on
1985 failures that result in restarting. Previously, the specified units
1986 would be activated each time the unit failed, even when the unit was
1987 going to be restarted automatically. This behaviour contradicted the
1988 documentation. With this release the code is adjusted to match the
1991 * systemd-tmpfiles will now print a notice whenever it encounters
1992 tmpfiles.d/ lines referencing the /var/run/ directory. It will
1993 recommend reworking them to use the /run/ directory instead (for
1994 which /var/run/ is simply a symlinked compatibility alias). This way
1995 systemd-tmpfiles can properly detect line conflicts and merge lines
1996 referencing the same file by two paths, without having to access
1999 * systemctl disable/unmask/preset/preset-all cannot be used with
2000 --runtime. Previously this was allowed, but resulted in unintuitive
2001 behaviour that wasn't useful. systemctl disable/unmask will now undo
2002 both runtime and persistent enablement/masking, i.e. it will remove
2003 any relevant symlinks both in /run and /etc.
2005 * Note that all long-running system services shipped with systemd will
2006 now default to a system call whitelist (rather than a blacklist, as
2007 before). In particular, systemd-udevd will now enforce one too. For
2008 most cases this should be safe, however downstream distributions
2009 which disabled sandboxing of systemd-udevd (specifically the
2010 MountFlags= setting), might want to disable this security feature
2011 too, as the default whitelisting will prohibit all mount, swap,
2012 reboot and clock changing operations from udev rules.
2014 * sd-boot acquired new loader configuration settings to optionally turn
2015 off Windows and MacOS boot partition discovery as well as
2016 reboot-into-firmware menu items. It is also able to pick a better
2017 screen resolution for HiDPI systems, and now provides loader
2018 configuration settings to change the resolution explicitly.
2020 * systemd-resolved now supports DNS-over-TLS. It's still
2021 turned off by default, use DNSOverTLS=opportunistic to turn it on in
2022 resolved.conf. We intend to make this the default as soon as couple
2023 of additional techniques for optimizing the initial latency caused by
2024 establishing a TLS/TCP connection are implemented.
2026 * systemd-resolved.service and systemd-networkd.service now set
2027 DynamicUser=yes. The users systemd-resolve and systemd-network are
2028 not created by systemd-sysusers anymore.
2030 NOTE: This has a chance of breaking nss-ldap and similar NSS modules
2031 that embed a network facing module into any process using getpwuid()
2032 or related call: the dynamic allocation of the user ID for
2033 systemd-resolved.service means the service manager has to check NSS
2034 if the user name is already taken when forking off the service. Since
2035 the user in the common case won't be defined in /etc/passwd the
2036 lookup is likely to trigger nss-ldap which in turn might use NSS to
2037 ask systemd-resolved for hostname lookups. This will hence result in
2038 a deadlock: a user name lookup in order to start
2039 systemd-resolved.service will result in a hostname lookup for which
2040 systemd-resolved.service needs to be started already. There are
2041 multiple ways to work around this problem: pre-allocate the
2042 "systemd-resolve" user on such systems, so that nss-ldap won't be
2043 triggered; or use a different NSS package that doesn't do networking
2044 in-process but provides a local asynchronous name cache; or configure
2045 the NSS package to avoid lookups for UIDs in the range `pkg-config
2046 systemd --variable=dynamicuidmin` … `pkg-config systemd
2047 --variable=dynamicuidmax`, so that it does not consider itself
2048 authoritative for the same UID range systemd allocates dynamic users
2051 * The systemd-resolve tool has been renamed to resolvectl (it also
2052 remains available under the old name, for compatibility), and its
2053 interface is now verb-based, similar in style to the other <xyz>ctl
2054 tools, such as systemctl or loginctl.
2056 * The resolvectl/systemd-resolve tool also provides 'resolvconf'
2057 compatibility. It may be symlinked under the 'resolvconf' name, in
2058 which case it will take arguments and input compatible with the
2059 Debian and FreeBSD resolvconf tool.
2061 * Support for suspend-then-hibernate has been added, i.e. a sleep mode
2062 where the system initially suspends, and after a timeout resumes and
2065 * networkd's ClientIdentifier= now accepts a new option "duid-only". If
2066 set the client will only send a DUID as client identifier.
2068 * The nss-systemd glibc NSS module will now enumerate dynamic users and
2069 groups in effect. Previously, it could resolve UIDs/GIDs to user
2070 names/groups and vice versa, but did not support enumeration.
2072 * journald's Compress= configuration setting now optionally accepts a
2073 byte threshold value. All journal objects larger than this threshold
2074 will be compressed, smaller ones will not. Previously this threshold
2075 was not configurable and set to 512.
2077 * A new system.conf setting NoNewPrivileges= is now available which may
2078 be used to turn off acquisition of new privileges system-wide
2079 (i.e. set Linux' PR_SET_NO_NEW_PRIVS for PID 1 itself, and thus also
2080 for all its children). Note that turning this option on means setuid
2081 binaries and file system capabilities lose their special powers.
2082 While turning on this option is a big step towards a more secure
2083 system, doing so is likely to break numerous pre-existing UNIX tools,
2084 in particular su and sudo.
2086 * A new service systemd-time-sync-wait.service has been added. If
2087 enabled it will delay the time-sync.target unit at boot until time
2088 synchronization has been received from the network. This
2089 functionality is useful on systems lacking a local RTC or where it is
2090 acceptable that the boot process shall be delayed by external network
2093 * When hibernating, systemd will now inform the kernel of the image
2094 write offset, on kernels new enough to support this. This means swap
2095 files should work for hibernation now.
2097 * When loading unit files, systemd will now look for drop-in unit files
2098 extensions in additional places. Previously, for a unit file name
2099 "foo-bar-baz.service" it would look for dropin files in
2100 "foo-bar-baz.service.d/*.conf". Now, it will also look in
2101 "foo-bar-.service.d/*.conf" and "foo-.service.d/", i.e. at the
2102 service name truncated after all inner dashes. This scheme allows
2103 writing drop-ins easily that apply to a whole set of unit files at
2104 once. It's particularly useful for mount and slice units (as their
2105 naming is prefix based), but is also useful for service and other
2106 units, for packages that install multiple unit files at once,
2107 following a strict naming regime of beginning the unit file name with
2108 the package's name. Two new specifiers are now supported in unit
2109 files to match this: %j and %J are replaced by the part of the unit
2110 name following the last dash.
2112 * Unit files and other configuration files that support specifier
2113 expansion now understand another three new specifiers: %T and %V will
2114 resolve to /tmp and /var/tmp respectively, or whatever temporary
2115 directory has been set for the calling user. %E will expand to either
2116 /etc (for system units) or $XDG_CONFIG_HOME (for user units).
2118 * The ExecStart= lines of unit files are no longer required to
2119 reference absolute paths. If non-absolute paths are specified the
2120 specified binary name is searched within the service manager's
2121 built-in $PATH, which may be queried with 'systemd-path
2122 search-binaries-default'. It's generally recommended to continue to
2123 use absolute paths for all binaries specified in unit files.
2125 * Units gained a new load state "bad-setting", which is used when a
2126 unit file was loaded, but contained fatal errors which prevent it
2127 from being started (for example, a service unit has been defined
2128 lacking both ExecStart= and ExecStop= lines).
2130 * coredumpctl's "gdb" verb has been renamed to "debug", in order to
2131 support alternative debuggers, for example lldb. The old name
2132 continues to be available however, for compatibility reasons. Use the
2133 new --debugger= switch or the $SYSTEMD_DEBUGGER environment variable
2134 to pick an alternative debugger instead of the default gdb.
2136 * systemctl and the other tools will now output escape sequences that
2137 generate proper clickable hyperlinks in various terminal emulators
2138 where useful (for example, in the "systemctl status" output you can
2139 now click on the unit file name to quickly open it in the
2140 editor/viewer of your choice). Note that not all terminal emulators
2141 support this functionality yet, but many do. Unfortunately, the
2142 "less" pager doesn't support this yet, hence this functionality is
2143 currently automatically turned off when a pager is started (which
2144 happens quite often due to auto-paging). We hope to remove this
2145 limitation as soon as "less" learns these escape sequences. This new
2146 behaviour may also be turned off explicitly with the $SYSTEMD_URLIFY
2147 environment variable. For details on these escape sequences see:
2148 https://gist.github.com/egmontkob/eb114294efbcd5adb1944c9f3cb5feda
2150 * networkd's .network files now support a new IPv6MTUBytes= option for
2151 setting the MTU used by IPv6 explicitly as well as a new MTUBytes=
2152 option in the [Route] section to configure the MTU to use for
2153 specific routes. It also gained support for configuration of the DHCP
2154 "UserClass" option through the new UserClass= setting. It gained
2155 three new options in the new [CAN] section for configuring CAN
2156 networks. The MULTICAST and ALLMULTI interface flags may now be
2157 controlled explicitly with the new Multicast= and AllMulticast=
2160 * networkd will now automatically make use of the kernel's route
2161 expiration feature, if it is available.
2163 * udevd's .link files now support setting the number of receive and
2164 transmit channels, using the RxChannels=, TxChannels=,
2165 OtherChannels=, CombinedChannels= settings.
2167 * Support for UDPSegmentationOffload= has been removed, given its
2168 limited support in hardware, and waning software support.
2170 * networkd's .netdev files now support creating "netdevsim" interfaces.
2172 * PID 1 learnt a new bus call GetUnitByControlGroup() which may be used
2173 to query the unit belonging to a specific kernel control group.
2175 * systemd-analyze gained a new verb "cat-config", which may be used to
2176 dump the contents of any configuration file, with all its matching
2177 drop-in files added in, and honouring the usual search and masking
2178 logic applied to systemd configuration files. For example use
2179 "systemd-analyze cat-config systemd/system.conf" to get the complete
2180 system configuration file of systemd how it would be loaded by PID 1
2181 itself. Similar to this, various tools such as systemd-tmpfiles or
2182 systemd-sysusers, gained a new option "--cat-config", which does the
2183 corresponding operation for their own configuration settings. For
2184 example, "systemd-tmpfiles --cat-config" will now output the full
2185 list of tmpfiles.d/ lines in place.
2187 * timedatectl gained three new verbs: "show" shows bus properties of
2188 systemd-timedated, "timesync-status" shows the current NTP
2189 synchronization state of systemd-timesyncd, and "show-timesync"
2190 shows bus properties of systemd-timesyncd.
2192 * systemd-timesyncd gained a bus interface on which it exposes details
2195 * A new environment variable $SYSTEMD_TIMEDATED_NTP_SERVICES is now
2196 understood by systemd-timedated. It takes a colon-separated list of
2197 unit names of NTP client services. The list is used by
2198 "timedatectl set-ntp".
2200 * systemd-nspawn gained a new --rlimit= switch for setting initial
2201 resource limits for the container payload. There's a new switch
2202 --hostname= to explicitly override the container's hostname. A new
2203 --no-new-privileges= switch may be used to control the
2204 PR_SET_NO_NEW_PRIVS flag for the container payload. A new
2205 --oom-score-adjust= switch controls the OOM scoring adjustment value
2206 for the payload. The new --cpu-affinity= switch controls the CPU
2207 affinity of the container payload. The new --resolv-conf= switch
2208 allows more detailed control of /etc/resolv.conf handling of the
2209 container. Similarly, the new --timezone= switch allows more detailed
2210 control of /etc/localtime handling of the container.
2212 * systemd-detect-virt gained a new --list switch, which will print a
2213 list of all currently known VM and container environments.
2215 * Support for "Portable Services" has been added, see
2216 doc/PORTABLE_SERVICES.md for details. Currently, the support is still
2217 experimental, but this is expected to change soon. Reflecting this
2218 experimental state, the "portablectl" binary is not installed into
2219 /usr/bin yet. The binary has to be called with the full path
2220 /usr/lib/systemd/portablectl instead.
2222 * journalctl's and systemctl's -o switch now knows a new log output
2223 mode "with-unit". The output it generates is very similar to the
2224 regular "short" mode, but displays the unit name instead of the
2225 syslog tag for each log line. Also, the date is shown with timezone
2226 information. This mode is probably more useful than the classic
2227 "short" output mode for most purposes, except where pixel-perfect
2228 compatibility with classic /var/log/messages formatting is required.
2230 * A new --dump-bus-properties switch has been added to the systemd
2231 binary, which may be used to dump all supported D-Bus properties.
2232 (Options which are still supported, but are deprecated, are *not*
2235 * sd-bus gained a set of new calls:
2236 sd_bus_slot_set_floating()/sd_bus_slot_get_floating() may be used to
2237 enable/disable the "floating" state of a bus slot object,
2238 i.e. whether the slot object pins the bus it is allocated for into
2239 memory or if the bus slot object gets disconnected when the bus goes
2240 away. sd_bus_open_with_description(),
2241 sd_bus_open_user_with_description(),
2242 sd_bus_open_system_with_description() may be used to allocate bus
2243 objects and set their description string already during allocation.
2245 * sd-event gained support for watching inotify events from the event
2246 loop, in an efficient way, sharing inotify handles between multiple
2247 users. For this a new function sd_event_add_inotify() has been added.
2249 * sd-event and sd-bus gained support for calling special user-supplied
2250 destructor functions for userdata pointers associated with
2251 sd_event_source, sd_bus_slot, and sd_bus_track objects. For this new
2252 functions sd_bus_slot_set_destroy_callback,
2253 sd_bus_slot_get_destroy_callback, sd_bus_track_set_destroy_callback,
2254 sd_bus_track_get_destroy_callback,
2255 sd_event_source_set_destroy_callback,
2256 sd_event_source_get_destroy_callback have been added.
2258 * The "net.ipv4.tcp_ecn" sysctl will now be turned on by default.
2260 * PID 1 will now automatically reschedule .timer units whenever the
2261 local timezone changes. (They previously got rescheduled
2262 automatically when the system clock changed.)
2264 * New documentation has been added to document cgroups delegation,
2265 portable services and the various code quality tools we have set up:
2267 https://github.com/systemd/systemd/blob/master/docs/CGROUP_DELEGATION.md
2268 https://github.com/systemd/systemd/blob/master/docs/PORTABLE_SERVICES.md
2269 https://github.com/systemd/systemd/blob/master/docs/CODE_QUALITY.md
2271 * The Boot Loader Specification has been added to the source tree.
2273 https://github.com/systemd/systemd/blob/master/docs/BOOT_LOADER_SPECIFICATION.md
2275 While moving it into our source tree we have updated it and further
2276 changes are now accepted through the usual github PR workflow.
2278 * pam_systemd will now look for PAM userdata fields systemd.memory_max,
2279 systemd.tasks_max, systemd.cpu_weight, systemd.io_weight set by
2280 earlier PAM modules. The data in these fields is used to initialize
2281 the session scope's resource properties. Thus external PAM modules
2282 may now configure per-session limits, for example sourced from
2283 external user databases.
2285 * socket units with Accept=yes will now maintain a "refused" counter in
2286 addition to the existing "accepted" counter, counting connections
2287 refused due to the enforced limits.
2289 * The "systemd-path search-binaries-default" command may now be use to
2290 query the default, built-in $PATH PID 1 will pass to the services it
2293 * A new unit file setting PrivateMounts= has been added. It's a boolean
2294 option. If enabled the unit's processes are invoked in their own file
2295 system namespace. Note that this behaviour is also implied if any
2296 other file system namespacing options (such as PrivateTmp=,
2297 PrivateDevices=, ProtectSystem=, …) are used. This option is hence
2298 primarily useful for services that do not use any of the other file
2299 system namespacing options. One such service is systemd-udevd.service
2300 where this is now used by default.
2302 * ConditionSecurity= gained a new value "uefi-secureboot" that is true
2303 when the system is booted in UEFI "secure mode".
2305 * A new unit "system-update-pre.target" is added, which defines an
2306 optional synchronization point for offline system updates, as
2307 implemented by the pre-existing "system-update.target" unit. It
2308 allows ordering services before the service that executes the actual
2309 update process in a generic way.
2311 * Systemd now emits warnings whenever .include syntax is used.
2313 Contributions from: Adam Duskett, Alan Jenkins, Alessandro Casale,
2314 Alexander Kurtz, Alex Gartrell, Anssi Hannula, Arnaud Rebillout, Brian
2315 J. Murrell, Bruno Vernay, Chris Lamb, Chris Lesiak, Christian Brauner,
2316 Christian Hesse, Christian Rebischke, Colin Guthrie, Daniel Dao, Daniel
2317 Lin, Danylo Korostil, Davide Cavalca, David Tardon, Dimitri John
2318 Ledkov, Dmitriy Geels, Douglas Christman, Elia Geretto, emelenas, Emil
2319 Velikov, Evgeny Vereshchagin, Felipe Sateler, Feng Sun, Filipe
2320 Brandenburger, Franck Bui, futpib, Giuseppe Scrivano, Guillem Jover,
2321 guixxx, Hannes Reinecke, Hans de Goede, Harald Hoyer, Henrique Dante de
2322 Almeida, Hiram van Paassen, Ian Miell, Igor Gnatenko, Ivan Shapovalov,
2323 Iwan Timmer, James Cowgill, Jan Janssen, Jan Synacek, Jared Kazimir,
2324 Jérémy Rosen, João Paulo Rechi Vita, Joost Heitbrink, Jui-Chi Ricky
2325 Liang, Jürg Billeter, Kai-Heng Feng, Karol Augustin, Kay Sievers,
2326 Krzysztof Nowicki, Lauri Tirkkonen, Lennart Poettering, Leonard König,
2327 Long Li, Luca Boccassi, Lucas Werkmeister, Marcel Hoppe, Marc
2328 Kleine-Budde, Mario Limonciello, Martin Jansa, Martin Wilck, Mathieu
2329 Malaterre, Matteo F. Vescovi, Matthew McGinn, Matthias-Christian Ott,
2330 Michael Biebl, Michael Olbrich, Michael Prokop, Michal Koutný, Michal
2331 Sekletar, Mike Gilbert, Mikhail Kasimov, Milan Broz, Milan Pässler,
2332 Mladen Pejaković, Muhammet Kara, Nicolas Boichat, Omer Katz, Paride
2333 Legovini, Paul Menzel, Paul Milliken, Pavel Hrdina, Peter A. Bigot,
2334 Peter D'Hoye, Peter Hutterer, Peter Jones, Philip Sequeira, Philip
2335 Withnall, Piotr Drąg, Radostin Stoyanov, Ricardo Salveti de Araujo,
2336 Ronny Chevalier, Rosen Penev, Rubén Suárez Alvarez, Ryan Gonzalez,
2337 Salvo Tomaselli, Sebastian Reichel, Sergey Ptashnick, Sergio Lindo
2338 Mansilla, Stefan Schweter, Stephen Hemminger, Stuart Hayes, Susant
2339 Sahani, Sylvain Plantefève, Thomas H. P. Andersen, Tobias Jungel,
2340 Tomasz Torcz, Vito Caputo, Will Dietz, Will Thompson, Wim van Mourik,
2341 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
2343 — Berlin, 2018-06-22
2347 * The MemoryAccounting= unit property now defaults to on. After
2348 discussions with the upstream control group maintainers we learnt
2349 that the negative impact of cgroup memory accounting on current
2350 kernels is finally relatively minimal, so that it should be safe to
2351 enable this by default without affecting system performance. Besides
2352 memory accounting only task accounting is turned on by default, all
2353 other forms of resource accounting (CPU, IO, IP) remain off for now,
2354 because it's not clear yet that their impact is small enough to move
2355 from opt-in to opt-out. We recommend downstreams to leave memory
2356 accounting on by default if kernel 4.14 or higher is primarily
2357 used. On very resource constrained systems or when support for old
2358 kernels is a necessity, -Dmemory-accounting-default=false can be used
2359 to revert this change.
2361 * rpm scriptlets to update the udev hwdb and rules (%udev_hwdb_update,
2362 %udev_rules_update) and the journal catalog (%journal_catalog_update)
2363 from the upgrade scriptlets of individual packages now do nothing.
2364 Transfiletriggers have been added which will perform those updates
2365 once at the end of the transaction.
2367 Similar transfiletriggers have been added to execute any sysctl.d
2368 and binfmt.d rules. Thus, it should be unnecessary to provide any
2369 scriptlets to execute this configuration from package installation
2372 * systemd-sysusers gained a mode where the configuration to execute is
2373 specified on the command line, but this configuration is not executed
2374 directly, but instead it is merged with the configuration on disk,
2375 and the result is executed. This is useful for package installation
2376 scripts which want to create the user before installing any files on
2377 disk (in case some of those files are owned by that user), while
2378 still allowing local admin overrides.
2380 This functionality is exposed to rpm scriptlets through a new
2381 %sysusers_create_package macro. Old %sysusers_create and
2382 %sysusers_create_inline macros are deprecated.
2384 A transfiletrigger for sysusers.d configuration is now installed,
2385 which means that it should be unnecessary to call systemd-sysusers from
2386 package installation scripts, unless the package installs any files
2387 owned by those newly-created users, in which case
2388 %sysusers_create_package should be used.
2390 * Analogous change has been done for systemd-tmpfiles: it gained a mode
2391 where the command-line configuration is merged with the configuration
2392 on disk. This is exposed as the new %tmpfiles_create_package macro,
2393 and %tmpfiles_create is deprecated. A transfiletrigger is installed
2394 for tmpfiles.d, hence it should be unnecessary to call systemd-tmpfiles
2395 from package installation scripts.
2397 * sysusers.d configuration for a user may now also specify the group
2398 number, in addition to the user number ("u username 123:456"), or
2399 without the user number ("u username -:456").
2401 * Configution items for systemd-sysusers can now be specified as
2402 positional arguments when the new --inline switch is used.
2404 * The login shell of users created through sysusers.d may now be
2405 specified (previously, it was always /bin/sh for root and
2406 /sbin/nologin for other users).
2408 * systemd-analyze gained a new --global switch to look at global user
2409 configuration. It also gained a unit-paths verb to list the unit load
2410 paths that are compiled into systemd (which can be used with
2411 --systemd, --user, or --global).
2413 * udevadm trigger gained a new --settle/-w option to wait for any
2414 triggered events to finish (but just those, and not any other events
2415 which are triggered meanwhile).
2417 * The action that systemd-logind takes when the lid is closed and the
2418 machine is connected to external power can now be configured using
2419 HandleLidSwitchExternalPower= in logind.conf. Previously, this action
2420 was determined by HandleLidSwitch=, and, for backwards compatibility,
2421 is still is, if HandleLidSwitchExternalPower= is not explicitly set.
2423 * journalctl will periodically call sd_journal_process() to make it
2424 resilient against inotify queue overruns when journal files are
2425 rotated very quickly.
2427 * Two new functions in libsystemd — sd_bus_get_n_queued_read and
2428 sd_bus_get_n_queued_write — may be used to check the number of
2429 pending bus messages.
2431 * systemd gained a new
2432 org.freedesktop.systemd1.Manager.AttachProcessesToUnit dbus call
2433 which can be used to migrate foreign processes to scope and service
2434 units. The primary user for this new API is systemd itself: the
2435 systemd --user instance uses this call of the systemd --system
2436 instance to migrate processes if it itself gets the request to
2437 migrate processes and the kernel refuses this due to access
2438 restrictions. Thanks to this "systemd-run --scope --user …" works
2439 again in pure cgroup v2 environments when invoked from the user
2442 * A new TemporaryFileSystem= setting can be used to mask out part of
2443 the real file system tree with tmpfs mounts. This may be combined
2444 with BindPaths= and BindReadOnlyPaths= to hide files or directories
2445 not relevant to the unit, while still allowing some paths lower in
2446 the tree to be accessed.
2448 ProtectHome=tmpfs may now be used to hide user home and runtime
2449 directories from units, in a way that is mostly equivalent to
2450 "TemporaryFileSystem=/home /run/user /root".
2452 * Non-service units are now started with KeyringMode=shared by default.
2453 This means that mount and swapon and other mount tools have access
2454 to keys in the main keyring.
2456 * /sys/fs/bpf is now mounted automatically.
2458 * QNX virtualization is now detected by systemd-detect-virt and may
2459 be used in ConditionVirtualization=.
2461 * IPAccounting= may now be enabled also for slice units.
2463 * A new -Dsplit-bin= build configuration switch may be used to specify
2464 whether bin and sbin directories are merged, or if they should be
2465 included separately in $PATH and various listings of executable
2466 directories. The build configuration scripts will try to autodetect
2467 the proper values of -Dsplit-usr= and -Dsplit-bin= based on build
2468 system, but distributions are encouraged to configure this
2471 * A new -Dok-color= build configuration switch may be used to change
2472 the colour of "OK" status messages.
2474 * UPGRADE ISSUE: serialization of units using JoinsNamespaceOf= with
2475 PrivateNetwork=yes was buggy in previous versions of systemd. This
2476 means that after the upgrade and daemon-reexec, any such units must
2479 * INCOMPATIBILITY: as announced in the NEWS for 237, systemd-tmpfiles
2480 will not exclude read-only files owned by root from cleanup.
2482 Contributions from: Alan Jenkins, Alexander F Rødseth, Alexis Jeandet,
2483 Andika Triwidada, Andrei Gherzan, Ansgar Burchardt, antizealot1337,
2484 Batuhan Osman Taşkaya, Beniamino Galvani, Bill Yodlowsky, Caio Marcelo
2485 de Oliveira Filho, CuBiC, Daniele Medri, Daniel Mouritzen, Daniel
2486 Rusek, Davide Cavalca, Dimitri John Ledkov, Douglas Christman, Evgeny
2487 Vereshchagin, Faalagorn, Filipe Brandenburger, Franck Bui, futpib,
2488 Giacomo Longo, Gunnar Hjalmarsson, Hans de Goede, Hermann Gausterer,
2489 Iago López Galeiras, Jakub Filak, Jan Synacek, Jason A. Donenfeld,
2490 Javier Martinez Canillas, Jérémy Rosen, Lennart Poettering, Lucas
2491 Werkmeister, Mao Huang, Marco Gulino, Michael Biebl, Michael Vogt,
2492 MilhouseVH, Neal Gompa (ニール・ゴンパ), Oleander Reis, Olof Mogren,
2493 Patrick Uiterwijk, Peter Hutterer, Peter Portante, Piotr Drąg, Robert
2494 Antoni Buj Gelonch, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
2495 Fowler, SjonHortensius, snorreflorre, Susant Sahani, Sylvain
2496 Plantefève, Thomas Blume, Thomas Haller, Vito Caputo, Yu Watanabe,
2497 Zbigniew Jędrzejewski-Szmek, Марко М. Костић (Marko M. Kostić)
2499 — Warsaw, 2018-03-05
2503 * Some keyboards come with a zoom see-saw or rocker which until now got
2504 mapped to the Linux "zoomin/out" keys in hwdb. However, these
2505 keycodes are not recognized by any major desktop. They now produce
2506 Up/Down key events so that they can be used for scrolling.
2508 * INCOMPATIBILITY: systemd-tmpfiles' "f" lines changed behaviour
2509 slightly: previously, if an argument was specified for lines of this
2510 type (i.e. the right-most column was set) this string was appended to
2511 existing files each time systemd-tmpfiles was run. This behaviour was
2512 different from what the documentation said, and not particularly
2513 useful, as repeated systemd-tmpfiles invocations would not be
2514 idempotent and grow such files without bounds. With this release
2515 behaviour has been altered to match what the documentation says:
2516 lines of this type only have an effect if the indicated files don't
2517 exist yet, and only then the argument string is written to the file.
2519 * FUTURE INCOMPATIBILITY: In systemd v238 we intend to slightly change
2520 systemd-tmpfiles behaviour: previously, read-only files owned by root
2521 were always excluded from the file "aging" algorithm (i.e. the
2522 automatic clean-up of directories like /tmp based on
2523 atime/mtime/ctime). We intend to drop this restriction, and age files
2524 by default even when owned by root and read-only. This behaviour was
2525 inherited from older tools, but there have been requests to remove
2526 it, and it's not obvious why this restriction was made in the first
2527 place. Please speak up now, if you are aware of software that reqires
2528 this behaviour, otherwise we'll remove the restriction in v238.
2530 * A new environment variable $SYSTEMD_OFFLINE is now understood by
2531 systemctl. It takes a boolean argument. If on, systemctl assumes it
2532 operates on an "offline" OS tree, and will not attempt to talk to the
2533 service manager. Previously, this mode was implicitly enabled if a
2534 chroot() environment was detected, and this new environment variable
2535 now provides explicit control.
2537 * .path and .socket units may now be created transiently, too.
2538 Previously only service, mount, automount and timer units were
2539 supported as transient units. The systemd-run tool has been updated
2540 to expose this new functionality, you may hence use it now to bind
2541 arbitrary commands to path or socket activation on-the-fly from the
2542 command line. Moreover, almost all properties are now exposed for the
2543 unit types that already supported transient operation.
2545 * The systemd-mount command gained support for a new --owner= parameter
2546 which takes a user name, which is then resolved and included in uid=
2547 and gid= mount options string of the file system to mount.
2549 * A new unit condition ConditionControlGroupController= has been added
2550 that checks whether a specific cgroup controller is available.
2552 * Unit files, udev's .link files, and systemd-networkd's .netdev and
2553 .network files all gained support for a new condition
2554 ConditionKernelVersion= for checking against specific kernel
2557 * In systemd-networkd, the [IPVLAN] section in .netdev files gained
2558 support for configuring device flags in the Flags= setting. In the
2559 same files, the [Tunnel] section gained support for configuring
2560 AllowLocalRemote=. The [Route] section in .network files gained
2561 support for configuring InitialCongestionWindow=,
2562 InitialAdvertisedReceiveWindow= and QuickAck=. The [DHCP] section now
2563 understands RapidCommit=.
2565 * systemd-networkd's DHCPv6 support gained support for Prefix
2568 * sd-bus gained support for a new "watch-bind" feature. When this
2569 feature is enabled, an sd_bus connection may be set up to connect to
2570 an AF_UNIX socket in the file system as soon as it is created. This
2571 functionality is useful for writing early-boot services that
2572 automatically connect to the system bus as soon as it is started,
2573 without ugly time-based polling. systemd-networkd and
2574 systemd-resolved have been updated to make use of this
2575 functionality. busctl exposes this functionality in a new
2576 --watch-bind= command line switch.
2578 * sd-bus will now optionally synthesize a local "Connected" signal as
2579 soon as a D-Bus connection is set up fully. This message mirrors the
2580 already existing "Disconnected" signal which is synthesized when the
2581 connection is terminated. This signal is generally useful but
2582 particularly handy in combination with the "watch-bind" feature
2583 described above. Synthesizing of this message has to be requested
2584 explicitly through the new API call sd_bus_set_connected_signal(). In
2585 addition a new call sd_bus_is_ready() has been added that checks
2586 whether a connection is fully set up (i.e. between the "Connected" and
2587 "Disconnected" signals).
2589 * sd-bus gained two new calls sd_bus_request_name_async() and
2590 sd_bus_release_name_async() for asynchronously registering bus
2591 names. Similar, there is now sd_bus_add_match_async() for installing
2592 a signal match asynchronously. All of systemd's own services have
2593 been updated to make use of these calls. Doing these operations
2594 asynchronously has two benefits: it reduces the risk of deadlocks in
2595 case of cyclic dependencies between bus services, and it speeds up
2596 service initialization since synchronization points for bus
2597 round-trips are removed.
2599 * sd-bus gained two new calls sd_bus_match_signal() and
2600 sd_bus_match_signal_async(), which are similar to sd_bus_add_match()
2601 and sd_bus_add_match_async() but instead of taking a D-Bus match
2602 string take match fields as normal function parameters.
2604 * sd-bus gained two new calls sd_bus_set_sender() and
2605 sd_bus_message_set_sender() for setting the sender name of outgoing
2606 messages (either for all outgoing messages or for just one specific
2607 one). These calls are only useful in direct connections as on
2608 brokered connections the broker fills in the sender anyway,
2609 overwriting whatever the client filled in.
2611 * sd-event gained a new pseudo-handle that may be specified on all API
2612 calls where an "sd_event*" object is expected: SD_EVENT_DEFAULT. When
2613 used this refers to the default event loop object of the calling
2614 thread. Note however that this does not implicitly allocate one —
2615 which has to be done prior by using sd_event_default(). Similarly
2616 sd-bus gained three new pseudo-handles SD_BUS_DEFAULT,
2617 SD_BUS_DEFAULT_USER, SD_BUS_DEFAULT_SYSTEM that may be used to refer
2618 to the default bus of the specified type of the calling thread. Here
2619 too this does not implicitly allocate bus connection objects, this
2620 has to be done prior with sd_bus_default() and friends.
2622 * sd-event gained a new call pair
2623 sd_event_source_{get|set}_io_fd_own(). This may be used to request
2624 automatic closure of the file descriptor an IO event source watches
2625 when the event source is destroyed.
2627 * systemd-networkd gained support for natively configuring WireGuard
2630 * In previous versions systemd synthesized user records both for the
2631 "nobody" (UID 65534) and "root" (UID 0) users in nss-systemd and
2632 internally. In order to simplify distribution-wide renames of the
2633 "nobody" user (like it is planned in Fedora: nfsnobody → nobody), a
2634 new transitional flag file has been added: if
2635 /etc/systemd/dont-synthesize-nobody exists synthesizing of the 65534
2636 user and group record within the systemd codebase is disabled.
2638 * systemd-notify gained a new --uid= option for selecting the source
2639 user/UID to use for notification messages sent to the service
2642 * journalctl gained a new --grep= option to list only entries in which
2643 the message matches a certain pattern. By default matching is case
2644 insensitive if the pattern is lowercase, and case sensitive
2645 otherwise. Option --case-sensitive=yes|no can be used to override
2646 this an specify case sensitivity or case insensitivity.
2648 * There's now a "systemd-analyze service-watchdogs" command for printing
2649 the current state of the service runtime watchdog, and optionally
2650 enabling or disabling the per-service watchdogs system-wide if given a
2651 boolean argument (i.e. the concept you configure in WatchdogSec=), for
2652 debugging purposes. There's also a kernel command line option
2653 systemd.service_watchdogs= for controlling the same.
2655 * Two new "log-level" and "log-target" options for systemd-analyze were
2656 added that merge the now deprecated get-log-level, set-log-level and
2657 get-log-target, set-log-target pairs. The deprecated options are still
2658 understood for backwards compatibility. The two new options print the
2659 current value when no arguments are given, and set them when a
2660 level/target is given as an argument.
2662 * sysusers.d's "u" lines now optionally accept both a UID and a GID
2663 specification, separated by a ":" character, in order to create users
2664 where UID and GID do not match.
2666 Contributions from: Adam Duskett, Alan Jenkins, Alexander Kuleshov,
2667 Alexis Deruelle, Andrew Jeddeloh, Armin Widegreen, Batuhan Osman
2668 Taşkaya, Björn Esser, bleep_blop, Bruce A. Johnson, Chris Down, Clinton
2669 Roy, Colin Walters, Daniel Rusek, Dimitri John Ledkov, Dmitry Rozhkov,
2670 Evgeny Vereshchagin, Ewout van Mansom, Felipe Sateler, Franck Bui,
2671 Frantisek Sumsal, George Gaydarov, Gianluca Boiano, Hans-Christian
2672 Noren Egtvedt, Hans de Goede, Henrik Grindal Bakken, Jan Alexander
2673 Steffens, Jan Klötzke, Jason A. Donenfeld, jdkbx, Jérémy Rosen,
2674 Jerónimo Borque, John Lin, John Paul Herold, Jonathan Rudenberg, Jörg
2675 Thalheim, Ken (Bitsko) MacLeod, Larry Bernstone, Lennart Poettering,
2676 Lucas Werkmeister, Maciej S. Szmigiero, Marek Čermák, Martin Pitt,
2677 Mathieu Malaterre, Matthew Thode, Matthias-Christian Ott, Max Harmathy,
2678 Michael Biebl, Michael Vogt, Michal Koutný, Michal Sekletar, Michał
2679 Szczepański, Mike Gilbert, Nathaniel McCallum, Nicolas Chauvet, Olaf
2680 Hering, Olivier Schwander, Patrik Flykt, Paul Cercueil, Peter Hutterer,
2681 Piotr Drąg, Raphael Vogelgsang, Reverend Homer, Robert Kolchmeyer,
2682 Samuel Dionne-Riel, Sergey Ptashnick, Shawn Landden, Susant Sahani,
2683 Sylvain Plantefève, Thomas H. P. Andersen, Thomas Huth, Tomasz
2684 Bachorski, Vladislav Vishnyakov, Wieland Hoffmann, Yu Watanabe, Zachary
2685 Winnerman, Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски, Дилян
2692 * The modprobe.d/ drop-in for the bonding.ko kernel module introduced
2693 in v235 has been extended to also set the dummy.ko module option
2694 numdummies=0, preventing the kernel from automatically creating
2695 dummy0. All dummy interfaces must now be explicitly created.
2697 * Unknown '%' specifiers in configuration files are now rejected. This
2698 applies to units and tmpfiles.d configuration. Any percent characters
2699 that are followed by a letter or digit that are not supposed to be
2700 interpreted as the beginning of a specifier should be escaped by
2701 doubling ("%%"). (So "size=5%" is still accepted, as well as
2702 "size=5%,foo=bar", but not "LABEL=x%y%z" since %y and %z are not
2703 valid specifiers today.)
2705 * systemd-resolved now maintains a new dynamic
2706 /run/systemd/resolve/stub-resolv.conf compatibility file. It is
2707 recommended to make /etc/resolv.conf a symlink to it. This file
2708 points at the systemd-resolved stub DNS 127.0.0.53 resolver and
2709 includes dynamically acquired search domains, achieving more correct
2710 DNS resolution by software that bypasses local DNS APIs such as NSS.
2712 * The "uaccess" udev tag has been dropped from /dev/kvm and
2713 /dev/dri/renderD*. These devices now have the 0666 permissions by
2714 default (but this may be changed at build-time). /dev/dri/renderD*
2715 will now be owned by the "render" group along with /dev/kfd.
2717 * "DynamicUser=yes" has been enabled for systemd-timesyncd.service,
2718 systemd-journal-gatewayd.service and
2719 systemd-journal-upload.service. This means "nss-systemd" must be
2720 enabled in /etc/nsswitch.conf to ensure the UIDs assigned to these
2721 services are resolved properly.
2723 * In /etc/fstab two new mount options are now understood:
2724 x-systemd.makefs and x-systemd.growfs. The former has the effect that
2725 the configured file system is formatted before it is mounted, the
2726 latter that the file system is resized to the full block device size
2727 after it is mounted (i.e. if the file system is smaller than the
2728 partition it resides on, it's grown). This is similar to the fsck
2729 logic in /etc/fstab, and pulls in systemd-makefs@.service and
2730 systemd-growfs@.service as necessary, similar to
2731 systemd-fsck@.service. Resizing is currently only supported on ext4
2734 * In systemd-networkd, the IPv6 RA logic now optionally may announce
2735 DNS server and domain information.
2737 * Support for the LUKS2 on-disk format for encrypted partitions has
2738 been added. This requires libcryptsetup2 during compilation and
2741 * The systemd --user instance will now signal "readiness" when its
2742 basic.target unit has been reached, instead of when the run queue ran
2743 empty for the first time.
2745 * Tmpfiles.d with user configuration are now also supported.
2746 systemd-tmpfiles gained a new --user switch, and snippets placed in
2747 ~/.config/user-tmpfiles.d/ and corresponding directories will be
2748 executed by systemd-tmpfiles --user running in the new
2749 systemd-tmpfiles-setup.service and systemd-tmpfiles-clean.service
2750 running in the user session.
2752 * Unit files and tmpfiles.d snippets learnt three new % specifiers:
2753 %S resolves to the top-level state directory (/var/lib for the system
2754 instance, $XDG_CONFIG_HOME for the user instance), %C resolves to the
2755 top-level cache directory (/var/cache for the system instance,
2756 $XDG_CACHE_HOME for the user instance), %L resolves to the top-level
2757 logs directory (/var/log for the system instance,
2758 $XDG_CONFIG_HOME/log/ for the user instance). This matches the
2759 existing %t specifier, that resolves to the top-level runtime
2760 directory (/run for the system instance, and $XDG_RUNTIME_DIR for the
2763 * journalctl learnt a new parameter --output-fields= for limiting the
2764 set of journal fields to output in verbose and JSON output modes.
2766 * systemd-timesyncd's configuration file gained a new option
2767 RootDistanceMaxSec= for setting the maximum root distance of servers
2768 it'll use, as well as the new options PollIntervalMinSec= and
2769 PollIntervalMaxSec= to tweak the minimum and maximum poll interval.
2771 * bootctl gained a new command "list" for listing all available boot
2772 menu items on systems that follow the boot loader specification.
2774 * systemctl gained a new --dry-run switch that shows what would be done
2775 instead of doing it, and is currently supported by the shutdown and
2778 * ConditionSecurity= can now detect the TOMOYO security module.
2780 * Unit file [Install] sections are now also respected in unit drop-in
2781 files. This is intended to be used by drop-ins under /usr/lib/.
2783 * systemd-firstboot may now also set the initial keyboard mapping.
2785 * Udev "changed" events for devices which are exposed as systemd
2786 .device units are now propagated to units specified in
2787 ReloadPropagatedFrom= as reload requests.
2789 * If a udev device has a SYSTEMD_WANTS= property containing a systemd
2790 unit template name (i.e. a name in the form of 'foobar@.service',
2791 without the instance component between the '@' and - the '.'), then
2792 the escaped sysfs path of the device is automatically used as the
2795 * SystemCallFilter= in unit files has been extended so that an "errno"
2796 can be specified individually for each system call. Example:
2797 SystemCallFilter=~uname:EILSEQ.
2799 * The cgroup delegation logic has been substantially updated. Delegate=
2800 now optionally takes a list of controllers (instead of a boolean, as
2801 before), which lists the controllers to delegate at least.
2803 * The networkd DHCPv6 client now implements the FQDN option (RFC 4704).
2805 * A new LogLevelMax= setting configures the maximum log level any
2806 process of the service may log at (i.e. anything with a lesser
2807 priority than what is specified is automatically dropped). A new
2808 LogExtraFields= setting allows configuration of additional journal
2809 fields to attach to all log records generated by any of the unit's
2812 * New StandardInputData= and StandardInputText= settings along with the
2813 new option StandardInput=data may be used to configure textual or
2814 binary data that shall be passed to the executed service process via
2815 standard input, encoded in-line in the unit file.
2817 * StandardInput=, StandardOutput= and StandardError= may now be used to
2818 connect stdin/stdout/stderr of executed processes directly with a
2819 file or AF_UNIX socket in the file system, using the new "file:" option.
2821 * A new unit file option CollectMode= has been added, that allows
2822 tweaking the garbage collection logic for units. It may be used to
2823 tell systemd to garbage collect units that have failed automatically
2824 (normally it only GCs units that exited successfully). systemd-run
2825 and systemd-mount expose this new functionality with a new -G option.
2827 * "machinectl bind" may now be used to bind mount non-directories
2828 (i.e. regularfiles, devices, fifos, sockets).
2830 * systemd-analyze gained a new verb "calendar" for validating and
2831 testing calendar time specifications to use for OnCalendar= in timer
2832 units. Besides validating the expression it will calculate the next
2833 time the specified expression would elapse.
2835 * In addition to the pre-existing FailureAction= unit file setting
2836 there's now SuccessAction=, for configuring a shutdown action to
2837 execute when a unit completes successfully. This is useful in
2838 particular inside containers that shall terminate after some workload
2839 has been completed. Also, both options are now supported for all unit
2840 types, not just services.
2842 * networkds's IP rule support gained two new options
2843 IncomingInterface= and OutgoingInterface= for configuring the incoming
2844 and outgoing interfaces of configured rules. systemd-networkd also
2845 gained support for "vxcan" network devices.
2847 * networkd gained a new setting RequiredForOnline=, taking a
2848 boolean. If set, systemd-wait-online will take it into consideration
2849 when determining that the system is up, otherwise it will ignore the
2850 interface for this purpose.
2852 * The sd_notify() protocol gained support for a new operation: with
2853 FDSTOREREMOVE=1 file descriptors may be removed from the per-service
2854 store again, ahead of POLLHUP or POLLERR when they are removed
2857 * A new document doc/UIDS-GIDS.md has been added to the source tree,
2858 that documents the UID/GID range and assignment assumptions and
2859 requirements of systemd.
2861 * The watchdog device PID 1 will ping may now be configured through the
2862 WatchdogDevice= configuration file setting, or by setting the
2863 systemd.watchdog_service= kernel commandline option.
2865 * systemd-resolved's gained support for registering DNS-SD services on
2866 the local network using MulticastDNS. Services may either be
2867 registered by dropping in a .dnssd file in /etc/systemd/dnssd/ (or
2868 the same dir below /run, /usr/lib), or through its D-Bus API.
2870 * The sd_notify() protocol can now with EXTEND_TIMEOUT_USEC=microsecond
2871 extend the effective start, runtime, and stop time. The service must
2872 continue to send EXTEND_TIMEOUT_USEC within the period specified to
2873 prevent the service manager from making the service as timedout.
2875 * systemd-resolved's DNSSEC support gained support for RFC 8080
2876 (Ed25519 keys and signatures).
2878 * The systemd-resolve command line tool gained a new set of options
2879 --set-dns=, --set-domain=, --set-llmnr=, --set-mdns=, --set-dnssec=,
2880 --set-nta= and --revert to configure per-interface DNS configuration
2881 dynamically during runtime. It's useful for pushing DNS information
2882 into systemd-resolved from DNS hook scripts that various interface
2883 managing software supports (such as pppd).
2885 * systemd-nspawn gained a new --network-namespace-path= command line
2886 option, which may be used to make a container join an existing
2887 network namespace, by specifying a path to a "netns" file.
2889 Contributions from: Alan Jenkins, Alan Robertson, Alessandro Ghedini,
2890 Andrew Jeddeloh, Antonio Rojas, Ari, asavah, bleep_blop, Carsten
2891 Strotmann, Christian Brauner, Christian Hesse, Clinton Roy, Collin
2892 Eggert, Cong Wang, Daniel Black, Daniel Lockyer, Daniel Rusek, Dimitri
2893 John Ledkov, Dmitry Rozhkov, Dongsu Park, Edward A. James, Evgeny
2894 Vereshchagin, Florian Klink, Franck Bui, Gwendal Grignou, Hans de
2895 Goede, Harald Hoyer, Hristo Venev, Iago López Galeiras, Ikey Doherty,
2896 Jakub Wilk, Jérémy Rosen, Jiahui Xie, John Lin, José Bollo, Josef
2897 Andersson, juga0, Krzysztof Nowicki, Kyle Walker, Lars Karlitski, Lars
2898 Kellogg-Stedman, Lauri Tirkkonen, Lennart Poettering, Lubomir Rintel,
2899 Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn, Lukáš Říha, Lukasz
2900 Rubaszewski, Maciej S. Szmigiero, Mantas Mikulėnas, Marcus Folkesson,
2901 Martin Steuer, Mathieu Trudel-Lapierre, Matija Skala,
2902 Matthias-Christian Ott, Max Resch, Michael Biebl, Michael Vogt, Michal
2903 Koutný, Michal Sekletar, Mike Gilbert, Muhammet Kara, Neil Brown, Olaf
2904 Hering, Ondrej Kozina, Patrik Flykt, Patryk Kocielnik, Peter Hutterer,
2905 Piotr Drąg, Razvan Cojocaru, Robin McCorkell, Roland Hieber, Saran
2906 Tunyasuvunakool, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
2907 Arlott, Simon Peeters, Stanislav Angelovič, Stefan Agner, Susant
2908 Sahani, Sylvain Plantefève, Thomas Blume, Thomas Haller, Tiago Salem
2909 Herrmann, Tinu Weber, Tom Stellard, Topi Miettinen, Torsten Hilbrich,
2910 Vito Caputo, Vladislav Vishnyakov, WaLyong Cho, Yu Watanabe, Zbigniew
2911 Jędrzejewski-Szmek, Zeal Jagannatha
2913 — Berlin, 2017-12-14
2917 * INCOMPATIBILITY: systemd-logind.service and other long-running
2918 services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP
2919 communication with the outside. This generally improves security of
2920 the system, and is in almost all cases a safe and good choice, as
2921 these services do not and should not provide any network-facing
2922 functionality. However, systemd-logind uses the glibc NSS API to
2923 query the user database. This creates problems on systems where NSS
2924 is set up to directly consult network services for user database
2925 lookups. In particular, this creates incompatibilities with the
2926 "nss-nis" module, which attempts to directly contact the NIS/YP
2927 network servers it is configured for, and will now consistently
2928 fail. In such cases, it is possible to turn off IP sandboxing for
2929 systemd-logind.service (set IPAddressDeny= in its [Service] section
2930 to the empty string, via a .d/ unit file drop-in). Downstream
2931 distributions might want to update their nss-nis packaging to include
2932 such a drop-in snippet, accordingly, to hide this incompatibility
2933 from the user. Another option is to make use of glibc's nscd service
2934 to proxy such network requests through a privilege-separated, minimal
2935 local caching daemon, or to switch to more modern technologies such
2936 sssd, whose NSS hook-ups generally do not involve direct network
2937 access. In general, we think it's definitely time to question the
2938 implementation choices of nss-nis, i.e. whether it's a good idea
2939 today to embed a network-facing loadable module into all local
2940 processes that need to query the user database, including the most
2941 trivial and benign ones, such as "ls". For more details about
2942 IPAddressDeny= see below.
2944 * A new modprobe.d drop-in is now shipped by default that sets the
2945 bonding module option max_bonds=0. This overrides the kernel default,
2946 to avoid conflicts and ambiguity as to whether or not bond0 should be
2947 managed by systemd-networkd or not. This resolves multiple issues
2948 with bond0 properties not being applied, when bond0 is configured
2949 with systemd-networkd. Distributors may choose to not package this,
2950 however in that case users will be prevented from correctly managing
2951 bond0 interface using systemd-networkd.
2953 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
2954 which print the logging level and target of the system manager. They
2955 complement the existing "set-log-level" and "set-log-target" verbs
2956 used to change those values.
2958 * journald.conf gained a new boolean setting ReadKMsg= which defaults
2959 to on. If turned off kernel log messages will not be read by
2960 systemd-journald or included in the logs. It also gained a new
2961 setting LineMax= for configuring the maximum line length in
2962 STDOUT/STDERR log streams. The new default for this value is 48K, up
2963 from the previous hardcoded 2048.
2965 * A new unit setting RuntimeDirectoryPreserve= has been added, which
2966 allows more detailed control of what to do with a runtime directory
2967 configured with RuntimeDirectory= (i.e. a directory below /run or
2968 $XDG_RUNTIME_DIR) after a unit is stopped.
2970 * The RuntimeDirectory= setting for units gained support for creating
2971 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
2972 one top-level directory.
2974 * Units gained new options StateDirectory=, CacheDirectory=,
2975 LogsDirectory= and ConfigurationDirectory= which are closely related
2976 to RuntimeDirectory= but manage per-service directories below
2977 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
2978 possible to write unit files which when activated automatically gain
2979 properly owned service specific directories in these locations, thus
2980 making unit files self-contained and increasing compatibility with
2981 stateless systems and factory reset where /etc or /var are
2982 unpopulated at boot. Matching these new settings there's also
2983 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
2984 ConfigurationDirectoryMode= for configuring the access mode of these
2985 directories. These settings are particularly useful in combination
2986 with DynamicUser=yes as they provide secure, properly-owned,
2987 writable, and stateful locations for storage, excluded from the
2988 sandbox that such services live in otherwise.
2990 * Automake support has been removed from this release. systemd is now
2993 * systemd-journald will now aggressively cache client metadata during
2994 runtime, speeding up log write performance under pressure. This comes
2995 at a small price though: as much of the metadata is read
2996 asynchronously from /proc/ (and isn't implicitly attached to log
2997 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
2998 metadata stored alongside a log entry might be slightly
2999 out-of-date. Previously it could only be slightly newer than the log
3000 message. The time window is small however, and given that the kernel
3001 is unlikely to be improved anytime soon in this regard, this appears
3004 * nss-myhostname/systemd-resolved will now by default synthesize an
3005 A/AAAA resource record for the "_gateway" hostname, pointing to the
3006 current default IP gateway. Previously it did that for the "gateway"
3007 name, hampering adoption, as some distributions wanted to leave that
3008 hostname open for local use. The old behaviour may still be
3009 requested at build time.
3011 * systemd-networkd's [Address] section in .network files gained a new
3012 Scope= setting for configuring the IP address scope. The [Network]
3013 section gained a new boolean setting ConfigureWithoutCarrier= that
3014 tells systemd-networkd to ignore link sensing when configuring the
3015 device. The [DHCP] section gained a new Anonymize= boolean option for
3016 turning on a number of options suggested in RFC 7844. A new
3017 [RoutingPolicyRule] section has been added for configuring the IP
3018 routing policy. The [Route] section has gained support for a new
3019 Type= setting which permits configuring
3020 blackhole/unreachable/prohibit routes.
3022 * The [VRF] section in .netdev files gained a new Table= setting for
3023 configuring the routing table to use. The [Tunnel] section gained a
3024 new Independent= boolean field for configuring tunnels independent of
3025 an underlying network interface. The [Bridge] section gained a new
3026 GroupForwardMask= option for configuration of propagation of link
3027 local frames between bridge ports.
3029 * The WakeOnLan= setting in .link files gained support for a number of
3030 new modes. A new TCP6SegmentationOffload= setting has been added for
3031 configuring TCP/IPv6 hardware segmentation offload.
3033 * The IPv6 RA sender implementation may now optionally send out RDNSS
3034 and RDNSSL records to supply DNS configuration to peers.
3036 * systemd-nspawn gained support for a new --system-call-filter= command
3037 line option for adding and removing entries in the default system
3038 call filter it applies. Moreover systemd-nspawn has been changed to
3039 implement a system call whitelist instead of a blacklist.
3041 * systemd-run gained support for a new --pipe command line option. If
3042 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
3043 are directly passed on to the activated transient service
3044 executable. This allows invoking arbitrary processes as systemd
3045 services (for example to take benefit of dependency management,
3046 accounting management, resource management or log management that is
3047 done automatically for services) — while still allowing them to be
3048 integrated in a classic UNIX shell pipeline.
3050 * When a service sends RELOAD=1 via sd_notify() and reload propagation
3051 using ReloadPropagationTo= is configured, a reload is now propagated
3052 to configured units. (Previously this was only done on explicitly
3053 requested reloads, using "systemctl reload" or an equivalent
3056 * For each service unit a restart counter is now kept: it is increased
3057 each time the service is restarted due to Restart=, and may be
3058 queried using "systemctl show -p NRestarts …".
3060 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
3061 @signal and @timer have been added, for usage with SystemCallFilter=
3062 in unit files and the new --system-call-filter= command line option
3063 of systemd-nspawn (see above).
3065 * ExecStart= lines in unit files gained two new modifiers: when a
3066 command line is prefixed with "!" the command will be executed as
3067 configured, except for the credentials applied by
3068 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
3069 "+", but does still apply namespacing options unlike "+". There's
3070 also "!!" now, which is mostly identical, but becomes a NOP on
3071 systems that support ambient capabilities. This is useful to write
3072 unit files that work with ambient capabilities where possible but
3073 automatically fall back to traditional privilege dropping mechanisms
3074 on systems where this is not supported.
3076 * ListenNetlink= settings in socket units now support RDMA netlink
3079 * A new unit file setting LockPersonality= has been added which permits
3080 locking down the chosen execution domain ("personality") of a service
3083 * A new special target "getty-pre.target" has been added, which is
3084 ordered before all text logins, and may be used to order services
3085 before textual logins acquire access to the console.
3087 * systemd will now attempt to load the virtio-rng.ko kernel module very
3088 early on if a VM environment supporting this is detected. This should
3089 improve entropy during early boot in virtualized environments.
3091 * A _netdev option is now supported in /etc/crypttab that operates in a
3092 similar way as the same option in /etc/fstab: it permits configuring
3093 encrypted devices that need to be ordered after the network is up.
3094 Following this logic, two new special targets
3095 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
3096 added that are to cryptsetup.target what remote-fs.target and
3097 remote-fs-pre.target are to local-fs.target.
3099 * Service units gained a new UnsetEnvironment= setting which permits
3100 unsetting specific environment variables for services that are
3101 normally passed to it (for example in order to mask out locale
3102 settings for specific services that can't deal with it).
3104 * Units acquired a new boolean option IPAccounting=. When turned on, IP
3105 traffic accounting (packet count as well as byte count) is done for
3106 the service, and shown as part of "systemctl status" or "systemd-run
3109 * Service units acquired two new options IPAddressAllow= and
3110 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
3111 for configuring a simple IP access control list for all sockets of
3112 the unit. These options are available also on .slice and .socket
3113 units, permitting flexible access list configuration for individual
3114 services as well as groups of services (as defined by a slice unit),
3115 including system-wide. Note that IP ACLs configured this way are
3116 enforced on every single IPv4 and IPv6 socket created by any process
3117 of the service unit, and apply to ingress as well as egress traffic.
3119 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
3120 structured log message is generated each time the unit is stopped,
3121 containing information about the consumed resources of this
3124 * A new setting KeyringMode= has been added to unit files, which may be
3125 used to control how the kernel keyring is set up for executed
3128 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
3129 "systemctl kexec" and "systemctl exit" are now always asynchronous in
3130 behaviour (that is: these commands return immediately after the
3131 operation was enqueued instead of waiting for the operation to
3132 complete). Previously, "systemctl poweroff" and "systemctl reboot"
3133 were asynchronous on systems using systemd-logind (i.e. almost
3134 always, and like they were on sysvinit), and the other three commands
3135 were unconditionally synchronous. With this release this is cleaned
3136 up, and callers will see the same asynchronous behaviour on all
3137 systems for all five operations.
3139 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
3142 * .timer units now accept calendar specifications in other timezones
3143 than UTC or the local timezone.
3145 * The tmpfiles snippet var.conf has been changed to create
3146 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
3147 the "utmp" group already, and it appears to be generally understood
3148 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
3149 databases. Previously this was implemented correctly for all these
3150 databases excepts btmp, which has been opened up like this now
3151 too. Note that while the other databases are world-readable
3152 (i.e. 0644), btmp is not and remains more restrictive.
3154 * The systemd-resolve tool gained a new --reset-server-features
3155 switch. When invoked like this systemd-resolved will forget
3156 everything it learnt about the features supported by the configured
3157 upstream DNS servers, and restarts the feature probing logic on the
3158 next resolver look-up for them at the highest feature level
3161 * The status dump systemd-resolved sends to the logs upon receiving
3162 SIGUSR1 now also includes information about all DNS servers it is
3163 configured to use, and the features levels it probed for them.
3165 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
3166 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
3167 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
3168 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
3169 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
3170 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
3171 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
3172 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
3173 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
3174 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
3175 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
3176 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
3177 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
3178 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
3179 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
3180 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
3181 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
3182 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
3183 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
3184 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
3186 — Berlin, 2017-10-06
3190 * Meson is now supported as build system in addition to Automake. It is
3191 our plan to remove Automake in one of our next releases, so that
3192 Meson becomes our exclusive build system. Hence, please start using
3193 the Meson build system in your downstream packaging. There's plenty
3194 of documentation around how to use Meson, the extremely brief
3197 ./autogen.sh && ./configure && make && sudo make install
3201 meson build && ninja -C build && sudo ninja -C build install
3203 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
3204 which permits configuring a timeout on the time a job is
3205 running. This is particularly useful for setting timeouts on jobs for
3208 * Unit files gained two new options ConditionUser= and ConditionGroup=
3209 for conditionalizing units based on the identity of the user/group
3210 running a systemd user instance.
3212 * systemd-networkd now understands a new FlowLabel= setting in the
3213 [VXLAN] section of .network files, as well as a Priority= in
3214 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
3215 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
3216 gained support for configuration of GENEVE links, and IPv6 address
3217 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
3219 * .link files now understand a new Port= setting.
3221 * systemd-networkd's DHCP support gained support for DHCP option 119
3222 (domain search list).
3224 * systemd-networkd gained support for serving IPv6 address ranges using
3225 the Router Advertisement protocol. The new .network configuration
3226 section [IPv6Prefix] may be used to configure the ranges to
3227 serve. This is implemented based on a new, minimal, native server
3228 implementation of RA.
3230 * journalctl's --output= switch gained support for a new parameter
3231 "short-iso-precise" for a mode where timestamps are shown as precise
3234 * systemd-udevd's "net_id" builtin may now generate stable network
3235 interface names from IBM PowerVM VIO devices as well as ACPI platform
3238 * MulticastDNS support in systemd-resolved may now be explicitly
3239 enabled/disabled using the new MulticastDNS= configuration file
3242 * systemd-resolved may now optionally use libidn2 instead of the libidn
3243 for processing internationalized domain names. Support for libidn2
3244 should be considered experimental and should not be enabled by
3247 * "machinectl pull-tar" and related call may now do verification of
3248 downloaded images using SUSE-style .sha256 checksum files in addition
3249 to the already existing support for validating using Ubuntu-style
3252 * sd-bus gained support for a new sd_bus_message_appendv() call which
3253 is va_list equivalent of sd_bus_message_append().
3255 * sd-boot gained support for validating images using SHIM/MOK.
3257 * The SMACK code learnt support for "onlycap".
3259 * systemd-mount --umount is now much smarter in figuring out how to
3260 properly unmount a device given its mount or device path.
3262 * The code to call libnss_dns as a fallback from libnss_resolve when
3263 the communication with systemd-resolved fails was removed. This
3264 fallback was redundant and interfered with the [!UNAVAIL=return]
3265 suffix. See nss-resolve(8) for the recommended configuration.
3267 * systemd-logind may now be restarted without losing state. It stores
3268 the file descriptors for devices it manages in the system manager
3269 using the FDSTORE= mechanism. Please note that further changes in
3270 other components may be required to make use of this (for example
3271 Xorg has code to listen for stops of systemd-logind and terminate
3272 itself when logind is stopped or restarted, in order to avoid using
3273 stale file descriptors for graphical devices, which is now
3274 counterproductive and must be reverted in order for restarts of
3275 systemd-logind to be safe. See
3276 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
3278 * All kernel-install plugins are called with the environment variable
3279 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
3280 /etc/machine-id. If the machine ID could not be determined,
3281 $KERNEL_INSTALL_MACHINE_ID will be empty. Plugins should not put
3282 anything in the entry directory (passed as the second argument) if
3283 $KERNEL_INSTALL_MACHINE_ID is empty. For backwards compatibility, a
3284 temporary directory is passed as the entry directory and removed
3285 after all the plugins exit.
3287 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
3288 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
3289 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
3290 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
3291 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
3292 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
3293 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
3294 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
3295 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
3296 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
3297 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
3298 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
3299 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
3300 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
3301 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
3302 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
3303 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
3304 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
3305 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
3306 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
3307 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
3308 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
3309 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
3310 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
3311 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
3312 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
3313 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
3314 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
3315 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
3318 — Berlin, 2017-07-12
3322 * The "hybrid" control group mode has been modified to improve
3323 compatibility with "legacy" cgroups-v1 setups. Specifically, the
3324 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
3325 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
3326 cgroups-v1 hierarchy), the only externally visible change being that
3327 the cgroups-v2 hierarchy is also mounted, to
3328 /sys/fs/cgroup/unified. This should provide a large degree of
3329 compatibility with "legacy" cgroups-v1, while taking benefit of the
3330 better management capabilities of cgroups-v2.
3332 * The default control group setup mode may be selected both a boot-time
3333 via a set of kernel command line parameters (specifically:
3334 systemd.unified_cgroup_hierarchy= and
3335 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
3336 default selected on the configure command line
3337 (--with-default-hierarchy=). The upstream default is "hybrid"
3338 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
3339 this will change in a future systemd version to be "unified" (pure
3340 cgroups-v2 mode). The third option for the compile time option is
3341 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
3342 distributions to default to "hybrid" mode for release distributions,
3343 starting with v233. We recommend "unified" for development
3344 distributions (specifically: distributions such as Fedora's rawhide)
3345 as that's where things are headed in the long run. Use "legacy" for
3346 greatest stability and compatibility only.
3348 * Note one current limitation of "unified" and "hybrid" control group
3349 setup modes: the kernel currently does not permit the systemd --user
3350 instance (i.e. unprivileged code) to migrate processes between two
3351 disconnected cgroup subtrees, even if both are managed and owned by
3352 the user. This effectively means "systemd-run --user --scope" doesn't
3353 work when invoked from outside of any "systemd --user" service or
3354 scope. Specifically, it is not supported from session scopes. We are
3355 working on fixing this in a future systemd version. (See #3388 for
3356 further details about this.)
3358 * DBus policy files are now installed into /usr rather than /etc. Make
3359 sure your system has dbus >= 1.9.18 running before upgrading to this
3360 version, or override the install path with --with-dbuspolicydir= .
3362 * All python scripts shipped with systemd (specifically: the various
3363 tests written in Python) now require Python 3.
3365 * systemd unit tests can now run standalone (without the source or
3366 build directories), and can be installed into /usr/lib/systemd/tests/
3367 with 'make install-tests'.
3369 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
3370 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
3373 * Support for the %c, %r, %R specifiers in unit files has been
3374 removed. Specifiers are not supposed to be dependent on configuration
3375 in the unit file itself (so that they resolve the same regardless
3376 where used in the unit files), but these specifiers were influenced
3377 by the Slice= option.
3379 * The shell invoked by debug-shell.service now defaults to /bin/sh in
3380 all cases. If distributions want to use a different shell for this
3381 purpose (for example Fedora's /sbin/sushell) they need to specify
3382 this explicitly at configure time using --with-debug-shell=.
3384 * The confirmation spawn prompt has been reworked to offer the
3387 (c)ontinue, proceed without asking anymore
3388 (D)ump, show the state of the unit
3389 (f)ail, don't execute the command and pretend it failed
3391 (i)nfo, show a short summary of the unit
3392 (j)obs, show jobs that are in progress
3393 (s)kip, don't execute the command and pretend it succeeded
3394 (y)es, execute the command
3396 The 'n' choice for the confirmation spawn prompt has been removed,
3397 because its meaning was confusing.
3399 The prompt may now also be redirected to an alternative console by
3400 specifying the console as parameter to systemd.confirm_spawn=.
3402 * Services of Type=notify require a READY=1 notification to be sent
3403 during startup. If no such message is sent, the service now fails,
3404 even if the main process exited with a successful exit code.
3406 * Services that fail to start up correctly now always have their
3407 ExecStopPost= commands executed. Previously, they'd enter "failed"
3408 state directly, without executing these commands.
3410 * The option MulticastDNS= of network configuration files has acquired
3411 an actual implementation. With MulticastDNS=yes a host can resolve
3412 names of remote hosts and reply to mDNS A and AAAA requests.
3414 * When units are about to be started an additional check is now done to
3415 ensure that all dependencies of type BindsTo= (when used in
3416 combination with After=) have been started.
3418 * systemd-analyze gained a new verb "syscall-filter" which shows which
3419 system call groups are defined for the SystemCallFilter= unit file
3420 setting, and which system calls they contain.
3422 * A new system call filter group "@filesystem" has been added,
3423 consisting of various file system related system calls. Group
3424 "@reboot" has been added, covering reboot, kexec and shutdown related
3425 calls. Finally, group "@swap" has been added covering swap
3426 configuration related calls.
3428 * A new unit file option RestrictNamespaces= has been added that may be
3429 used to restrict access to the various process namespace types the
3430 Linux kernel provides. Specifically, it may be used to take away the
3431 right for a service unit to create additional file system, network,
3432 user, and other namespaces. This sandboxing option is particularly
3433 relevant due to the high amount of recently discovered namespacing
3434 related vulnerabilities in the kernel.
3436 * systemd-udev's .link files gained support for a new AutoNegotiation=
3437 setting for configuring Ethernet auto-negotiation.
3439 * systemd-networkd's .network files gained support for a new
3440 ListenPort= setting in the [DHCP] section to explicitly configure the
3441 UDP client port the DHCP client shall listen on.
3443 * .network files gained a new Unmanaged= boolean setting for explicitly
3444 excluding one or more interfaces from management by systemd-networkd.
3446 * The systemd-networkd ProxyARP= option has been renamed to
3447 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
3448 renamed to ReduceARPProxy=. The old names continue to be available
3451 * systemd-networkd gained support for configuring IPv6 Proxy NDP
3452 addresses via the new IPv6ProxyNDPAddress= .network file setting.
3454 * systemd-networkd's bonding device support gained support for two new
3455 configuration options ActiveSlave= and PrimarySlave=.
3457 * The various options in the [Match] section of .network files gained
3458 support for negative matching.
3460 * New systemd-specific mount options are now understood in /etc/fstab:
3462 x-systemd.mount-timeout= may be used to configure the maximum
3463 permitted runtime of the mount command.
3465 x-systemd.device-bound may be set to bind a mount point to its
3466 backing device unit, in order to automatically remove a mount point
3467 if its backing device is unplugged. This option may also be
3468 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
3469 on the block device, which is now automatically set for all CDROM
3470 drives, so that mounted CDs are automatically unmounted when they are
3471 removed from the drive.
3473 x-systemd.after= and x-systemd.before= may be used to explicitly
3474 order a mount after or before another unit or mount point.
3476 * Enqueued start jobs for device units are now automatically garbage
3477 collected if there are no jobs waiting for them anymore.
3479 * systemctl list-jobs gained two new switches: with --after, for every
3480 queued job the jobs it's waiting for are shown; with --before the
3481 jobs which it's blocking are shown.
3483 * systemd-nspawn gained support for ephemeral boots from disk images
3484 (or in other words: --ephemeral and --image= may now be
3485 combined). Moreover, ephemeral boots are now supported for normal
3486 directories, even if the backing file system is not btrfs. Of course,
3487 if the file system does not support file system snapshots or
3488 reflinks, the initial copy operation will be relatively expensive, but
3489 this should still be suitable for many use cases.
3491 * Calendar time specifications in .timer units now support
3492 specifications relative to the end of a month by using "~" instead of
3493 "-" as separator between month and day. For example, "*-02~03" means
3494 "the third last day in February". In addition a new syntax for
3495 repeated events has been added using the "/" character. For example,
3496 "9..17/2:00" means "every two hours from 9am to 5pm".
3498 * systemd-socket-proxyd gained a new parameter --connections-max= for
3499 configuring the maximum number of concurrent connections.
3501 * sd-id128 gained a new API for generating unique IDs for the host in a
3502 way that does not leak the machine ID. Specifically,
3503 sd_id128_get_machine_app_specific() derives an ID based on the
3504 machine ID a in well-defined, non-reversible, stable way. This is
3505 useful whenever an identifier for the host is needed but where the
3506 identifier shall not be useful to identify the system beyond the
3507 scope of the application itself. (Internally this uses HMAC-SHA256 as
3508 keyed hash function using the machine ID as input.)
3510 * NotifyAccess= gained a new supported value "exec". When set
3511 notifications are accepted from all processes systemd itself invoked,
3512 including all control processes.
3514 * .nspawn files gained support for defining overlay mounts using the
3515 Overlay= and OverlayReadOnly= options. Previously this functionality
3516 was only available on the systemd-nspawn command line.
3518 * systemd-nspawn's --bind= and --overlay= options gained support for
3519 bind/overlay mounts whose source lies within the container tree by
3520 prefixing the source path with "+".
3522 * systemd-nspawn's --bind= and --overlay= options gained support for
3523 automatically allocating a temporary source directory in /var/tmp
3524 that is removed when the container dies. Specifically, if the source
3525 directory is specified as empty string this mechanism is selected. An
3526 example usage is --overlay=+/var::/var, which creates an overlay
3527 mount based on the original /var contained in the image, overlaid
3528 with a temporary directory in the host's /var/tmp. This way changes
3529 to /var are automatically flushed when the container shuts down.
3531 * systemd-nspawn --image= option does now permit raw file system block
3532 devices (in addition to images containing partition tables, as
3535 * The disk image dissection logic in systemd-nspawn gained support for
3536 automatically setting up LUKS encrypted as well as Verity protected
3537 partitions. When a container is booted from an encrypted image the
3538 passphrase is queried at start-up time. When a container with Verity
3539 data is started, the root hash is search in a ".roothash" file
3540 accompanying the disk image (alternatively, pass the root hash via
3541 the new --root-hash= command line option).
3543 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
3544 be used to dissect disk images the same way as systemd-nspawn does
3545 it, following the Bootable Partition Specification. It may even be
3546 used to mount disk images with complex partition setups (including
3547 LUKS and Verity partitions) to a local host directory, in order to
3548 inspect them. This tool is not considered public API (yet), and is
3549 thus not installed into /usr/bin. Please do not rely on its
3550 existence, since it might go away or be changed in later systemd
3553 * A new generator "systemd-verity-generator" has been added, similar in
3554 style to "systemd-cryptsetup-generator", permitting automatic setup of
3555 Verity root partitions when systemd boots up. In order to make use of
3556 this your partition setup should follow the Discoverable Partitions
3557 Specification, and the GPT partition ID of the root file system
3558 partition should be identical to the upper 128bit of the Verity root
3559 hash. The GPT partition ID of the Verity partition protecting it
3560 should be the lower 128bit of the Verity root hash. If the partition
3561 image follows this model it is sufficient to specify a single
3562 "roothash=" kernel command line argument to both configure which root
3563 image and verity partition to use as well as the root hash for
3564 it. Note that systemd-nspawn's Verity support follows the same
3565 semantics, meaning that disk images with proper Verity data in place
3566 may be booted in containers with systemd-nspawn as well as on
3567 physical systems via the verity generator. Also note that the "mkosi"
3568 tool available at https://github.com/systemd/mkosi has been updated
3569 to generate Verity protected disk images following this scheme. In
3570 fact, it has been updated to generate disk images that optionally
3571 implement a complete UEFI SecureBoot trust chain, involving a signed
3572 kernel and initrd image that incorporates such a root hash as well as
3573 a Verity-enabled root partition.
3575 * The hardware database (hwdb) udev supports has been updated to carry
3576 accelerometer quirks.
3578 * All system services are now run with a fresh kernel keyring set up
3579 for them. The invocation ID is stored by default in it, thus
3580 providing a safe, non-overridable way to determine the invocation
3583 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
3584 options for bind mounting arbitrary paths in a service-specific
3585 way. When these options are used, arbitrary host or service files and
3586 directories may be mounted to arbitrary locations in the service's
3589 * Documentation has been added that lists all of systemd's low-level
3590 environment variables:
3592 https://github.com/systemd/systemd/blob/master/docs/ENVIRONMENT.md
3594 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
3595 whether a specific socket file descriptor matches a specified socket
3598 * systemd-firstboot has been updated to check for the
3599 systemd.firstboot= kernel command line option. It accepts a boolean
3600 and when set to false the first boot questions are skipped.
3602 * systemd-fstab-generator has been updated to check for the
3603 systemd.volatile= kernel command line option, which either takes an
3604 optional boolean parameter or the special value "state". If used the
3605 system may be booted in a "volatile" boot mode. Specifically,
3606 "systemd.volatile" is used, the root directory will be mounted as
3607 tmpfs, and only /usr is mounted from the actual root file system. If
3608 "systemd.volatile=state" is used, the root directory will be mounted
3609 as usual, but /var is mounted as tmpfs. This concept provides similar
3610 functionality as systemd-nspawn's --volatile= option, but provides it
3611 on physical boots. Use this option for implementing stateless
3612 systems, or testing systems with all state and/or configuration reset
3613 to the defaults. (Note though that many distributions are not
3614 prepared to boot up without a populated /etc or /var, though.)
3616 * systemd-gpt-auto-generator gained support for LUKS encrypted root
3617 partitions. Previously it only supported LUKS encrypted partitions
3618 for all other uses, except for the root partition itself.
3620 * Socket units gained support for listening on AF_VSOCK sockets for
3621 communication in virtualized QEMU environments.
3623 * The "configure" script gained a new option --with-fallback-hostname=
3624 for specifying the fallback hostname to use if none is configured in
3625 /etc/hostname. For example, by specifying
3626 --with-fallback-hostname=fedora it is possible to default to a
3627 hostname of "fedora" on pristine installations.
3629 * systemd-cgls gained support for a new --unit= switch for listing only
3630 the control groups of a specific unit. Similar --user-unit= has been
3631 added for listing only the control groups of a specific user unit.
3633 * systemd-mount gained a new --umount switch for unmounting a mount or
3634 automount point (and all mount/automount points below it).
3636 * systemd will now refuse full configuration reloads (via systemctl
3637 daemon-reload and related calls) unless at least 16MiB of free space
3638 are available in /run. This is a safety precaution in order to ensure
3639 that generators can safely operate after the reload completed.
3641 * A new unit file option RootImage= has been added, which has a similar
3642 effect as RootDirectory= but mounts the service's root directory from
3643 a disk image instead of plain directory. This logic reuses the same
3644 image dissection and mount logic that systemd-nspawn already uses,
3645 and hence supports any disk images systemd-nspawn supports, including
3646 those following the Discoverable Partition Specification, as well as
3647 Verity enabled images. This option enables systemd to run system
3648 services directly off disk images acting as resource bundles,
3649 possibly even including full integrity data.
3651 * A new MountAPIVFS= unit file option has been added, taking a boolean
3652 argument. If enabled /proc, /sys and /dev (collectively called the
3653 "API VFS") will be mounted for the service. This is only relevant if
3654 RootDirectory= or RootImage= is used for the service, as these mounts
3655 are of course in place in the host mount namespace anyway.
3657 * systemd-nspawn gained support for a new --pivot-root= switch. If
3658 specified the root directory within the container image is pivoted to
3659 the specified mount point, while the original root disk is moved to a
3660 different place. This option enables booting of ostree images
3661 directly with systemd-nspawn.
3663 * The systemd build scripts will no longer complain if the NTP server
3664 addresses are not changed from the defaults. Google now supports
3665 these NTP servers officially. We still recommend downstreams to
3666 properly register an NTP pool with the NTP pool project though.
3668 * coredumpctl gained a new "--reverse" option for printing the list
3669 of coredumps in reverse order.
3671 * coredumpctl will now show additional information about truncated and
3672 inaccessible coredumps, as well as coredumps that are still being
3673 processed. It also gained a new --quiet switch for suppressing
3674 additional informational message in its output.
3676 * coredumpctl gained support for only showing coredumps newer and/or
3677 older than specific timestamps, using the new --since= and --until=
3678 options, reminiscent of journalctl's options by the same name.
3680 * The systemd-coredump logic has been improved so that it may be reused
3681 to collect backtraces in non-compiled languages, for example in
3682 scripting languages such as Python.
3684 * machinectl will now show the UID shift of local containers, if user
3685 namespacing is enabled for them.
3687 * systemd will now optionally run "environment generator" binaries at
3688 configuration load time. They may be used to add environment
3689 variables to the environment block passed to services invoked. One
3690 user environment generator is shipped by default that sets up
3691 environment variables based on files dropped into /etc/environment.d
3692 and ~/.config/environment.d/.
3694 * systemd-resolved now includes the new, recently published 2017 DNSSEC
3697 * hostnamed has been updated to report a new chassis type of
3698 "convertible" to cover "foldable" laptops that can both act as a
3699 tablet and as a laptop, such as various Lenovo Yoga devices.
3701 Contributions from: Adrián López, Alexander Galanin, Alexander
3702 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
3703 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
3704 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
3705 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
3706 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
3707 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
3708 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
3709 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
3710 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
3711 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
3712 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
3713 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
3714 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
3715 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
3716 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
3717 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
3718 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
3719 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
3720 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
3721 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
3722 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
3723 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
3724 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
3725 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
3726 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
3727 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
3728 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
3731 — Berlin, 2017-03-01
3735 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
3736 RestrictAddressFamilies= enabled. These sandboxing options should
3737 generally be compatible with the various external udev call-out
3738 binaries we are aware of, however there may be exceptions, in
3739 particular when exotic languages for these call-outs are used. In
3740 this case, consider turning off these settings locally.
3742 * The new RemoveIPC= option can be used to remove IPC objects owned by
3743 the user or group of a service when that service exits.
3745 * The new ProtectKernelModules= option can be used to disable explicit
3746 load and unload operations of kernel modules by a service. In
3747 addition access to /usr/lib/modules is removed if this option is set.
3749 * ProtectSystem= option gained a new value "strict", which causes the
3750 whole file system tree with the exception of /dev, /proc, and /sys,
3751 to be remounted read-only for a service.
3753 * The new ProtectKernelTunables= option can be used to disable
3754 modification of configuration files in /sys and /proc by a service.
3755 Various directories and files are remounted read-only, so access is
3756 restricted even if the file permissions would allow it.
3758 * The new ProtectControlGroups= option can be used to disable write
3759 access by a service to /sys/fs/cgroup.
3761 * Various systemd services have been hardened with
3762 ProtectKernelTunables=yes, ProtectControlGroups=yes,
3763 RestrictAddressFamilies=.
3765 * Support for dynamically creating users for the lifetime of a service
3766 has been added. If DynamicUser=yes is specified, user and group IDs
3767 will be allocated from the range 61184..65519 for the lifetime of the
3768 service. They can be resolved using the new nss-systemd.so NSS
3769 module. The module must be enabled in /etc/nsswitch.conf. Services
3770 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
3771 any resources allocated by the service will be cleaned up when the
3772 service exits. They also have ProtectHome=read-only and
3773 ProtectSystem=strict enabled, so they are not able to make any
3774 permanent modifications to the system.
3776 * The nss-systemd module also always resolves root and nobody, making
3777 it possible to have no /etc/passwd or /etc/group files in minimal
3778 container or chroot environments.
3780 * Services may be started with their own user namespace using the new
3781 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
3782 under which the service is running are mapped. All other users are
3785 * Support for the cgroup namespace has been added to systemd-nspawn. If
3786 supported by kernel, the container system started by systemd-nspawn
3787 will have its own view of the cgroup hierarchy. This new behaviour
3788 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
3790 * The new MemorySwapMax= option can be used to limit the maximum swap
3791 usage under the unified cgroup hierarchy.
3793 * Support for the CPU controller in the unified cgroup hierarchy has
3794 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
3795 options. This controller requires out-of-tree patches for the kernel
3796 and the support is provisional.
3798 * Mount and automount units may now be created transiently
3799 (i.e. dynamically at runtime via the bus API, instead of requiring
3800 unit files in the file system).
3802 * systemd-mount is a new tool which may mount file systems – much like
3803 mount(8), optionally pulling in additional dependencies through
3804 transient .mount and .automount units. For example, this tool
3805 automatically runs fsck on a backing block device before mounting,
3806 and allows the automount logic to be used dynamically from the
3807 command line for establishing mount points. This tool is particularly
3808 useful when dealing with removable media, as it will ensure fsck is
3809 run – if necessary – before the first access and that the file system
3810 is quickly unmounted after each access by utilizing the automount
3811 logic. This maximizes the chance that the file system on the
3812 removable media stays in a clean state, and if it isn't in a clean
3813 state is fixed automatically.
3815 * LazyUnmount=yes option for mount units has been added to expose the
3816 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
3819 * /efi will be used as the mount point of the EFI boot partition, if
3820 the directory is present, and the mount point was not configured
3821 through other means (e.g. fstab). If /efi directory does not exist,
3822 /boot will be used as before. This makes it easier to automatically
3823 mount the EFI partition on systems where /boot is used for something
3826 * When operating on GPT disk images for containers, systemd-nspawn will
3827 now mount the ESP to /boot or /efi according to the same rules as PID
3828 1 running on a host. This allows tools like "bootctl" to operate
3829 correctly within such containers, in order to make container images
3830 bootable on physical systems.
3832 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
3834 * Two new user session targets have been added to support running
3835 graphical sessions under the systemd --user instance:
3836 graphical-session.target and graphical-session-pre.target. See
3837 systemd.special(7) for a description of how those targets should be
3840 * The vconsole initialization code has been significantly reworked to
3841 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
3842 support unicode keymaps. Font and keymap configuration will now be
3843 copied to all allocated virtual consoles.
3845 * FreeBSD's bhyve virtualization is now detected.
3847 * Information recorded in the journal for core dumps now includes the
3848 contents of /proc/mountinfo and the command line of the process at
3849 the top of the process hierarchy (which is usually the init process
3852 * systemd-journal-gatewayd learned the --directory= option to serve
3853 files from the specified location.
3855 * journalctl --root=… can be used to peruse the journal in the
3856 /var/log/ directories inside of a container tree. This is similar to
3857 the existing --machine= option, but does not require the container to
3860 * The hardware database has been extended to support
3861 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
3864 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
3865 specify the click rate for mice which include a horizontal wheel with
3866 a click rate that is different than the one for the vertical wheel.
3868 * systemd-run gained a new --wait option that makes service execution
3869 synchronous. (Specifically, the command will not return until the
3870 specified service binary exited.)
3872 * systemctl gained a new --wait option that causes the start command to
3873 wait until the units being started have terminated again.
3875 * A new journal output mode "short-full" has been added which displays
3876 timestamps with abbreviated English day names and adds a timezone
3877 suffix. Those timestamps include more information than the default
3878 "short" output mode, and can be passed directly to journalctl's
3879 --since= and --until= options.
3881 * /etc/resolv.conf will be bind-mounted into containers started by
3882 systemd-nspawn, if possible, so any changes to resolv.conf contents
3883 are automatically propagated to the container.
3885 * The number of instances for socket-activated services originating
3886 from a single IP address can be limited with
3887 MaxConnectionsPerSource=, extending the existing setting of
3890 * systemd-networkd gained support for vcan ("Virtual CAN") interface
3893 * .netdev and .network configuration can now be extended through
3896 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
3897 Segmentation Offload, Generic Receive Offload, Large Receive Offload
3898 can be enabled and disabled using the new UDPSegmentationOffload=,
3899 TCPSegmentationOffload=, GenericSegmentationOffload=,
3900 GenericReceiveOffload=, LargeReceiveOffload= options in the
3901 [Link] section of .link files.
3903 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
3904 Port VLAN ID can be configured for bridge devices using the new STP=,
3905 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
3906 section of .netdev files.
3908 * The route table to which routes received over DHCP or RA should be
3909 added can be configured with the new RouteTable= option in the [DHCP]
3910 and [IPv6AcceptRA] sections of .network files.
3912 * The Address Resolution Protocol can be disabled on links managed by
3913 systemd-networkd using the ARP=no setting in the [Link] section of
3916 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
3917 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
3918 encode information about the result and exit codes of the current
3919 service runtime cycle.
3921 * systemd-sysctl will now configure kernel parameters in the order
3922 they occur in the configuration files. This matches what sysctl
3923 has been traditionally doing.
3925 * kernel-install "plugins" that are executed to perform various
3926 tasks after a new kernel is added and before an old one is removed
3927 can now return a special value to terminate the procedure and
3928 prevent any later plugins from running.
3930 * Journald's SplitMode=login setting has been deprecated. It has been
3931 removed from documentation, and its use is discouraged. In a future
3932 release it will be completely removed, and made equivalent to current
3933 default of SplitMode=uid.
3935 * Storage=both option setting in /etc/systemd/coredump.conf has been
3936 removed. With fast LZ4 compression storing the core dump twice is not
3939 * The --share-system systemd-nspawn option has been replaced with an
3940 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
3941 this functionality is discouraged. In addition the variables
3942 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
3943 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
3944 individual namespaces.
3946 * "machinectl list" now shows the IP address of running containers in
3947 the output, as well as OS release information.
3949 * "loginctl list" now shows the TTY of each session in the output.
3951 * sd-bus gained new API calls sd_bus_track_set_recursive(),
3952 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
3953 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
3954 tracking objects in a "recursive" mode, where a single client can be
3955 counted multiple times, if it takes multiple references.
3957 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
3958 sd_bus_get_exit_on_disconnect(). They may be used to make a
3959 process using sd-bus automatically exit if the bus connection is
3962 * Bus clients of the service manager may now "pin" loaded units into
3963 memory, by taking an explicit reference on them. This is useful to
3964 ensure the client can retrieve runtime data about the service even
3965 after the service completed execution. Taking such a reference is
3966 available only for privileged clients and should be helpful to watch
3967 running services in a race-free manner, and in particular collect
3968 information about exit statuses and results.
3970 * The nss-resolve module has been changed to strictly return UNAVAIL
3971 when communication via D-Bus with resolved failed, and NOTFOUND when
3972 a lookup completed but was negative. This means it is now possible to
3973 neatly configure fallbacks using nsswitch.conf result checking
3974 expressions. Taking benefit of this, the new recommended
3975 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
3977 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
3979 * A new setting CtrlAltDelBurstAction= has been added to
3980 /etc/systemd/system.conf which may be used to configure the precise
3981 behaviour if the user on the console presses Ctrl-Alt-Del more often
3982 than 7 times in 2s. Previously this would unconditionally result in
3983 an expedited, immediate reboot. With this new setting the precise
3984 operation may be configured in more detail, and also turned off
3987 * In .netdev files two new settings RemoteChecksumTx= and
3988 RemoteChecksumRx= are now understood that permit configuring the
3989 remote checksumming logic for VXLAN networks.
3991 * The service manager learnt a new "invocation ID" concept for invoked
3992 services. Each runtime cycle of a service will get a new invocation
3993 ID (a 128bit random UUID) assigned that identifies the current
3994 run of the service uniquely and globally. A new invocation ID
3995 is generated each time a service starts up. The journal will store
3996 the invocation ID of a service along with any logged messages, thus
3997 making the invocation ID useful for matching the online runtime of a
3998 service with the offline log data it generated in a safe way without
3999 relying on synchronized timestamps. In many ways this new service
4000 invocation ID concept is similar to the kernel's boot ID concept that
4001 uniquely and globally identifies the runtime of each boot. The
4002 invocation ID of a service is passed to the service itself via an
4003 environment variable ($INVOCATION_ID). A new bus call
4004 GetUnitByInvocationID() has been added that is similar to GetUnit()
4005 but instead of retrieving the bus path for a unit by its name
4006 retrieves it by its invocation ID. The returned path is valid only as
4007 long as the passed invocation ID is current.
4009 * systemd-resolved gained a new "DNSStubListener" setting in
4010 resolved.conf. It either takes a boolean value or the special values
4011 "udp" and "tcp", and configures whether to enable the stub DNS
4012 listener on 127.0.0.53:53.
4014 * IP addresses configured via networkd may now carry additional
4015 configuration settings supported by the kernel. New options include:
4016 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
4017 PrefixRoute=, AutoJoin=.
4019 * The PAM configuration fragment file for "user@.service" shipped with
4020 systemd (i.e. the --user instance of systemd) has been stripped to
4021 the minimum necessary to make the system boot. Previously, it
4022 contained Fedora-specific stanzas that did not apply to other
4023 distributions. It is expected that downstream distributions add
4024 additional configuration lines, matching their needs to this file,
4025 using it only as rough template of what systemd itself needs. Note
4026 that this reduced fragment does not even include an invocation of
4027 pam_limits which most distributions probably want to add, even though
4028 systemd itself does not need it. (There's also the new build time
4029 option --with-pamconfdir=no to disable installation of the PAM
4032 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
4033 capability is now also dropped from its set (in addition to
4034 CAP_SYS_MKNOD as before).
4036 * In service unit files it is now possible to connect a specific named
4037 file descriptor with stdin/stdout/stdout of an executed service. The
4038 name may be specified in matching .socket units using the
4039 FileDescriptorName= setting.
4041 * A number of journal settings may now be configured on the kernel
4042 command line. Specifically, the following options are now understood:
4043 systemd.journald.max_level_console=,
4044 systemd.journald.max_level_store=,
4045 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
4046 systemd.journald.max_level_wall=.
4048 * "systemctl is-enabled --full" will now show by which symlinks a unit
4049 file is enabled in the unit dependency tree.
4051 * Support for VeraCrypt encrypted partitions has been added to the
4052 "cryptsetup" logic and /etc/crypttab.
4054 * systemd-detect-virt gained support for a new --private-users switch
4055 that checks whether the invoking processes are running inside a user
4056 namespace. Similar, a new special value "private-users" for the
4057 existing ConditionVirtualization= setting has been added, permitting
4058 skipping of specific units in user namespace environments.
4060 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
4061 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
4062 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
4063 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
4064 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
4065 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
4066 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
4067 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
4068 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
4069 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
4070 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
4071 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
4072 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
4073 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
4074 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
4075 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
4076 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
4077 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
4078 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
4079 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
4080 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
4081 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
4082 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
4083 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
4084 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
4085 Jędrzejewski-Szmek, Zeal Jagannatha
4087 — Santa Fe, 2016-11-03
4091 * In service units the various ExecXYZ= settings have been extended
4092 with an additional special character as first argument of the
4093 assigned value: if the character '+' is used the specified command
4094 line it will be run with full privileges, regardless of User=,
4095 Group=, CapabilityBoundingSet= and similar options. The effect is
4096 similar to the existing PermissionsStartOnly= option, but allows
4097 configuration of this concept for each executed command line
4100 * Services may now alter the service watchdog timeout at runtime by
4101 sending a WATCHDOG_USEC= message via sd_notify().
4103 * MemoryLimit= and related unit settings now optionally take percentage
4104 specifications. The percentage is taken relative to the amount of
4105 physical memory in the system (or in case of containers, the assigned
4106 amount of memory). This allows scaling service resources neatly with
4107 the amount of RAM available on the system. Similarly, systemd-logind's
4108 RuntimeDirectorySize= option now also optionally takes percentage
4111 * In similar fashion TasksMax= takes percentage values now, too. The
4112 value is taken relative to the configured maximum number of processes
4113 on the system. The per-service task maximum has been changed to 15%
4114 using this functionality. (Effectively this is an increase of 512 →
4115 4915 for service units, given the kernel's default pid_max setting.)
4117 * Calendar time specifications in .timer units now understand a ".."
4118 syntax for time ranges. Example: "4..7:10" may now be used for
4119 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
4122 * The InaccessableDirectories=, ReadOnlyDirectories= and
4123 ReadWriteDirectories= unit file settings have been renamed to
4124 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
4125 applied to all kinds of file nodes, and not just directories, with
4126 the exception of symlinks. Specifically these settings may now be
4127 used on block and character device nodes, UNIX sockets and FIFOS as
4128 well as regular files. The old names of these settings remain
4129 available for compatibility.
4131 * systemd will now log about all service processes it kills forcibly
4132 (using SIGKILL) because they remained after the clean shutdown phase
4133 of the service completed. This should help identifying services that
4134 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
4135 systemd-logind's configuration a similar log message is generated for
4136 processes killed at the end of each session due to this setting.
4138 * systemd will now set the $JOURNAL_STREAM environment variable for all
4139 services whose stdout/stderr are connected to the Journal (which
4140 effectively means by default: all services). The variable contains
4141 the device and inode number of the file descriptor used for
4142 stdout/stderr. This may be used by invoked programs to detect whether
4143 their stdout/stderr is connected to the Journal, in which case they
4144 can switch over to direct Journal communication, thus being able to
4145 pass extended, structured metadata along with their log messages. As
4146 one example, this is now used by glib's logging primitives.
4148 * When using systemd's default tmp.mount unit for /tmp, the mount point
4149 will now be established with the "nosuid" and "nodev" options. This
4150 avoids privilege escalation attacks that put traps and exploits into
4151 /tmp. However, this might cause problems if you e. g. put container
4152 images or overlays into /tmp; if you need this, override tmp.mount's
4153 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
4156 * systemd now supports the "memory" cgroup controller also on
4159 * The systemd-cgtop tool now optionally takes a control group path as
4160 command line argument. If specified, the control group list shown is
4161 limited to subgroups of that group.
4163 * The SystemCallFilter= unit file setting gained support for
4164 pre-defined, named system call filter sets. For example
4165 SystemCallFilter=@clock is now an effective way to make all clock
4166 changing-related system calls unavailable to a service. A number of
4167 similar pre-defined groups are defined. Writing system call filters
4168 for system services is simplified substantially with this new
4169 concept. Accordingly, all of systemd's own, long-running services now
4170 enable system call filtering based on this, by default.
4172 * A new service setting MemoryDenyWriteExecute= has been added, taking
4173 a boolean value. If turned on, a service may no longer create memory
4174 mappings that are writable and executable at the same time. This
4175 enhances security for services where this is enabled as it becomes
4176 harder to dynamically write and then execute memory in exploited
4177 service processes. This option has been enabled for all of systemd's
4178 own long-running services.
4180 * A new RestrictRealtime= service setting has been added, taking a
4181 boolean argument. If set the service's processes may no longer
4182 acquire realtime scheduling. This improves security as realtime
4183 scheduling may otherwise be used to easily freeze the system.
4185 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
4186 value. This may be used for requesting that the system manager inside
4187 of the container reports start-up completion to nspawn which then
4188 propagates this notification further to the service manager
4189 supervising nspawn itself. A related option NotifyReady= in .nspawn
4190 files has been added too. This functionality allows ordering of the
4191 start-up of multiple containers using the usual systemd ordering
4194 * machinectl gained a new command "stop" that is an alias for
4197 * systemd-resolved gained support for contacting DNS servers on
4198 link-local IPv6 addresses.
4200 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
4201 its caches. A method call for requesting the same operation has been
4202 added to the bus API too, and is made available via "systemd-resolve
4205 * systemd-resolve gained a new --status switch. If passed a brief
4206 summary of the used DNS configuration with per-interface information
4209 * resolved.conf gained a new Cache= boolean option, defaulting to
4210 on. If turned off local DNS caching is disabled. This comes with a
4211 performance penalty in particular when DNSSEC is enabled. Note that
4212 resolved disables its internal caching implicitly anyway, when the
4213 configured DNS server is on a host-local IP address such as ::1 or
4214 127.0.0.1, thus automatically avoiding double local caching.
4216 * systemd-resolved now listens on the local IP address 127.0.0.53:53
4217 for DNS requests. This improves compatibility with local programs
4218 that do not use the libc NSS or systemd-resolved's bus APIs for name
4219 resolution. This minimal DNS service is only available to local
4220 programs and does not implement the full DNS protocol, but enough to
4221 cover local DNS clients. A new, static resolv.conf file, listing just
4222 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
4223 now recommended to make /etc/resolv.conf a symlink to this file in
4224 order to route all DNS lookups to systemd-resolved, regardless if
4225 done via NSS, the bus API or raw DNS packets. Note that this local
4226 DNS service is not as fully featured as the libc NSS or
4227 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
4228 used to deliver link-local address information (as this implies
4229 sending a local interface index along), LLMNR/mDNS support via this
4230 interface is severely restricted. It is thus strongly recommended for
4231 all applications to use the libc NSS API or native systemd-resolved
4234 * systemd-networkd's bridge support learned a new setting
4235 VLANFiltering= for controlling VLAN filtering. Moreover a new section
4236 in .network files has been added for configuring VLAN bridging in
4237 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
4239 * systemd-networkd's IPv6 Router Advertisement code now makes use of
4240 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
4241 now be acquired without relying on DHCPv6. Two new options
4242 UseDomains= and UseDNS= have been added to configure this behaviour.
4244 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
4245 renamed IPv6AcceptRA=, without altering its behaviour. The old
4246 setting name remains available for compatibility reasons.
4248 * The systemd-networkd VTI/VTI6 tunneling support gained new options
4249 Key=, InputKey= and OutputKey=.
4251 * systemd-networkd gained support for VRF ("Virtual Routing Function")
4252 interface configuration.
4254 * "systemctl edit" may now be used to create new unit files by
4255 specifying the --force switch.
4257 * sd-event gained a new function sd_event_get_iteration() for
4258 requesting the current iteration counter of the event loop. It starts
4259 at zero and is increased by one with each event loop iteration.
4261 * A new rpm macro %systemd_ordering is provided by the macros.systemd
4262 file. It can be used in lieu of %systemd_requires in packages which
4263 don't use any systemd functionality and are intended to be installed
4264 in minimal containers without systemd present. This macro provides
4265 ordering dependencies to ensure that if the package is installed in
4266 the same rpm transaction as systemd, systemd will be installed before
4267 the scriptlets for the package are executed, allowing unit presets
4270 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
4271 been added to simplify packaging of generators.
4273 * The os-release file gained VERSION_CODENAME field for the
4274 distribution nickname (e.g. VERSION_CODENAME=woody).
4276 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
4277 can be set to disable parsing of metadata and the creation
4278 of persistent symlinks for that device.
4280 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
4281 to make them available to logged-in users has been reverted.
4283 * Much of the common code of the various systemd components is now
4284 built into an internal shared library libsystemd-shared-231.so
4285 (incorporating the systemd version number in the name, to be updated
4286 with future releases) that the components link to. This should
4287 decrease systemd footprint both in memory during runtime and on
4288 disk. Note that the shared library is not for public use, and is
4289 neither API nor ABI stable, but is likely to change with every new
4290 released update. Packagers need to make sure that binaries
4291 linking to libsystemd-shared.so are updated in step with the
4294 * Configuration for "mkosi" is now part of the systemd
4295 repository. mkosi is a tool to easily build legacy-free OS images,
4296 and is available on github: https://github.com/systemd/mkosi. If
4297 "mkosi" is invoked in the build tree a new raw OS image is generated
4298 incorporating the systemd sources currently being worked on and a
4299 clean, fresh distribution installation. The generated OS image may be
4300 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
4301 UEFI PC. This functionality is particularly useful to easily test
4302 local changes made to systemd in a pristine, defined environment. See
4303 doc/HACKING for details.
4305 * configure learned the --with-support-url= option to specify the
4306 distribution's bugtracker.
4308 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
4309 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
4310 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
4311 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
4312 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
4313 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
4314 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
4315 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
4316 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
4317 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
4318 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
4319 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
4320 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
4321 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
4322 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
4323 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
4324 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
4325 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
4326 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
4328 — Berlin, 2016-07-25
4332 * DNSSEC is now turned on by default in systemd-resolved (in
4333 "allow-downgrade" mode), but may be turned off during compile time by
4334 passing "--with-default-dnssec=no" to "configure" (and of course,
4335 during runtime with DNSSEC= in resolved.conf). We recommend
4336 downstreams to leave this on at least during development cycles and
4337 report any issues with the DNSSEC logic upstream. We are very
4338 interested in collecting feedback about the DNSSEC validator and its
4339 limitations in the wild. Note however, that DNSSEC support is
4340 probably nothing downstreams should turn on in stable distros just
4341 yet, as it might create incompatibilities with a few DNS servers and
4342 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
4343 automatically whenever we detect such incompatible setups, but there
4344 might be systems we do not cover yet. Hence: please help us testing
4345 the DNSSEC code, leave this on where you can, report back, but then
4346 again don't consider turning this on in your stable, LTS or
4347 production release just yet. (Note that you have to enable
4348 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
4349 and its DNSSEC mode for hostname resolution from local
4352 * systemd-resolve conveniently resolves DANE records with the --tlsa
4353 option and OPENPGPKEY records with the --openpgp option. It also
4354 supports dumping raw DNS record data via the new --raw= switch.
4356 * systemd-logind will now by default terminate user processes that are
4357 part of the user session scope unit (session-XX.scope) when the user
4358 logs out. This behavior is controlled by the KillUserProcesses=
4359 setting in logind.conf, and the previous default of "no" is now
4360 changed to "yes". This means that user sessions will be properly
4361 cleaned up after, but additional steps are necessary to allow
4362 intentionally long-running processes to survive logout.
4364 While the user is logged in at least once, user@.service is running,
4365 and any service that should survive the end of any individual login
4366 session can be started at a user service or scope using systemd-run.
4367 systemd-run(1) man page has been extended with an example which shows
4368 how to run screen in a scope unit underneath user@.service. The same
4369 command works for tmux.
4371 After the user logs out of all sessions, user@.service will be
4372 terminated too, by default, unless the user has "lingering" enabled.
4373 To effectively allow users to run long-term tasks even if they are
4374 logged out, lingering must be enabled for them. See loginctl(1) for
4375 details. The default polkit policy was modified to allow users to
4376 set lingering for themselves without authentication.
4378 Previous defaults can be restored at compile time by the
4379 --without-kill-user-processes option to "configure".
4381 * systemd-logind gained new configuration settings SessionsMax= and
4382 InhibitorsMax=, both with a default of 8192. It will not register new
4383 user sessions or inhibitors above this limit.
4385 * systemd-logind will now reload configuration on SIGHUP.
4387 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
4388 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
4389 enable. Also, support for the "io" cgroup controller in the unified
4390 hierarchy has been added, so that the "memory", "pids" and "io" are
4391 now the controllers that are supported on the unified hierarchy.
4393 WARNING: it is not possible to use previous systemd versions with
4394 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
4395 is necessary to also update systemd in the initramfs if using the
4396 unified hierarchy. An updated SELinux policy is also required.
4398 * LLDP support has been extended, and both passive (receive-only) and
4399 active (sender) modes are supported. Passive mode ("routers-only") is
4400 enabled by default in systemd-networkd. Active LLDP mode is enabled
4401 by default for containers on the internal network. The "networkctl
4402 lldp" command may be used to list information gathered. "networkctl
4403 status" will also show basic LLDP information on connected peers now.
4405 * The IAID and DUID unique identifier sent in DHCP requests may now be
4406 configured for the system and each .network file managed by
4407 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
4409 * systemd-networkd gained support for configuring proxy ARP support for
4410 each interface, via the ProxyArp= setting in .network files. It also
4411 gained support for configuring the multicast querier feature of
4412 bridge devices, via the new MulticastQuerier= setting in .netdev
4413 files. Similarly, snooping on the IGMP traffic can be controlled
4414 via the new setting MulticastSnooping=.
4416 A new setting PreferredLifetime= has been added for addresses
4417 configured in .network file to configure the lifetime intended for an
4420 The systemd-networkd DHCP server gained the option EmitRouter=, which
4421 defaults to yes, to configure whether the DHCP Option 3 (Router)
4424 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
4425 systemd-socket-activate and installed into /usr/bin. It is now fully
4428 * systemd-journald now uses separate threads to flush changes to disk
4429 when closing journal files, thus reducing impact of slow disk I/O on
4430 logging performance.
4432 * The sd-journal API gained two new calls
4433 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
4434 can be used to open journal files using file descriptors instead of
4435 file or directory paths. sd_journal_open_container() has been
4436 deprecated, sd_journal_open_directory_fd() should be used instead
4437 with the flag SD_JOURNAL_OS_ROOT.
4439 * journalctl learned a new output mode "-o short-unix" that outputs log
4440 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
4441 UTC). It also gained support for a new --no-hostname setting to
4442 suppress the hostname column in the family of "short" output modes.
4444 * systemd-ask-password now optionally skips printing of the password to
4445 stdout with --no-output which can be useful in scripts.
4447 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
4448 (devices tagged with ID_MAKER_TOOL) are now tagged with
4449 "uaccess" and are available to logged in users.
4451 * The DeviceAllow= unit setting now supports specifiers (with "%").
4453 * "systemctl show" gained a new --value switch, which allows print a
4454 only the contents of a specific unit property, without also printing
4455 the property's name. Similar support was added to "show*" verbs
4456 of loginctl and machinectl that output "key=value" lists.
4458 * A new unit type "generated" was added for files dynamically generated
4459 by generator tools. Similarly, a new unit type "transient" is used
4460 for unit files created using the runtime API. "systemctl enable" will
4461 refuse to operate on such files.
4463 * A new command "systemctl revert" has been added that may be used to
4464 revert to the vendor version of a unit file, in case local changes
4465 have been made by adding drop-ins or overriding the unit file.
4467 * "machinectl clean" gained a new verb to automatically remove all or
4468 just hidden container images.
4470 * systemd-tmpfiles gained support for a new line type "e" for emptying
4471 directories, if they exist, without creating them if they don't.
4473 * systemd-nspawn gained support for automatically patching the UID/GIDs
4474 of the owners and the ACLs of all files and directories in a
4475 container tree to match the UID/GID user namespacing range selected
4476 for the container invocation. This mode is enabled via the new
4477 --private-users-chown switch. It also gained support for
4478 automatically choosing a free, previously unused UID/GID range when
4479 starting a container, via the new --private-users=pick setting (which
4480 implies --private-users-chown). Together, these options for the first
4481 time make user namespacing for nspawn containers fully automatic and
4482 thus deployable. The systemd-nspawn@.service template unit file has
4483 been changed to use this functionality by default.
4485 * systemd-nspawn gained a new --network-zone= switch, that allows
4486 creating ad-hoc virtual Ethernet links between multiple containers,
4487 that only exist as long as at least one container referencing them is
4488 running. This allows easy connecting of multiple containers with a
4489 common link that implements an Ethernet broadcast domain. Each of
4490 these network "zones" may be named relatively freely by the user, and
4491 may be referenced by any number of containers, but each container may
4492 only reference one of these "zones". On the lower level, this is
4493 implemented by an automatically managed bridge network interface for
4494 each zone, that is created when the first container referencing its
4495 zone is created and removed when the last one referencing its zone
4498 * The default start timeout may now be configured on the kernel command
4499 line via systemd.default_timeout_start_sec=. It was already
4500 configurable via the DefaultTimeoutStartSec= option in
4501 /etc/systemd/system.conf.
4503 * Socket units gained a new TriggerLimitIntervalSec= and
4504 TriggerLimitBurst= setting to configure a limit on the activation
4505 rate of the socket unit.
4507 * The LimitNICE= setting now optionally takes normal UNIX nice values
4508 in addition to the raw integer limit value. If the specified
4509 parameter is prefixed with "+" or "-" and is in the range -20..19 the
4510 value is understood as UNIX nice value. If not prefixed like this it
4511 is understood as raw RLIMIT_NICE limit.
4513 * Note that the effect of the PrivateDevices= unit file setting changed
4514 slightly with this release: the per-device /dev file system will be
4515 mounted read-only from this version on, and will have "noexec"
4516 set. This (minor) change of behavior might cause some (exceptional)
4517 legacy software to break, when PrivateDevices=yes is set for its
4518 service. Please leave PrivateDevices= off if you run into problems
4521 * systemd-bootchart has been split out to a separate repository:
4522 https://github.com/systemd/systemd-bootchart
4524 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
4525 merged into the kernel in its current form.
4527 * The compatibility libraries libsystemd-daemon.so,
4528 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
4529 which have been deprecated since systemd-209 have been removed along
4530 with the corresponding pkg-config files. All symbols provided by
4531 those libraries are provided by libsystemd.so.
4533 * The Capabilities= unit file setting has been removed (it is ignored
4534 for backwards compatibility). AmbientCapabilities= and
4535 CapabilityBoundingSet= should be used instead.
4537 * A new special target has been added, initrd-root-device.target,
4538 which creates a synchronization point for dependencies of the root
4539 device in early userspace. Initramfs builders must ensure that this
4540 target is now included in early userspace.
4542 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
4543 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
4544 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
4545 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
4546 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
4547 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
4548 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
4549 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
4550 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
4551 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
4552 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
4553 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
4554 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
4555 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
4556 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
4557 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
4558 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
4559 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
4560 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
4561 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
4562 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
4563 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
4564 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
4565 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
4566 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
4569 — Fairfax, 2016-05-21
4573 * The systemd-resolved DNS resolver service has gained a substantial
4574 set of new features, most prominently it may now act as a DNSSEC
4575 validating stub resolver. DNSSEC mode is currently turned off by
4576 default, but is expected to be turned on by default in one of the
4577 next releases. For now, we invite everybody to test the DNSSEC logic
4578 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
4579 service also gained a full set of D-Bus interfaces, including calls
4580 to configure DNS and DNSSEC settings per link (for use by external
4581 network management software). systemd-resolved and systemd-networkd
4582 now distinguish between "search" and "routing" domains. The former
4583 are used to qualify single-label names, the latter are used purely
4584 for routing lookups within certain domains to specific links.
4585 resolved now also synthesizes RRs for all entries from /etc/hosts.
4587 * The systemd-resolve tool (which is a client utility for
4588 systemd-resolved) has been improved considerably and is now fully
4589 supported and documented. Hence it has moved from /usr/lib/systemd to
4592 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
4595 * The coredump collection logic has been reworked: when a coredump is
4596 collected it is now written to disk, compressed and processed
4597 (including stacktrace extraction) from a new instantiated service
4598 systemd-coredump@.service, instead of directly from the
4599 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
4600 processing large coredumps can take up a substantial amount of
4601 resources and time, and this previously happened entirely outside of
4602 systemd's service supervision. With the new logic the core_pattern
4603 hook only does minimal metadata collection before passing off control
4604 to the new instantiated service, which is configured with a time
4605 limit, a nice level and other settings to minimize negative impact on
4606 the rest of the system. Also note that the new logic will honour the
4607 RLIMIT_CORE setting of the crashed process, which now allows users
4608 and processes to turn off coredumping for their processes by setting
4611 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
4612 and all forked processes by default. Previously, PID 1 would leave
4613 the setting at "0" for all processes, as set by the kernel. Note that
4614 the resource limit traditionally has no effect on the generated
4615 coredumps on the system if the /proc/sys/kernel/core_pattern hook
4616 logic is used. Since the limit is now honoured (see above) its
4617 default has been changed so that the coredumping logic is enabled by
4618 default for all processes, while allowing specific opt-out.
4620 * When the stacktrace is extracted from processes of system users, this
4621 is now done as "systemd-coredump" user, in order to sandbox this
4622 potentially security sensitive parsing operation. (Note that when
4623 processing coredumps of normal users this is done under the user ID
4624 of process that crashed, as before.) Packagers should take notice
4625 that it is now necessary to create the "systemd-coredump" system user
4626 and group at package installation time.
4628 * The systemd-activate socket activation testing tool gained support
4629 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
4630 and --seqpacket switches. It also has been extended to support both
4631 new-style and inetd-style file descriptor passing. Use the new
4632 --inetd switch to request inetd-style file descriptor passing.
4634 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
4635 variable, which takes a boolean value. If set to false, ANSI color
4636 output is disabled in the tools even when run on a terminal that
4639 * The VXLAN support in networkd now supports two new settings
4640 DestinationPort= and PortRange=.
4642 * A new systemd.machine_id= kernel command line switch has been added,
4643 that may be used to set the machine ID in /etc/machine-id if it is
4644 not initialized yet. This command line option has no effect if the
4645 file is already initialized.
4647 * systemd-nspawn gained a new --as-pid2 switch that invokes any
4648 specified command line as PID 2 rather than PID 1 in the
4649 container. In this mode PID 1 is a minimal stub init process that
4650 implements the special POSIX and Linux semantics of PID 1 regarding
4651 signal and child process management. Note that this stub init process
4652 is implemented in nspawn itself and requires no support from the
4653 container image. This new logic is useful to support running
4654 arbitrary commands in the container, as normal processes are
4655 generally not prepared to run as PID 1.
4657 * systemd-nspawn gained a new --chdir= switch for setting the current
4658 working directory for the process started in the container.
4660 * "journalctl /dev/sda" will now output all kernel log messages for
4661 specified device from the current boot, in addition to all devices
4662 that are parents of it. This should make log output about devices
4663 pretty useful, as long as kernel drivers attach enough metadata to
4664 the log messages. (The usual SATA drivers do.)
4666 * The sd-journal API gained two new calls
4667 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
4668 that report whether log data from /run or /var has been found.
4670 * journalctl gained a new switch "--fields" that prints all journal
4671 record field names currently in use in the journal. This is backed
4672 by two new sd-journal API calls sd_journal_enumerate_fields() and
4673 sd_journal_restart_fields().
4675 * Most configurable timeouts in systemd now expect an argument of
4676 "infinity" to turn them off, instead of "0" as before. The semantics
4677 from now on is that a timeout of "0" means "now", and "infinity"
4678 means "never". To maintain backwards compatibility, "0" continues to
4679 turn off previously existing timeout settings.
4681 * "systemctl reload-or-try-restart" has been renamed to "systemctl
4682 try-reload-or-restart" to clarify what it actually does: the "try"
4683 logic applies to both reloading and restarting, not just restarting.
4684 The old name continues to be accepted for compatibility.
4686 * On boot-up, when PID 1 detects that the system clock is behind the
4687 release date of the systemd version in use, the clock is now set
4688 to the latter. Previously, this was already done in timesyncd, in order
4689 to avoid running with clocks set to the various clock epochs such as
4690 1902, 1938 or 1970. With this change the logic is now done in PID 1
4691 in addition to timesyncd during early boot-up, so that it is enforced
4692 before the first process is spawned by systemd. Note that the logic
4693 in timesyncd remains, as it is more comprehensive and ensures
4694 clock monotonicity by maintaining a persistent timestamp file in
4695 /var. Since /var is generally not available in earliest boot or the
4696 initrd, this part of the logic remains in timesyncd, and is not done
4699 * Support for tweaking details in net_cls.class_id through the
4700 NetClass= configuration directive has been removed, as the kernel
4701 people have decided to deprecate that controller in cgroup v2.
4702 Userspace tools such as nftables are moving over to setting rules
4703 that are specific to the full cgroup path of a task, which obsoletes
4704 these controllers anyway. The NetClass= directive is kept around for
4705 legacy compatibility reasons. For a more in-depth description of the
4706 kernel change, please refer to the respective upstream commit:
4708 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
4710 * A new service setting RuntimeMaxSec= has been added that may be used
4711 to specify a maximum runtime for a service. If the timeout is hit, the
4712 service is terminated and put into a failure state.
4714 * A new service setting AmbientCapabilities= has been added. It allows
4715 configuration of additional Linux process capabilities that are
4716 passed to the activated processes. This is only available on very
4719 * The process resource limit settings in service units may now be used
4720 to configure hard and soft limits individually.
4722 * The various libsystemd APIs such as sd-bus or sd-event now publicly
4723 expose support for gcc's __attribute__((cleanup())) C extension.
4724 Specifically, for many object destructor functions alternative
4725 versions have been added that have names suffixed with "p" and take a
4726 pointer to a pointer to the object to destroy, instead of just a
4727 pointer to the object itself. This is useful because these destructor
4728 functions may be used directly as parameters to the cleanup
4729 construct. Internally, systemd has been a heavy user of this GCC
4730 extension for a long time, and with this change similar support is
4731 now available to consumers of the library outside of systemd. Note
4732 that by using this extension in your sources compatibility with old
4733 and strictly ANSI compatible C compilers is lost. However, all gcc or
4734 LLVM versions of recent years support this extension.
4736 * Timer units gained support for a new setting RandomizedDelaySec= that
4737 allows configuring some additional randomized delay to the configured
4738 time. This is useful to spread out timer events to avoid load peaks in
4739 clusters or larger setups.
4741 * Calendar time specifications now support sub-second accuracy.
4743 * Socket units now support listening on SCTP and UDP-lite protocol
4746 * The sd-event API now comes with a full set of man pages.
4748 * Older versions of systemd contained experimental support for
4749 compressing journal files and coredumps with the LZ4 compressor that
4750 was not compatible with the lz4 binary (due to API limitations of the
4751 lz4 library). This support has been removed; only support for files
4752 compatible with the lz4 binary remains. This LZ4 logic is now
4753 officially supported and no longer considered experimental.
4755 * The dkr image import logic has been removed again from importd. dkr's
4756 micro-services focus doesn't fit into the machine image focus of
4757 importd, and quickly got out of date with the upstream dkr API.
4759 * Creation of the /run/lock/lockdev/ directory was dropped from
4760 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
4761 been available for many years. If you still need this, you need to
4762 create your own tmpfiles.d config file with:
4764 d /run/lock/lockdev 0775 root lock -
4766 * The settings StartLimitBurst=, StartLimitInterval=, StartLimitAction=
4767 and RebootArgument= have been moved from the [Service] section of
4768 unit files to [Unit], and they are now supported on all unit types,
4769 not just service units. Of course, systemd will continue to
4770 understand these settings also at the old location, in order to
4771 maintain compatibility.
4773 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
4774 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
4775 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
4776 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
4777 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
4778 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
4779 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
4780 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
4781 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
4782 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
4783 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
4784 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
4785 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
4786 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
4787 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
4788 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
4789 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
4790 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
4791 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
4793 — Berlin, 2016-02-11
4797 * A number of properties previously only settable in unit
4798 files are now also available as properties to set when
4799 creating transient units programmatically via the bus, as it
4800 is exposed with systemd-run's --property=
4801 setting. Specifically, these are: SyslogIdentifier=,
4802 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
4803 EnvironmentFile=, ReadWriteDirectories=,
4804 ReadOnlyDirectories=, InaccessibleDirectories=,
4805 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
4807 * When creating transient services via the bus API it is now
4808 possible to pass in a set of file descriptors to use as
4809 STDIN/STDOUT/STDERR for the invoked process.
4811 * Slice units may now be created transiently via the bus APIs,
4812 similar to the way service and scope units may already be
4813 created transiently.
4815 * Wherever systemd expects a calendar timestamp specification
4816 (like in journalctl's --since= and --until= switches) UTC
4817 timestamps are now supported. Timestamps suffixed with "UTC"
4818 are now considered to be in Universal Time Coordinated
4819 instead of the local timezone. Also, timestamps may now
4820 optionally be specified with sub-second accuracy. Both of
4821 these additions also apply to recurring calendar event
4822 specification, such as OnCalendar= in timer units.
4824 * journalctl gained a new "--sync" switch that asks the
4825 journal daemon to write all so far unwritten log messages to
4826 disk and sync the files, before returning.
4828 * systemd-tmpfiles learned two new line types "q" and "Q" that
4829 operate like "v", but also set up a basic btrfs quota
4830 hierarchy when used on a btrfs file system with quota
4833 * tmpfiles' "v", "q" and "Q" will now create a plain directory
4834 instead of a subvolume (even on a btrfs file system) if the
4835 root directory is a plain directory, and not a
4836 subvolume. This should simplify things with certain chroot()
4837 environments which are not aware of the concept of btrfs
4840 * systemd-detect-virt gained a new --chroot switch to detect
4841 whether execution takes place in a chroot() environment.
4843 * CPUAffinity= now takes CPU index ranges in addition to
4846 * The various memory-related resource limit settings (such as
4847 LimitAS=) now understand the usual K, M, G, ... suffixes to
4848 the base of 1024 (IEC). Similar, the time-related resource
4849 limit settings understand the usual min, h, day, ...
4852 * There's a new system.conf setting DefaultTasksMax= to
4853 control the default TasksMax= setting for services and
4854 scopes running on the system. (TasksMax= is the primary
4855 setting that exposes the "pids" cgroup controller on systemd
4856 and was introduced in the previous systemd release.) The
4857 setting now defaults to 512, which means services that are
4858 not explicitly configured otherwise will only be able to
4859 create 512 processes or threads at maximum, from this
4860 version on. Note that this means that thread- or
4861 process-heavy services might need to be reconfigured to set
4862 TasksMax= to a higher value. It is sufficient to set
4863 TasksMax= in these specific unit files to a higher value, or
4864 even "infinity". Similar, there's now a logind.conf setting
4865 UserTasksMax= that defaults to 4096 and limits the total
4866 number of processes or tasks each user may own
4867 concurrently. nspawn containers also have the TasksMax=
4868 value set by default now, to 8192. Note that all of this
4869 only has an effect if the "pids" cgroup controller is
4870 enabled in the kernel. The general benefit of these changes
4871 should be a more robust and safer system, that provides a
4872 certain amount of per-service fork() bomb protection.
4874 * systemd-nspawn gained the new --network-veth-extra= switch
4875 to define additional and arbitrarily-named virtual Ethernet
4876 links between the host and the container.
4878 * A new service execution setting PassEnvironment= has been
4879 added that allows importing select environment variables
4880 from PID1's environment block into the environment block of
4883 * Timer units gained support for a new RemainAfterElapse=
4884 setting which takes a boolean argument. It defaults to on,
4885 exposing behaviour unchanged to previous releases. If set to
4886 off, timer units are unloaded after they elapsed if they
4887 cannot elapse again. This is particularly useful for
4888 transient timer units, which shall not stay around longer
4889 than until they first elapse.
4891 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
4892 default now (the kernel default is 16). This is beneficial
4893 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
4894 allows substantially larger numbers of queued
4895 datagrams. This should increase the capability of systemd to
4896 parallelize boot-up, as logging and sd_notify() are unlikely
4897 to stall execution anymore. If you need to change the value
4898 from the new defaults, use the usual sysctl.d/ snippets.
4900 * The compression framing format used by the journal or
4901 coredump processing has changed to be in line with what the
4902 official LZ4 tools generate. LZ4 compression support in
4903 systemd was considered unsupported previously, as the format
4904 was not compatible with the normal tools. With this release
4905 this has changed now, and it is hence safe for downstream
4906 distributions to turn it on. While not compressing as well
4907 as the XZ, LZ4 is substantially faster, which makes
4908 it a good default choice for the compression logic in the
4909 journal and in coredump handling.
4911 * Any reference to /etc/mtab has been dropped from
4912 systemd. The file has been obsolete since a while, but
4913 systemd refused to work on systems where it was incorrectly
4914 set up (it should be a symlink or non-existent). Please make
4915 sure to update to util-linux 2.27.1 or newer in conjunction
4916 with this systemd release, which also drops any reference to
4917 /etc/mtab. If you maintain a distribution make sure that no
4918 software you package still references it, as this is a
4919 likely source of bugs. There's also a glibc bug pending,
4920 asking for removal of any reference to this obsolete file:
4922 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
4924 Note that only util-linux versions built with
4925 --enable-libmount-force-mountinfo are supported.
4927 * Support for the ".snapshot" unit type has been removed. This
4928 feature turned out to be little useful and little used, and
4929 has now been removed from the core and from systemctl.
4931 * The dependency types RequiresOverridable= and
4932 RequisiteOverridable= have been removed from systemd. They
4933 have been used only very sparingly to our knowledge and
4934 other options that provide a similar effect (such as
4935 systemctl --mode=ignore-dependencies) are much more useful
4936 and commonly used. Moreover, they were only half-way
4937 implemented as the option to control behaviour regarding
4938 these dependencies was never added to systemctl. By removing
4939 these dependency types the execution engine becomes a bit
4940 simpler. Unit files that use these dependencies should be
4941 changed to use the non-Overridable dependency types
4942 instead. In fact, when parsing unit files with these
4943 options, that's what systemd will automatically convert them
4944 too, but it will also warn, asking users to fix the unit
4945 files accordingly. Removal of these dependency types should
4946 only affect a negligible number of unit files in the wild.
4948 * Behaviour of networkd's IPForward= option changed
4949 (again). It will no longer maintain a per-interface setting,
4950 but propagate one way from interfaces where this is enabled
4951 to the global kernel setting. The global setting will be
4952 enabled when requested by a network that is set up, but
4953 never be disabled again. This change was made to make sure
4954 IPv4 and IPv6 behaviour regarding packet forwarding is
4955 similar (as the Linux IPv6 stack does not support
4956 per-interface control of this setting) and to minimize
4959 * In unit files the behaviour of %u, %U, %h, %s has
4960 changed. These specifiers will now unconditionally resolve
4961 to the various user database fields of the user that the
4962 systemd instance is running as, instead of the user
4963 configured in the specific unit via User=. Note that this
4964 effectively doesn't change much, as resolving of these
4965 specifiers was already turned off in the --system instance
4966 of systemd, as we cannot do NSS lookups from PID 1. In the
4967 --user instance of systemd these specifiers where correctly
4968 resolved, but hardly made any sense, since the user instance
4969 lacks privileges to do user switches anyway, and User= is
4970 hence useless. Moreover, even in the --user instance of
4971 systemd behaviour was awkward as it would only take settings
4972 from User= assignment placed before the specifier into
4973 account. In order to unify and simplify the logic around
4974 this the specifiers will now always resolve to the
4975 credentials of the user invoking the manager (which in case
4976 of PID 1 is the root user).
4978 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
4979 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
4980 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
4981 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
4982 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
4983 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
4984 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
4985 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
4986 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
4987 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
4988 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
4989 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
4990 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
4991 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
4994 — Berlin, 2015-11-18
4998 * systemd now depends on util-linux v2.27. More specifically,
4999 the newly added mount monitor feature in libmount now
5000 replaces systemd's former own implementation.
5002 * libmount mandates /etc/mtab not to be regular file, and
5003 systemd now enforces this condition at early boot.
5004 /etc/mtab has been deprecated and warned about for a very
5005 long time, so systems running systemd should already have
5006 stopped having this file around as anything else than a
5007 symlink to /proc/self/mounts.
5009 * Support for the "pids" cgroup controller has been added. It
5010 allows accounting the number of tasks in a cgroup and
5011 enforcing limits on it. This adds two new setting
5012 TasksAccounting= and TasksMax= to each unit, as well as a
5013 global option DefaultTasksAccounting=.
5015 * Support for the "net_cls" cgroup controller has been added.
5016 It allows assigning a net class ID to each task in the
5017 cgroup, which can then be used in firewall rules and traffic
5018 shaping configurations. Note that the kernel netfilter net
5019 class code does not currently work reliably for ingress
5020 packets on unestablished sockets.
5022 This adds a new config directive called NetClass= to CGroup
5023 enabled units. Allowed values are positive numbers for fixed
5024 assignments and "auto" for picking a free value
5027 * 'systemctl is-system-running' now returns 'offline' if the
5028 system is not booted with systemd. This command can now be
5029 used as a substitute for 'systemd-notify --booted'.
5031 * Watchdog timeouts have been increased to 3 minutes for all
5032 in-tree service files. Apparently, disk IO issues are more
5033 frequent than we hoped, and user reported >1 minute waiting
5036 * 'machine-id-commit' functionality has been merged into
5037 'machine-id-setup --commit'. The separate binary has been
5040 * The WorkingDirectory= directive in unit files may now be set
5041 to the special value '~'. In this case, the working
5042 directory is set to the home directory of the user
5043 configured in User=.
5045 * "machinectl shell" will now open the shell in the home
5046 directory of the selected user by default.
5048 * The CrashChVT= configuration file setting is renamed to
5049 CrashChangeVT=, following our usual logic of not
5050 abbreviating unnecessarily. The old directive is still
5051 supported for compat reasons. Also, this directive now takes
5052 an integer value between 1 and 63, or a boolean value. The
5053 formerly supported '-1' value for disabling stays around for
5056 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
5057 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
5058 RootDirectory= properties can now be set for transient
5061 * The systemd-analyze tool gained a new "set-log-target" verb
5062 to change the logging target the system manager logs to
5063 dynamically during runtime. This is similar to how
5064 "systemd-analyze set-log-level" already changes the log
5067 * In nspawn /sys is now mounted as tmpfs, with only a selected
5068 set of subdirectories mounted in from the real sysfs. This
5069 enhances security slightly, and is useful for ensuring user
5070 namespaces work correctly.
5072 * Support for USB FunctionFS activation has been added. This
5073 allows implementation of USB gadget services that are
5074 activated as soon as they are requested, so that they don't
5075 have to run continuously, similar to classic socket
5078 * The "systemctl exit" command now optionally takes an
5079 additional parameter that sets the exit code to return from
5080 the systemd manager when exiting. This is only relevant when
5081 running the systemd user instance, or when running the
5082 system instance in a container.
5084 * sd-bus gained the new API calls sd_bus_path_encode_many()
5085 and sd_bus_path_decode_many() that allow easy encoding and
5086 decoding of multiple identifier strings inside a D-Bus
5087 object path. Another new call sd_bus_default_flush_close()
5088 has been added to flush and close per-thread default
5091 * systemd-cgtop gained support for a -M/--machine= switch to
5092 show the control groups within a certain container only.
5094 * "systemctl kill" gained support for an optional --fail
5095 switch. If specified the requested operation will fail of no
5096 processes have been killed, because the unit had no
5097 processes attached, or similar.
5099 * A new systemd.crash_reboot=1 kernel command line option has
5100 been added that triggers a reboot after crashing. This can
5101 also be set through CrashReboot= in systemd.conf.
5103 * The RuntimeDirectory= setting now understands unit
5104 specifiers like %i or %f.
5106 * A new (still internal) library API sd-ipv4acd has been added,
5107 that implements address conflict detection for IPv4. It's
5108 based on code from sd-ipv4ll, and will be useful for
5109 detecting DHCP address conflicts.
5111 * File descriptors passed during socket activation may now be
5112 named. A new API sd_listen_fds_with_names() is added to
5113 access the names. The default names may be overridden,
5114 either in the .socket file using the FileDescriptorName=
5115 parameter, or by passing FDNAME= when storing the file
5116 descriptors using sd_notify().
5118 * systemd-networkd gained support for:
5120 - Setting the IPv6 Router Advertisement settings via
5121 IPv6AcceptRouterAdvertisements= in .network files.
5123 - Configuring the HelloTimeSec=, MaxAgeSec= and
5124 ForwardDelaySec= bridge parameters in .netdev files.
5126 - Configuring PreferredSource= for static routes in
5129 * The "ask-password" framework used to query for LUKS harddisk
5130 passwords or SSL passwords during boot gained support for
5131 caching passwords in the kernel keyring, if it is
5132 available. This makes sure that the user only has to type in
5133 a passphrase once if there are multiple objects to unlock
5134 with the same one. Previously, such password caching was
5135 available only when Plymouth was used; this moves the
5136 caching logic into the systemd codebase itself. The
5137 "systemd-ask-password" utility gained a new --keyname=
5138 switch to control which kernel keyring key to use for
5139 caching a password in. This functionality is also useful for
5140 enabling display managers such as gdm to automatically
5141 unlock the user's GNOME keyring if its passphrase, the
5142 user's password and the harddisk password are the same, if
5143 gdm-autologin is used.
5145 * When downloading tar or raw images using "machinectl
5146 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
5147 file is now also downloaded, if it is available and stored
5148 next to the image file.
5150 * Units of type ".socket" gained a new boolean setting
5151 Writable= which is only useful in conjunction with
5152 ListenSpecial=. If true, enables opening the specified
5153 special file in O_RDWR mode rather than O_RDONLY mode.
5155 * systemd-rfkill has been reworked to become a singleton
5156 service that is activated through /dev/rfkill on each rfkill
5157 state change and saves the settings to disk. This way,
5158 systemd-rfkill is now compatible with devices that exist
5159 only intermittendly, and even restores state if the previous
5160 system shutdown was abrupt rather than clean.
5162 * The journal daemon gained support for vacuuming old journal
5163 files controlled by the number of files that shall remain,
5164 in addition to the already existing control by size and by
5165 date. This is useful as journal interleaving performance
5166 degrades with too many separate journal files, and allows
5167 putting an effective limit on them. The new setting defaults
5168 to 100, but this may be changed by setting SystemMaxFiles=
5169 and RuntimeMaxFiles= in journald.conf. Also, the
5170 "journalctl" tool gained the new --vacuum-files= switch to
5171 manually vacuum journal files to leave only the specified
5172 number of files in place.
5174 * udev will now create /dev/disk/by-path links for ATA devices
5175 on kernels where that is supported.
5177 * Galician, Serbian, Turkish and Korean translations were added.
5179 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
5180 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
5181 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
5182 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
5183 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
5184 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
5185 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
5186 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
5187 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
5188 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
5189 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
5190 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
5191 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
5192 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
5193 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
5194 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
5195 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
5196 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
5198 — Berlin, 2015-10-07
5202 * The DHCP implementation of systemd-networkd gained a set of
5205 - The DHCP server now supports emitting DNS and NTP
5206 information. It may be enabled and configured via
5207 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
5208 and NTP information is enabled, but no servers are
5209 configured, the corresponding uplink information (if there
5210 is any) is propagated.
5212 - Server and client now support transmission and reception
5213 of timezone information. It can be configured via the
5214 newly introduced network options UseTimezone=,
5215 EmitTimezone=, and Timezone=. Transmission of timezone
5216 information is enabled between host and containers by
5217 default now: the container will change its local timezone
5218 to what the host has set.
5220 - Lease timeouts can now be configured via
5221 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
5223 - The DHCP server improved on the stability of
5224 leases. Clients are more likely to get the same lease
5225 information back, even if the server loses state.
5227 - The DHCP server supports two new configuration options to
5228 control the lease address pool metrics, PoolOffset= and
5231 * The encapsulation limit of tunnels in systemd-networkd may
5232 now be configured via 'EncapsulationLimit='. It allows
5233 modifying the maximum additional levels of encapsulation
5234 that are permitted to be prepended to a packet.
5236 * systemd now supports the concept of user buses replacing
5237 session buses, if used with dbus-1.10 (and enabled via dbus
5238 --enable-user-session). It previously only supported this on
5239 kdbus-enabled systems, and this release expands this to
5240 'dbus-daemon' systems.
5242 * systemd-networkd now supports predictable interface names
5245 * systemd now optionally supports the new Linux kernel
5246 "unified" control group hierarchy. If enabled via the kernel
5247 command-line option 'systemd.unified_cgroup_hierarchy=1',
5248 systemd will try to mount the unified cgroup hierarchy
5249 directly on /sys/fs/cgroup. If not enabled, or not
5250 available, systemd will fall back to the legacy cgroup
5251 hierarchy setup, as before. Host system and containers can
5252 mix and match legacy and unified hierarchies as they
5253 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
5254 environment variable to individually select the hierarchy to
5255 use for executed containers. By default, nspawn will use the
5256 unified hierarchy for the containers if the host uses the
5257 unified hierarchy, and the legacy hierarchy otherwise.
5258 Please note that at this point the unified hierarchy is an
5259 experimental kernel feature and is likely to change in one
5260 of the next kernel releases. Therefore, it should not be
5261 enabled by default in downstream distributions yet. The
5262 minimum required kernel version for the unified hierarchy to
5263 work is 4.2. Note that when the unified hierarchy is used
5264 for the first time delegated access to controllers is
5265 safe. Because of this systemd-nspawn containers will get
5266 access to controllers now, as will systemd user
5267 sessions. This means containers and user sessions may now
5268 manage their own resources, partitioning up what the system
5271 * A new special scope unit "init.scope" has been introduced
5272 that encapsulates PID 1 of the system. It may be used to
5273 determine resource usage and enforce resource limits on PID
5274 1 itself. PID 1 hence moved out of the root of the control
5277 * The cgtop tool gained support for filtering out kernel
5278 threads when counting tasks in a control group. Also, the
5279 count of processes is now recursively summed up by
5280 default. Two options -k and --recursive= have been added to
5281 revert to old behaviour. The tool has also been updated to
5282 work correctly in containers now.
5284 * systemd-nspawn's --bind= and --bind-ro= options have been
5285 extended to allow creation of non-recursive bind mounts.
5287 * libsystemd gained two new calls sd_pid_get_cgroup() and
5288 sd_peer_get_cgroup() which return the control group path of
5289 a process or peer of a connected AF_UNIX socket. This
5290 function call is particularly useful when implementing
5291 delegated subtrees support in the control group hierarchy.
5293 * The "sd-event" event loop API of libsystemd now supports
5294 correct dequeuing of real-time signals, without losing
5297 * When systemd requests a polkit decision when managing units it
5298 will now add additional fields to the request, including unit
5299 name and desired operation. This enables more powerful polkit
5300 policies, that make decisions depending on these parameters.
5302 * nspawn learnt support for .nspawn settings files, that may
5303 accompany the image files or directories of containers, and
5304 may contain additional settings for the container. This is
5305 an alternative to configuring container parameters via the
5306 nspawn command line.
5308 Contributions from: Cristian Rodríguez, Daniel Mack, David
5309 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
5310 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
5311 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
5312 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
5313 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
5314 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
5315 Andersen, Tom Gundersen, Torstein Husebø
5317 — Berlin, 2015-09-08
5321 * machinectl gained a new verb 'shell' which opens a fresh
5322 shell on the target container or the host. It is similar to
5323 the existing 'login' command of machinectl, but spawns the
5324 shell directly without prompting for username or
5325 password. The pseudo machine '.host' now refers to the local
5326 host and is used by default. Hence, 'machinectl shell' can
5327 be used as replacement for 'su -' which spawns a session as
5328 a fresh systemd unit in a way that is fully isolated from
5329 the originating session.
5331 * systemd-networkd learned to cope with private-zone DHCP
5332 options and allows other programs to query the values.
5334 * SELinux access control when enabling/disabling units is no
5335 longer enforced with this release. The previous implementation
5336 was incorrect, and a new corrected implementation is not yet
5337 available. As unit file operations are still protected via
5338 polkit and D-Bus policy this is not a security problem. Yet,
5339 distributions which care about optimal SELinux support should
5340 probably not stabilize on this release.
5342 * sd-bus gained support for matches of type "arg0has=", that
5343 test for membership of strings in string arrays sent in bus
5346 * systemd-resolved now dumps the contents of its DNS and LLMNR
5347 caches to the logs on reception of the SIGUSR1 signal. This
5348 is useful to debug DNS behaviour.
5350 * The coredumpctl tool gained a new --directory= option to
5351 operate on journal files in a specific directory.
5353 * "systemctl reboot" and related commands gained a new
5354 "--message=" option which may be used to set a free-text
5355 wall message when shutting down or rebooting the
5356 system. This message is also logged, which is useful for
5357 figuring out the reason for a reboot or shutdown a
5360 * The "systemd-resolve-host" tool's -i switch now takes
5361 network interface numbers as alternative to interface names.
5363 * A new unit file setting for services has been introduced:
5364 UtmpMode= allows configuration of how precisely systemd
5365 handles utmp and wtmp entries for the service if this is
5366 enabled. This allows writing services that appear similar to
5367 user sessions in the output of the "w", "who", "last" and
5370 * systemd-resolved will now locally synthesize DNS resource
5371 records for the "localhost" and "gateway" domains as well as
5372 the local hostname. This should ensure that clients querying
5373 RRs via resolved will get similar results as those going via
5374 NSS, if nss-myhostname is enabled.
5376 Contributions from: Alastair Hughes, Alex Crawford, Daniel
5377 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
5378 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
5379 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
5380 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
5381 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
5382 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
5383 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
5384 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
5385 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
5386 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
5387 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
5389 — Berlin, 2015-08-27
5393 * The systemd-efi-boot-generator functionality was merged into
5394 systemd-gpt-auto-generator.
5396 * systemd-networkd now supports Group Policy for vxlan
5397 devices. It can be enabled via the new boolean configuration
5398 option called 'GroupPolicyExtension='.
5400 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
5401 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
5402 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
5404 — Berlin, 2015-07-31
5408 * The python-systemd code has been removed from the systemd repository.
5409 A new repository has been created which accommodates the code from
5410 now on, and we kindly ask distributions to create a separate package
5411 for this: https://github.com/systemd/python-systemd
5413 * The systemd daemon will now reload its main configuration
5414 (/etc/systemd/system.conf) on daemon-reload.
5416 * sd-dhcp now exposes vendor specific extensions via
5417 sd_dhcp_lease_get_vendor_specific().
5419 * systemd-networkd gained a number of new configuration options.
5421 - A new boolean configuration option for TAP devices called
5422 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
5423 device, thus allowing to send and receive GSO packets.
5425 - A new tunnel configuration option called 'CopyDSCP='.
5426 If enabled, the DSCP field of ip6 tunnels is copied into the
5427 decapsulated packet.
5429 - A set of boolean bridge configuration options were added.
5430 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
5431 and 'UnicastFlood=' are now parsed by networkd and applied to the
5432 respective bridge link device via the respective IFLA_BRPORT_*
5435 - A new string configuration option to override the hostname sent
5436 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
5437 is true, networkd will use the configured hostname instead of the
5438 system hostname when sending DHCP requests.
5440 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
5441 networkd will configure the IPv6 flow-label of the tunnel device
5442 according to RFC2460.
5444 - The 'macvtap' virtual network devices are now supported, similar to
5445 the already supported 'macvlan' devices.
5447 * systemd-resolved now implements RFC5452 to improve resilience against
5448 cache poisoning. Additionally, source port randomization is enabled
5449 by default to further protect against DNS spoofing attacks.
5451 * nss-mymachines now supports translating UIDs and GIDs of running
5452 containers with user-namespaces enabled. If a container 'foo'
5453 translates a host uid 'UID' to the container uid 'TUID', then
5454 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
5455 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
5456 mapped as 'vg-foo-TGID'.
5458 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
5459 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
5460 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
5461 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
5462 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
5463 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
5464 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
5465 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
5466 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
5467 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
5469 — Berlin, 2015-07-29
5473 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
5474 There are no known issues with current sysfs, and udev does not need
5475 or should be used to work around such bugs.
5477 * udev does no longer enable USB HID power management. Several reports
5478 indicate, that some devices cannot handle that setting.
5480 * The udev accelerometer helper was removed. The functionality
5481 is now fully included in iio-sensor-proxy. But this means,
5482 older iio-sensor-proxy versions will no longer provide
5483 accelerometer/orientation data with this systemd version.
5484 Please upgrade iio-sensor-proxy to version 1.0.
5486 * networkd gained a new configuration option IPv6PrivacyExtensions=
5487 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
5488 for Stateless Address") on selected networks.
5490 * For the sake of fewer build-time dependencies and less code in the
5491 main repository, the python bindings are about to be removed in the
5492 next release. A new repository has been created which accommodates
5493 the code from now on, and we kindly ask distributions to create a
5494 separate package for this. The removal will take place in v223.
5496 https://github.com/systemd/python-systemd
5498 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
5499 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
5500 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
5501 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
5502 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
5503 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
5504 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
5505 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
5506 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
5507 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
5509 — Berlin, 2015-07-07
5513 * The sd-bus.h and sd-event.h APIs have now been declared
5514 stable and have been added to the official interface of
5515 libsystemd.so. sd-bus implements an alternative D-Bus client
5516 library, that is relatively easy to use, very efficient and
5517 supports both classic D-Bus as well as kdbus as transport
5518 backend. sd-event is a generic event loop abstraction that
5519 is built around Linux epoll, but adds features such as event
5520 prioritization or efficient timer handling. Both APIs are good
5521 choices for C programs looking for a bus and/or event loop
5522 implementation that is minimal and does not have to be
5523 portable to other kernels.
5525 * kdbus support is no longer compile-time optional. It is now
5526 always built-in. However, it can still be disabled at
5527 runtime using the kdbus=0 kernel command line setting, and
5528 that setting may be changed to default to off, by specifying
5529 --disable-kdbus at build-time. Note though that the kernel
5530 command line setting has no effect if the kdbus.ko kernel
5531 module is not installed, in which case kdbus is (obviously)
5532 also disabled. We encourage all downstream distributions to
5533 begin testing kdbus by adding it to the kernel images in the
5534 development distributions, and leaving kdbus support in
5537 * The minimal required util-linux version has been bumped to
5540 * Support for chkconfig (--enable-chkconfig) was removed in
5541 favor of calling an abstraction tool
5542 /lib/systemd/systemd-sysv-install. This needs to be
5543 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
5544 in README for details.
5546 * If there's a systemd unit and a SysV init script for the
5547 same service name, and the user executes "systemctl enable"
5548 for it (or a related call), then this will now enable both
5549 (or execute the related operation on both), not just the
5552 * The libudev API documentation has been converted from gtkdoc
5555 * gudev has been removed from the systemd tree, it is now an
5558 * The systemd-cgtop tool learnt a new --raw switch to generate
5559 "raw" (machine parsable) output.
5561 * networkd's IPForwarding= .network file setting learnt the
5562 new setting "kernel", which ensures that networkd does not
5563 change the IP forwarding sysctl from the default kernel
5566 * The systemd-logind bus API now exposes a new boolean
5567 property "Docked" that reports whether logind considers the
5568 system "docked", i.e. connected to a docking station or not.
5570 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
5571 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
5572 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
5573 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
5574 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
5575 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
5576 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
5577 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
5578 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
5579 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
5580 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
5581 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
5582 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
5583 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
5584 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
5585 Fink, Zbigniew Jędrzejewski-Szmek
5587 — Berlin, 2015-06-19
5591 * The gudev library has been extracted into a separate repository
5592 available at: https://git.gnome.org/browse/libgudev/
5593 It is now managed as part of the Gnome project. Distributions
5594 are recommended to pass --disable-gudev to systemd and use
5595 gudev from the Gnome project instead. gudev is still included
5596 in systemd, for now. It will be removed soon, though. Please
5597 also see the announcement-thread on systemd-devel:
5598 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
5600 * systemd now exposes a CPUUsageNSec= property for each
5601 service unit on the bus, that contains the overall consumed
5602 CPU time of a service (the sum of what each process of the
5603 service consumed). This value is only available if
5604 CPUAccounting= is turned on for a service, and is then shown
5605 in the "systemctl status" output.
5607 * Support for configuring alternative mappings of the old SysV
5608 runlevels to systemd targets has been removed. They are now
5609 hardcoded in a way that runlevels 2, 3, 4 all map to
5610 multi-user.target and 5 to graphical.target (which
5611 previously was already the default behaviour).
5613 * The auto-mounter logic gained support for mount point
5614 expiry, using a new TimeoutIdleSec= setting in .automount
5615 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
5617 * The EFI System Partition (ESP) as mounted to /boot by
5618 systemd-efi-boot-generator will now be unmounted
5619 automatically after 2 minutes of not being used. This should
5620 minimize the risk of ESP corruptions.
5622 * New /etc/fstab options x-systemd.requires= and
5623 x-systemd.requires-mounts-for= are now supported to express
5624 additional dependencies for mounts. This is useful for
5625 journalling file systems that support external journal
5626 devices or overlay file systems that require underlying file
5627 systems to be mounted.
5629 * systemd does not support direct live-upgrades (via systemctl
5630 daemon-reexec) from versions older than v44 anymore. As no
5631 distribution we are aware of shipped such old versions in a
5632 stable release this should not be problematic.
5634 * When systemd forks off a new per-connection service instance
5635 it will now set the $REMOTE_ADDR environment variable to the
5636 remote IP address, and $REMOTE_PORT environment variable to
5637 the remote IP port. This behaviour is similar to the
5638 corresponding environment variables defined by CGI.
5640 * systemd-networkd gained support for uplink failure
5641 detection. The BindCarrier= option allows binding interface
5642 configuration dynamically to the link sense of other
5643 interfaces. This is useful to achieve behaviour like in
5646 * systemd-networkd gained support for configuring the DHCP
5647 client identifier to use when requesting leases.
5649 * systemd-networkd now has a per-network UseNTP= option to
5650 configure whether NTP server information acquired via DHCP
5651 is passed on to services like systemd-timesyncd.
5653 * systemd-networkd gained support for vti6 tunnels.
5655 * Note that systemd-networkd manages the sysctl variable
5656 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
5657 it is configured for since v219. The variable controls IP
5658 forwarding, and is a per-interface alternative to the global
5659 /proc/sys/net/ipv[46]/ip_forward. This setting is
5660 configurable in the IPForward= option, which defaults to
5661 "no". This means if networkd is used for an interface it is
5662 no longer sufficient to set the global sysctl option to turn
5663 on IP forwarding! Instead, the .network file option
5664 IPForward= needs to be turned on! Note that the
5665 implementation of this behaviour was broken in v219 and has
5668 * Many bonding and vxlan options are now configurable in
5671 * systemd-nspawn gained a new --property= setting to set unit
5672 properties for the container scope. This is useful for
5673 setting resource parameters (e.g. "CPUShares=500") on
5674 containers started from the command line.
5676 * systemd-nspawn gained a new --private-users= switch to make
5677 use of user namespacing available on recent Linux kernels.
5679 * systemd-nspawn may now be called as part of a shell pipeline
5680 in which case the pipes used for stdin and stdout are passed
5681 directly to the process invoked in the container, without
5682 indirection via a pseudo tty.
5684 * systemd-nspawn gained a new switch to control the UNIX
5685 signal to use when killing the init process of the container
5688 * systemd-nspawn gained a new --overlay= switch for mounting
5689 overlay file systems into the container using the new kernel
5692 * When a container image is imported via systemd-importd and
5693 the host file system is not btrfs, a loopback block device
5694 file is created in /var/lib/machines.raw with a btrfs file
5695 system inside. It is then mounted to /var/lib/machines to
5696 enable btrfs features for container management. The loopback
5697 file and btrfs file system is grown as needed when container
5698 images are imported via systemd-importd.
5700 * systemd-machined/systemd-importd gained support for btrfs
5701 quota, to enforce container disk space limits on disk. This
5702 is exposed in "machinectl set-limit".
5704 * systemd-importd now can import containers from local .tar,
5705 .raw and .qcow2 images, and export them to .tar and .raw. It
5706 can also import dkr v2 images now from the network (on top
5709 * systemd-importd gained support for verifying downloaded
5710 images with gpg2 (previously only gpg1 was supported).
5712 * systemd-machined, systemd-logind, systemd: most bus calls are
5713 now accessible to unprivileged processes via polkit. Also,
5714 systemd-logind will now allow users to kill their own sessions
5715 without further privileges or authorization.
5717 * systemd-shutdownd has been removed. This service was
5718 previously responsible for implementing scheduled shutdowns
5719 as exposed in /usr/bin/shutdown's time parameter. This
5720 functionality has now been moved into systemd-logind and is
5721 accessible via a bus interface.
5723 * "systemctl reboot" gained a new switch --firmware-setup that
5724 can be used to reboot into the EFI firmware setup, if that
5725 is available. systemd-logind now exposes an API on the bus
5726 to trigger such reboots, in case graphical desktop UIs want
5727 to cover this functionality.
5729 * "systemctl enable", "systemctl disable" and "systemctl mask"
5730 now support a new "--now" switch. If specified the units
5731 that are enabled will also be started, and the ones
5732 disabled/masked also stopped.
5734 * The Gummiboot EFI boot loader tool has been merged into
5735 systemd, and renamed to "systemd-boot". The bootctl tool has been
5736 updated to support systemd-boot.
5738 * An EFI kernel stub has been added that may be used to create
5739 kernel EFI binaries that contain not only the actual kernel,
5740 but also an initrd, boot splash, command line and OS release
5741 information. This combined binary can then be signed as a
5742 single image, so that the firmware can verify it all in one
5743 step. systemd-boot has special support for EFI binaries created
5744 like this and can extract OS release information from them
5745 and show them in the boot menu. This functionality is useful
5746 to implement cryptographically verified boot schemes.
5748 * Optional support has been added to systemd-fsck to pass
5749 fsck's progress report to an AF_UNIX socket in the file
5752 * udev will no longer create device symlinks for all block
5753 devices by default. A blacklist for excluding special block
5754 devices from this logic has been turned into a whitelist
5755 that requires picking block devices explicitly that require
5758 * A new (currently still internal) API sd-device.h has been
5759 added to libsystemd. This modernized API is supposed to
5760 replace libudev eventually. In fact, already much of libudev
5761 is now just a wrapper around sd-device.h.
5763 * A new hwdb database for storing metadata about pointing
5764 stick devices has been added.
5766 * systemd-tmpfiles gained support for setting file attributes
5767 similar to the "chattr" tool with new 'h' and 'H' lines.
5769 * systemd-journald will no longer unconditionally set the
5770 btrfs NOCOW flag on new journal files. This is instead done
5771 with tmpfiles snippet using the new 'h' line type. This
5772 allows easy disabling of this logic, by masking the
5773 journal-nocow.conf tmpfiles file.
5775 * systemd-journald will now translate audit message types to
5776 human readable identifiers when writing them to the
5777 journal. This should improve readability of audit messages.
5779 * The LUKS logic gained support for the offset= and skip=
5780 options in /etc/crypttab, as previously implemented by
5783 * /usr/lib/os-release gained a new optional field VARIANT= for
5784 distributions that support multiple variants (such as a
5785 desktop edition, a server edition, ...)
5787 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
5788 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
5789 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
5790 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
5791 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
5792 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
5793 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
5794 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
5795 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
5796 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
5797 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
5798 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
5799 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
5800 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
5801 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
5802 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
5803 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
5804 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
5805 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
5806 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
5807 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
5808 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
5809 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
5810 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
5811 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
5812 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
5813 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
5815 — Berlin, 2015-05-22
5819 * Introduce a new API "sd-hwdb.h" for querying the hardware
5820 metadata database. With this minimal interface one can query
5821 and enumerate the udev hwdb, decoupled from the old libudev
5822 library. libudev's interface for this is now only a wrapper
5823 around sd-hwdb. A new tool systemd-hwdb has been added to
5824 interface with and update the database.
5826 * When any of systemd's tools copies files (for example due to
5827 tmpfiles' C lines) a btrfs reflink will attempted first,
5828 before bytewise copying is done.
5830 * systemd-nspawn gained a new --ephemeral switch. When
5831 specified a btrfs snapshot is taken of the container's root
5832 directory, and immediately removed when the container
5833 terminates again. Thus, a container can be started whose
5834 changes never alter the container's root directory, and are
5835 lost on container termination. This switch can also be used
5836 for starting a container off the root file system of the
5837 host without affecting the host OS. This switch is only
5838 available on btrfs file systems.
5840 * systemd-nspawn gained a new --template= switch. It takes the
5841 path to a container tree to use as template for the tree
5842 specified via --directory=, should that directory be
5843 missing. This allows instantiating containers dynamically,
5844 on first run. This switch is only available on btrfs file
5847 * When a .mount unit refers to a mount point on which multiple
5848 mounts are stacked, and the .mount unit is stopped all of
5849 the stacked mount points will now be unmounted until no
5850 mount point remains.
5852 * systemd now has an explicit notion of supported and
5853 unsupported unit types. Jobs enqueued for unsupported unit
5854 types will now fail with an "unsupported" error code. More
5855 specifically .swap, .automount and .device units are not
5856 supported in containers, .busname units are not supported on
5857 non-kdbus systems. .swap and .automount are also not
5858 supported if their respective kernel compile time options
5861 * machinectl gained support for two new "copy-from" and
5862 "copy-to" commands for copying files from a running
5863 container to the host or vice versa.
5865 * machinectl gained support for a new "bind" command to bind
5866 mount host directories into local containers. This is
5867 currently only supported for nspawn containers.
5869 * networkd gained support for configuring bridge forwarding
5870 database entries (fdb) from .network files.
5872 * A new tiny daemon "systemd-importd" has been added that can
5873 download container images in tar, raw, qcow2 or dkr formats,
5874 and make them available locally in /var/lib/machines, so
5875 that they can run as nspawn containers. The daemon can GPG
5876 verify the downloads (not supported for dkr, since it has no
5877 provisions for verifying downloads). It will transparently
5878 decompress bz2, xz, gzip compressed downloads if necessary,
5879 and restore sparse files on disk. The daemon uses privilege
5880 separation to ensure the actual download logic runs with
5881 fewer privileges than the daemon itself. machinectl has
5882 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
5883 make the functionality of importd available to the
5884 user. With this in place the Fedora and Ubuntu "Cloud"
5885 images can be downloaded and booted as containers unmodified
5886 (the Fedora images lack the appropriate GPG signature files
5887 currently, so they cannot be verified, but this will change
5888 soon, hopefully). Note that downloading images is currently
5889 only fully supported on btrfs.
5891 * machinectl is now able to list container images found in
5892 /var/lib/machines, along with some metadata about sizes of
5893 disk and similar. If the directory is located on btrfs and
5894 quota is enabled, this includes quota display. A new command
5895 "image-status" has been added that shows additional
5896 information about images.
5898 * machinectl is now able to clone container images
5899 efficiently, if the underlying file system (btrfs) supports
5900 it, with the new "machinectl clone" command. It also
5901 gained commands for renaming and removing images, as well as
5902 marking them read-only or read-write (supported also on
5903 legacy file systems).
5905 * networkd gained support for collecting LLDP network
5906 announcements, from hardware that supports this. This is
5907 shown in networkctl output.
5909 * systemd-run gained support for a new -t (--pty) switch for
5910 invoking a binary on a pty whose input and output is
5911 connected to the invoking terminal. This allows executing
5912 processes as system services while interactively
5913 communicating with them via the terminal. Most interestingly
5914 this is supported across container boundaries. Invoking
5915 "systemd-run -t /bin/bash" is an alternative to running a
5916 full login session, the difference being that the former
5917 will not register a session, nor go through the PAM session
5920 * tmpfiles gained support for a new "v" line type for creating
5921 btrfs subvolumes. If the underlying file system is a legacy
5922 file system, this automatically degrades to creating a
5923 normal directory. Among others /var/lib/machines is now
5924 created like this at boot, should it be missing.
5926 * The directory /var/lib/containers/ has been deprecated and
5927 been replaced by /var/lib/machines. The term "machines" has
5928 been used in the systemd context as generic term for both
5929 VMs and containers, and hence appears more appropriate for
5930 this, as the directory can also contain raw images bootable
5933 * systemd-nspawn when invoked with -M but without --directory=
5934 or --image= is now capable of searching for the container
5935 root directory, subvolume or disk image automatically, in
5936 /var/lib/machines. systemd-nspawn@.service has been updated
5937 to make use of this, thus allowing it to be used for raw
5940 * A new machines.target unit has been introduced that is
5941 supposed to group all containers/VMs invoked as services on
5942 the system. systemd-nspawn@.service has been updated to
5943 integrate with that.
5945 * machinectl gained a new "start" command, for invoking a
5946 container as a service. "machinectl start foo" is mostly
5947 equivalent to "systemctl start systemd-nspawn@foo.service",
5948 but handles escaping in a nicer way.
5950 * systemd-nspawn will now mount most of the cgroupfs tree
5951 read-only into each container, with the exception of the
5952 container's own subtree in the name=systemd hierarchy.
5954 * journald now sets the special FS_NOCOW file flag for its
5955 journal files. This should improve performance on btrfs, by
5956 avoiding heavy fragmentation when journald's write-pattern
5957 is used on COW file systems. It degrades btrfs' data
5958 integrity guarantees for the files to the same levels as for
5959 ext3/ext4 however. This should be OK though as journald does
5960 its own data integrity checks and all its objects are
5961 checksummed on disk. Also, journald should handle btrfs disk
5962 full events a lot more gracefully now, by processing SIGBUS
5963 errors, and not relying on fallocate() anymore.
5965 * When journald detects that journal files it is writing to
5966 have been deleted it will immediately start new journal
5969 * systemd now provides a way to store file descriptors
5970 per-service in PID 1. This is useful for daemons to ensure
5971 that fds they require are not lost during a daemon
5972 restart. The fds are passed to the daemon on the next
5973 invocation in the same way socket activation fds are
5974 passed. This is now used by journald to ensure that the
5975 various sockets connected to all the system's stdout/stderr
5976 are not lost when journald is restarted. File descriptors
5977 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
5978 an extension to sd_notify(). Note that a limit is enforced
5979 on the number of fds a service can store in PID 1, and it
5980 defaults to 0, so that no fds may be stored, unless this is
5981 explicitly turned on.
5983 * The default TERM variable to use for units connected to a
5984 terminal, when no other value is explicitly is set is now
5985 vt220 rather than vt102. This should be fairly safe still,
5986 but allows PgUp/PgDn work.
5988 * The /etc/crypttab option header= as known from Debian is now
5991 * "loginctl user-status" and "loginctl session-status" will
5992 now show the last 10 lines of log messages of the
5993 user/session following the status output. Similar,
5994 "machinectl status" will show the last 10 log lines
5995 associated with a virtual machine or container
5996 service. (Note that this is usually not the log messages
5997 done in the VM/container itself, but simply what the
5998 container manager logs. For nspawn this includes all console
6001 * "loginctl session-status" without further argument will now
6002 show the status of the session of the caller. Similar,
6003 "lock-session", "unlock-session", "activate",
6004 "enable-linger", "disable-linger" may now be called without
6005 session/user parameter in which case they apply to the
6006 caller's session/user.
6008 * An X11 session scriptlet is now shipped that uploads
6009 $DISPLAY and $XAUTHORITY into the environment of the systemd
6010 --user daemon if a session begins. This should improve
6011 compatibility with X11 enabled applications run as systemd
6014 * Generators are now subject to masking via /etc and /run, the
6015 same way as unit files.
6017 * networkd .network files gained support for configuring
6018 per-link IPv4/IPv6 packet forwarding as well as IPv4
6019 masquerading. This is by default turned on for veth links to
6020 containers, as registered by systemd-nspawn. This means that
6021 nspawn containers run with --network-veth will now get
6022 automatic routed access to the host's networks without any
6023 further configuration or setup, as long as networkd runs on
6026 * systemd-nspawn gained the --port= (-p) switch to expose TCP
6027 or UDP posts of a container on the host. With this in place
6028 it is possible to run containers with private veth links
6029 (--network-veth), and have their functionality exposed on
6030 the host as if their services were running directly on the
6033 * systemd-nspawn's --network-veth switch now gained a short
6034 version "-n", since with the changes above it is now truly
6035 useful out-of-the-box. The systemd-nspawn@.service has been
6036 updated to make use of it too by default.
6038 * systemd-nspawn will now maintain a per-image R/W lock, to
6039 ensure that the same image is not started more than once
6040 writable. (It's OK to run an image multiple times
6041 simultaneously in read-only mode.)
6043 * systemd-nspawn's --image= option is now capable of
6044 dissecting and booting MBR and GPT disk images that contain
6045 only a single active Linux partition. Previously it
6046 supported only GPT disk images with proper GPT type
6047 IDs. This allows running cloud images from major
6048 distributions directly with systemd-nspawn, without
6051 * In addition to collecting mouse dpi data in the udev
6052 hardware database, there's now support for collecting angle
6053 information for mouse scroll wheels. The database is
6054 supposed to guarantee similar scrolling behavior on mice
6055 that it knows about. There's also support for collecting
6056 information about Touchpad types.
6058 * udev's input_id built-in will now also collect touch screen
6059 dimension data and attach it to probed devices.
6061 * /etc/os-release gained support for a Distribution Privacy
6064 * networkd gained support for creating "ipvlan", "gretap",
6065 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
6067 * systemd-tmpfiles gained support for "a" lines for setting
6070 * systemd-nspawn will now mount /tmp in the container to
6071 tmpfs, automatically.
6073 * systemd now exposes the memory.usage_in_bytes cgroup
6074 attribute and shows it for each service in the "systemctl
6075 status" output, if available.
6077 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
6078 immediate reboot is triggered. This useful if shutdown is
6079 hung and is unable to complete, to expedite the
6080 operation. Note that this kind of reboot will still unmount
6081 all file systems, and hence should not result in fsck being
6084 * A .device unit for an optical block device will now be
6085 considered active only when a medium is in the drive. Also,
6086 mount units are now bound to their backing devices thus
6087 triggering automatic unmounting when devices become
6088 unavailable. With this in place systemd will now
6089 automatically unmount left-over mounts when a CD-ROM is
6090 ejected or an USB stick is yanked from the system.
6092 * networkd-wait-online now has support for waiting for
6093 specific interfaces only (with globbing), and for giving up
6094 after a configurable timeout.
6096 * networkd now exits when idle. It will be automatically
6097 restarted as soon as interfaces show up, are removed or
6098 change state. networkd will stay around as long as there is
6099 at least one DHCP state machine or similar around, that keep
6102 * networkd may now configure IPv6 link-local addressing in
6103 addition to IPv4 link-local addressing.
6105 * The IPv6 "token" for use in SLAAC may now be configured for
6106 each .network interface in networkd.
6108 * Routes configured with networkd may now be assigned a scope
6111 * networkd's [Match] sections now support globbing and lists
6112 of multiple space-separated matches per item.
6114 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
6115 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
6116 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
6117 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
6118 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
6119 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
6120 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
6121 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
6122 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
6123 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
6124 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
6125 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
6126 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
6127 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
6128 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
6129 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
6130 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
6131 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
6132 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
6133 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
6134 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
6135 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
6136 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
6137 Hoffmann, Zbigniew Jędrzejewski-Szmek
6139 — Berlin, 2015-02-16
6143 * When querying unit file enablement status (for example via
6144 "systemctl is-enabled"), a new state "indirect" is now known
6145 which indicates that a unit might not be enabled itself, but
6146 another unit listed in its Also= setting might be.
6148 * Similar to the various existing ConditionXYZ= settings for
6149 units, there are now matching AssertXYZ= settings. While
6150 failing conditions cause a unit to be skipped, but its job
6151 to succeed, failing assertions declared like this will cause
6152 a unit start operation and its job to fail.
6154 * hostnamed now knows a new chassis type "embedded".
6156 * systemctl gained a new "edit" command. When used on a unit
6157 file, this allows extending unit files with .d/ drop-in
6158 configuration snippets or editing the full file (after
6159 copying it from /usr/lib to /etc). This will invoke the
6160 user's editor (as configured with $EDITOR), and reload the
6161 modified configuration after editing.
6163 * "systemctl status" now shows the suggested enablement state
6164 for a unit, as declared in the (usually vendor-supplied)
6165 system preset files.
6167 * nss-myhostname will now resolve the single-label hostname
6168 "gateway" to the locally configured default IP routing
6169 gateways, ordered by their metrics. This assigns a stable
6170 name to the used gateways, regardless which ones are
6171 currently configured. Note that the name will only be
6172 resolved after all other name sources (if nss-myhostname is
6173 configured properly) and should hence not negatively impact
6174 systems that use the single-label hostname "gateway" in
6177 * systemd-inhibit now allows filtering by mode when listing
6180 * Scope and service units gained a new "Delegate" boolean
6181 property, which, when set, allows processes running inside the
6182 unit to further partition resources. This is primarily
6183 useful for systemd user instances as well as container
6186 * journald will now pick up audit messages directly from
6187 the kernel, and log them like any other log message. The
6188 audit fields are split up and fully indexed. This means that
6189 journalctl in many ways is now a (nicer!) alternative to
6190 ausearch, the traditional audit client. Note that this
6191 implements only a minimal audit client. If you want the
6192 special audit modes like reboot-on-log-overflow, please use
6193 the traditional auditd instead, which can be used in
6194 parallel to journald.
6196 * The ConditionSecurity= unit file option now understands the
6197 special string "audit" to check whether auditing is
6200 * journalctl gained two new commands --vacuum-size= and
6201 --vacuum-time= to delete old journal files until the
6202 remaining ones take up no more than the specified size on disk,
6203 or are not older than the specified time.
6205 * A new, native PPPoE library has been added to sd-network,
6206 systemd's library of light-weight networking protocols. This
6207 library will be used in a future version of networkd to
6208 enable PPPoE communication without an external pppd daemon.
6210 * The busctl tool now understands a new "capture" verb that
6211 works similar to "monitor", but writes a packet capture
6212 trace to STDOUT that can be redirected to a file which is
6213 compatible with libcap's capture file format. This can then
6214 be loaded in Wireshark and similar tools to inspect bus
6217 * The busctl tool now understands a new "tree" verb that shows
6218 the object trees of a specific service on the bus, or of all
6221 * The busctl tool now understands a new "introspect" verb that
6222 shows all interfaces and members of objects on the bus,
6223 including their signature and values. This is particularly
6224 useful to get more information about bus objects shown by
6225 the new "busctl tree" command.
6227 * The busctl tool now understands new verbs "call",
6228 "set-property" and "get-property" for invoking bus method
6229 calls, setting and getting bus object properties in a
6232 * busctl gained a new --augment-creds= argument that controls
6233 whether the tool shall augment credential information it
6234 gets from the bus with data from /proc, in a possibly
6237 * nspawn's --link-journal= switch gained two new values
6238 "try-guest" and "try-host" that work like "guest" and
6239 "host", but do not fail if the host has no persistent
6240 journalling enabled. -j is now equivalent to
6241 --link-journal=try-guest.
6243 * macvlan network devices created by nspawn will now have
6244 stable MAC addresses.
6246 * A new SmackProcessLabel= unit setting has been added, which
6247 controls the SMACK security label processes forked off by
6248 the respective unit shall use.
6250 * If compiled with --enable-xkbcommon, systemd-localed will
6251 verify x11 keymap settings by compiling the given keymap. It
6252 will spew out warnings if the compilation fails. This
6253 requires libxkbcommon to be installed.
6255 * When a coredump is collected, a larger number of metadata
6256 fields is now collected and included in the journal records
6257 created for it. More specifically, control group membership,
6258 environment variables, memory maps, working directory,
6259 chroot directory, /proc/$PID/status, and a list of open file
6260 descriptors is now stored in the log entry.
6262 * The udev hwdb now contains DPI information for mice. For
6265 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
6267 * All systemd programs that read standalone configuration
6268 files in /etc now also support a corresponding series of
6269 .conf.d configuration directories in /etc/, /run/,
6270 /usr/local/lib/, /usr/lib/, and (if configured with
6271 --enable-split-usr) /lib/. In particular, the following
6272 configuration files now have corresponding configuration
6273 directories: system.conf user.conf, logind.conf,
6274 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
6275 resolved.conf, timesyncd.conf, journal-remote.conf, and
6276 journal-upload.conf. Note that distributions should use the
6277 configuration directories in /usr/lib/; the directories in
6278 /etc/ are reserved for the system administrator.
6280 * systemd-rfkill will no longer take the rfkill device name
6281 into account when storing rfkill state on disk, as the name
6282 might be dynamically assigned and not stable. Instead, the
6283 ID_PATH udev variable combined with the rfkill type (wlan,
6284 bluetooth, ...) is used.
6286 * A new service systemd-machine-id-commit.service has been
6287 added. When used on systems where /etc is read-only during
6288 boot, and /etc/machine-id is not initialized (but an empty
6289 file), this service will copy the temporary machine ID
6290 created as replacement into /etc after the system is fully
6291 booted up. This is useful for systems that are freshly
6292 installed with a non-initialized machine ID, but should get
6293 a fixed machine ID for subsequent boots.
6295 * networkd's .netdev files now provide a large set of
6296 configuration parameters for VXLAN devices. Similarly, the
6297 bridge port cost parameter is now configurable in .network
6298 files. There's also new support for configuring IP source
6299 routing. networkd .link files gained support for a new
6300 OriginalName= match that is useful to match against the
6301 original interface name the kernel assigned. .network files
6302 may include MTU= and MACAddress= fields for altering the MTU
6303 and MAC address while being connected to a specific network
6306 * The LUKS logic gained supported for configuring
6307 UUID-specific key files. There's also new support for naming
6308 LUKS device from the kernel command line, using the new
6309 luks.name= argument.
6311 * Timer units may now be transiently created via the bus API
6312 (this was previously already available for scope and service
6313 units). In addition it is now possible to create multiple
6314 transient units at the same time with a single bus call. The
6315 "systemd-run" tool has been updated to make use of this for
6316 running commands on a specified time, in at(1)-style.
6318 * tmpfiles gained support for "t" lines, for assigning
6319 extended attributes to files. Among other uses this may be
6320 used to assign SMACK labels to files.
6322 Contributions from: Alin Rauta, Alison Chaiken, Andrej
6323 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
6324 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
6325 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
6326 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
6327 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
6328 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
6329 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
6330 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
6331 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
6332 Michael Biebl, Michael Chapman, Michael Marineau, Michal
6333 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
6334 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
6335 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
6336 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
6337 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
6338 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
6339 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
6341 — Berlin, 2014-12-10
6345 * journalctl gained the new options -t/--identifier= to match
6346 on the syslog identifier (aka "tag"), as well as --utc to
6347 show log timestamps in the UTC timezone. journalctl now also
6348 accepts -n/--lines=all to disable line capping in a pager.
6350 * journalctl gained a new switch, --flush, that synchronously
6351 flushes logs from /run/log/journal to /var/log/journal if
6352 persistent storage is enabled. systemd-journal-flush.service
6353 now waits until the operation is complete.
6355 * Services can notify the manager before they start a reload
6356 (by sending RELOADING=1) or shutdown (by sending
6357 STOPPING=1). This allows the manager to track and show the
6358 internal state of daemons and closes a race condition when
6359 the process is still running but has closed its D-Bus
6362 * Services with Type=oneshot do not have to have any ExecStart
6365 * User units are now loaded also from
6366 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
6367 /run/systemd/user directory that was already previously
6368 supported, but is under the control of the user.
6370 * Job timeouts (i.e. timeouts on the time a job that is
6371 queued stays in the run queue) can now optionally result in
6372 immediate reboot or power-off actions (JobTimeoutAction= and
6373 JobTimeoutRebootArgument=). This is useful on ".target"
6374 units, to limit the maximum time a target remains
6375 undispatched in the run queue, and to trigger an emergency
6376 operation in such a case. This is now used by default to
6377 turn off the system if boot-up (as defined by everything in
6378 basic.target) hangs and does not complete for at least
6379 15min. Also, if power-off or reboot hang for at least 30min
6380 an immediate power-off/reboot operation is triggered. This
6381 functionality is particularly useful to increase reliability
6382 on embedded devices, but also on laptops which might
6383 accidentally get powered on when carried in a backpack and
6384 whose boot stays stuck in a hard disk encryption passphrase
6387 * systemd-logind can be configured to also handle lid switch
6388 events even when the machine is docked or multiple displays
6389 are attached (HandleLidSwitchDocked= option).
6391 * A helper binary and a service have been added which can be
6392 used to resume from hibernation in the initramfs. A
6393 generator will parse the resume= option on the kernel
6394 command line to trigger resume.
6396 * A user console daemon systemd-consoled has been
6397 added. Currently, it is a preview, and will so far open a
6398 single terminal on each session of the user marked as
6399 Desktop=systemd-console.
6401 * Route metrics can be specified for DHCP routes added by
6404 * The SELinux context of socket-activated services can be set
6405 from the information provided by the networking stack
6406 (SELinuxContextFromNet= option).
6408 * Userspace firmware loading support has been removed and
6409 the minimum supported kernel version is thus bumped to 3.7.
6411 * Timeout for udev workers has been increased from 1 to 3
6412 minutes, but a warning will be printed after 1 minute to
6413 help diagnose kernel modules that take a long time to load.
6415 * Udev rules can now remove tags on devices with TAG-="foobar".
6417 * systemd's readahead implementation has been removed. In many
6418 circumstances it didn't give expected benefits even for
6419 rotational disk drives and was becoming less relevant in the
6420 age of SSDs. As none of the developers has been using
6421 rotating media anymore, and nobody stepped up to actively
6422 maintain this component of systemd it has now been removed.
6424 * Swap units can use Options= to specify discard options.
6425 Discard options specified for swaps in /etc/fstab are now
6428 * Docker containers are now detected as a separate type of
6431 * The Password Agent protocol gained support for queries where
6432 the user input is shown, useful e.g. for user names.
6433 systemd-ask-password gained a new --echo option to turn that
6436 * The default sysctl.d/ snippets will now set:
6438 net.core.default_qdisc = fq_codel
6440 This selects Fair Queuing Controlled Delay as the default
6441 queuing discipline for network interfaces. fq_codel helps
6442 fight the network bufferbloat problem. It is believed to be
6443 a good default with no tuning required for most workloads.
6444 Downstream distributions may override this choice. On 10Gbit
6445 servers that do not do forwarding, "fq" may perform better.
6446 Systems without a good clocksource should use "pfifo_fast".
6448 * If kdbus is enabled during build a new option BusPolicy= is
6449 available for service units, that allows locking all service
6450 processes into a stricter bus policy, in order to limit
6451 access to various bus services, or even hide most of them
6452 from the service's view entirely.
6454 * networkctl will now show the .network and .link file
6455 networkd has applied to a specific interface.
6457 * sd-login gained a new API call sd_session_get_desktop() to
6458 query which desktop environment has been selected for a
6461 * UNIX utmp support is now compile-time optional to support
6462 legacy-free systems.
6464 * systemctl gained two new commands "add-wants" and
6465 "add-requires" for pulling in units from specific targets
6468 * If the word "rescue" is specified on the kernel command line
6469 the system will now boot into rescue mode (aka
6470 rescue.target), which was previously available only by
6471 specifying "1" or "systemd.unit=rescue.target" on the kernel
6472 command line. This new kernel command line option nicely
6473 mirrors the already existing "emergency" kernel command line
6476 * New kernel command line options mount.usr=, mount.usrflags=,
6477 mount.usrfstype= have been added that match root=, rootflags=,
6478 rootfstype= but allow mounting a specific file system to
6481 * The $NOTIFY_SOCKET is now also passed to control processes of
6482 services, not only the main process.
6484 * This version reenables support for fsck's -l switch. This
6485 means at least version v2.25 of util-linux is required for
6486 operation, otherwise dead-locks on device nodes may
6487 occur. Again: you need to update util-linux to at least
6488 v2.25 when updating systemd to v217.
6490 * The "multi-seat-x" tool has been removed from systemd, as
6491 its functionality has been integrated into X servers 1.16,
6492 and the tool is hence redundant. It is recommended to update
6493 display managers invoking this tool to simply invoke X
6494 directly from now on, again.
6496 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
6497 message flag has been added for all of systemd's polkit
6498 authenticated method calls has been added. In particular this
6499 now allows optional interactive authorization via polkit for
6500 many of PID1's privileged operations such as unit file
6501 enabling and disabling.
6503 * "udevadm hwdb --update" learnt a new switch "--usr" for
6504 placing the rebuilt hardware database in /usr instead of
6505 /etc. When used only hardware database entries stored in
6506 /usr will be used, and any user database entries in /etc are
6507 ignored. This functionality is useful for vendors to ship a
6508 pre-built database on systems where local configuration is
6509 unnecessary or unlikely.
6511 * Calendar time specifications in .timer units now also
6512 understand the strings "semi-annually", "quarterly" and
6513 "minutely" as shortcuts (in addition to the preexisting
6514 "annually", "hourly", ...).
6516 * systemd-tmpfiles will now correctly create files in /dev
6517 at boot which are marked for creation only at boot. It is
6518 recommended to always create static device nodes with 'c!'
6519 and 'b!', so that they are created only at boot and not
6520 overwritten at runtime.
6522 * When the watchdog logic is used for a service (WatchdogSec=)
6523 and the watchdog timeout is hit the service will now be
6524 terminated with SIGABRT (instead of just SIGTERM), in order
6525 to make sure a proper coredump and backtrace is
6526 generated. This ensures that hanging services will result in
6527 similar coredump/backtrace behaviour as services that hit a
6530 Contributions from: Andreas Henriksson, Andrei Borzenkov,
6531 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
6532 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
6533 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
6534 Herrmann, David Sommerseth, David Strauss, Emil Renner
6535 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
6536 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
6537 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
6538 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
6539 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
6540 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
6541 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
6542 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
6543 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
6544 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
6545 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
6546 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
6547 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
6548 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
6549 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
6550 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
6553 — Berlin, 2014-10-28
6557 * timedated no longer reads NTP implementation unit names from
6558 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
6559 implementations should add a
6561 Conflicts=systemd-timesyncd.service
6563 to their unit files to take over and replace systemd's NTP
6564 default functionality.
6566 * systemd-sysusers gained a new line type "r" for configuring
6567 which UID/GID ranges to allocate system users/groups
6568 from. Lines of type "u" may now add an additional column
6569 that specifies the home directory for the system user to be
6570 created. Also, systemd-sysusers may now optionally read user
6571 information from STDIN instead of a file. This is useful for
6572 invoking it from RPM preinst scriptlets that need to create
6573 users before the first RPM file is installed since these
6574 files might need to be owned by them. A new
6575 %sysusers_create_inline RPM macro has been introduced to do
6576 just that. systemd-sysusers now updates the shadow files as
6577 well as the user/group databases, which should enhance
6578 compatibility with certain tools like grpck.
6580 * A number of bus APIs of PID 1 now optionally consult polkit to
6581 permit access for otherwise unprivileged clients under certain
6582 conditions. Note that this currently doesn't support
6583 interactive authentication yet, but this is expected to be
6584 added eventually, too.
6586 * /etc/machine-info now has new fields for configuring the
6587 deployment environment of the machine, as well as the
6588 location of the machine. hostnamectl has been updated with
6589 new command to update these fields.
6591 * systemd-timesyncd has been updated to automatically acquire
6592 NTP server information from systemd-networkd, which might
6593 have been discovered via DHCP.
6595 * systemd-resolved now includes a caching DNS stub resolver
6596 and a complete LLMNR name resolution implementation. A new
6597 NSS module "nss-resolve" has been added which can be used
6598 instead of glibc's own "nss-dns" to resolve hostnames via
6599 systemd-resolved. Hostnames, addresses and arbitrary RRs may
6600 be resolved via systemd-resolved D-Bus APIs. In contrast to
6601 the glibc internal resolver systemd-resolved is aware of
6602 multi-homed system, and keeps DNS server and caches separate
6603 and per-interface. Queries are sent simultaneously on all
6604 interfaces that have DNS servers configured, in order to
6605 properly handle VPNs and local LANs which might resolve
6606 separate sets of domain names. systemd-resolved may acquire
6607 DNS server information from systemd-networkd automatically,
6608 which in turn might have discovered them via DHCP. A tool
6609 "systemd-resolve-host" has been added that may be used to
6610 query the DNS logic in resolved. systemd-resolved implements
6611 IDNA and automatically uses IDNA or UTF-8 encoding depending
6612 on whether classic DNS or LLMNR is used as transport. In the
6613 next releases we intend to add a DNSSEC and mDNS/DNS-SD
6614 implementation to systemd-resolved.
6616 * A new NSS module nss-mymachines has been added, that
6617 automatically resolves the names of all local registered
6618 containers to their respective IP addresses.
6620 * A new client tool "networkctl" for systemd-networkd has been
6621 added. It currently is entirely passive and will query
6622 networking configuration from udev, rtnetlink and networkd,
6623 and present it to the user in a very friendly
6624 way. Eventually, we hope to extend it to become a full
6625 control utility for networkd.
6627 * .socket units gained a new DeferAcceptSec= setting that
6628 controls the kernels' TCP_DEFER_ACCEPT sockopt for
6629 TCP. Similarly, support for controlling TCP keep-alive
6630 settings has been added (KeepAliveTimeSec=,
6631 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
6632 turning off Nagle's algorithm on TCP has been added
6635 * logind learned a new session type "web", for use in projects
6636 like Cockpit which register web clients as PAM sessions.
6638 * timer units with at least one OnCalendar= setting will now
6639 be started only after time-sync.target has been
6640 reached. This way they will not elapse before the system
6641 clock has been corrected by a local NTP client or
6642 similar. This is particular useful on RTC-less embedded
6643 machines, that come up with an invalid system clock.
6645 * systemd-nspawn's --network-veth= switch should now result in
6646 stable MAC addresses for both the outer and the inner side
6649 * systemd-nspawn gained a new --volatile= switch for running
6650 container instances with /etc or /var unpopulated.
6652 * The kdbus client code has been updated to use the new Linux
6653 3.17 memfd subsystem instead of the old kdbus-specific one.
6655 * systemd-networkd's DHCP client and server now support
6656 FORCERENEW. There are also new configuration options to
6657 configure the vendor client identifier and broadcast mode
6660 * systemd will no longer inform the kernel about the current
6661 timezone, as this is necessarily incorrect and racy as the
6662 kernel has no understanding of DST and similar
6663 concepts. This hence means FAT timestamps will be always
6664 considered UTC, similar to what Android is already
6665 doing. Also, when the RTC is configured to the local time
6666 (rather than UTC) systemd will never synchronize back to it,
6667 as this might confuse Windows at a later boot.
6669 * systemd-analyze gained a new command "verify" for offline
6670 validation of unit files.
6672 * systemd-networkd gained support for a couple of additional
6673 settings for bonding networking setups. Also, the metric for
6674 statically configured routes may now be configured. For
6675 network interfaces where this is appropriate the peer IP
6676 address may now be configured.
6678 * systemd-networkd's DHCP client will no longer request
6679 broadcasting by default, as this tripped up some networks.
6680 For hardware where broadcast is required the feature should
6681 be switched back on using RequestBroadcast=yes.
6683 * systemd-networkd will now set up IPv4LL addresses (when
6684 enabled) even if DHCP is configured successfully.
6686 * udev will now default to respect network device names given
6687 by the kernel when the kernel indicates that these are
6688 predictable. This behavior can be tweaked by changing
6689 NamePolicy= in the relevant .link file.
6691 * A new library systemd-terminal has been added that
6692 implements full TTY stream parsing and rendering. This
6693 library is supposed to be used later on for implementing a
6694 full userspace VT subsystem, replacing the current kernel
6697 * A new tool systemd-journal-upload has been added to push
6698 journal data to a remote system running
6699 systemd-journal-remote.
6701 * journald will no longer forward all local data to another
6702 running syslog daemon. This change has been made because
6703 rsyslog (which appears to be the most commonly used syslog
6704 implementation these days) no longer makes use of this, and
6705 instead pulls the data out of the journal on its own. Since
6706 forwarding the messages to a non-existent syslog server is
6707 more expensive than we assumed we have now turned this
6708 off. If you run a syslog server that is not a recent rsyslog
6709 version, you have to turn this option on again
6710 (ForwardToSyslog= in journald.conf).
6712 * journald now optionally supports the LZ4 compressor for
6713 larger journal fields. This compressor should perform much
6714 better than XZ which was the previous default.
6716 * machinectl now shows the IP addresses of local containers,
6717 if it knows them, plus the interface name of the container.
6719 * A new tool "systemd-escape" has been added that makes it
6720 easy to escape strings to build unit names and similar.
6722 * sd_notify() messages may now include a new ERRNO= field
6723 which is parsed and collected by systemd and shown among the
6724 "systemctl status" output for a service.
6726 * A new component "systemd-firstboot" has been added that
6727 queries the most basic systemd information (timezone,
6728 hostname, root password) interactively on first
6729 boot. Alternatively it may also be used to provision these
6730 things offline on OS images installed into directories.
6732 * The default sysctl.d/ snippets will now set
6734 net.ipv4.conf.default.promote_secondaries=1
6736 This has the benefit of no flushing secondary IP addresses
6737 when primary addresses are removed.
6739 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
6740 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
6741 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
6742 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
6743 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
6744 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
6745 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
6746 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
6747 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
6748 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
6749 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
6750 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
6751 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
6752 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
6753 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
6755 — Berlin, 2014-08-19
6759 * A new tool systemd-sysusers has been added. This tool
6760 creates system users and groups in /etc/passwd and
6761 /etc/group, based on static declarative system user/group
6762 definitions in /usr/lib/sysusers.d/. This is useful to
6763 enable factory resets and volatile systems that boot up with
6764 an empty /etc directory, and thus need system users and
6765 groups created during early boot. systemd now also ships
6766 with two default sysusers.d/ files for the most basic
6767 users and groups systemd and the core operating system
6770 * A new tmpfiles snippet has been added that rebuilds the
6771 essential files in /etc on boot, should they be missing.
6773 * A directive for ensuring automatic clean-up of
6774 /var/cache/man/ has been removed from the default
6775 configuration. This line should now be shipped by the man
6776 implementation. The necessary change has been made to the
6777 man-db implementation. Note that you need to update your man
6778 implementation to one that ships this line, otherwise no
6779 automatic clean-up of /var/cache/man will take place.
6781 * A new condition ConditionNeedsUpdate= has been added that
6782 may conditionalize services to only run when /etc or /var
6783 are "older" than the vendor operating system resources in
6784 /usr. This is useful for reconstructing or updating /etc
6785 after an offline update of /usr or a factory reset, on the
6786 next reboot. Services that want to run once after such an
6787 update or reset should use this condition and order
6788 themselves before the new systemd-update-done.service, which
6789 will mark the two directories as fully updated. A number of
6790 service files have been added making use of this, to rebuild
6791 the udev hardware database, the journald message catalog and
6792 dynamic loader cache (ldconfig). The systemd-sysusers tool
6793 described above also makes use of this now. With this in
6794 place it is now possible to start up a minimal operating
6795 system with /etc empty cleanly. For more information on the
6796 concepts involved see this recent blog story:
6798 http://0pointer.de/blog/projects/stateless.html
6800 * A new system group "input" has been introduced, and all
6801 input device nodes get this group assigned. This is useful
6802 for system-level software to get access to input devices. It
6803 complements what is already done for "audio" and "video".
6805 * systemd-networkd learnt minimal DHCPv4 server support in
6806 addition to the existing DHCPv4 client support. It also
6807 learnt DHCPv6 client and IPv6 Router Solicitation client
6808 support. The DHCPv4 client gained support for static routes
6809 passed in from the server. Note that the [DHCPv4] section
6810 known in older systemd-networkd versions has been renamed to
6811 [DHCP] and is now also used by the DHCPv6 client. Existing
6812 .network files using settings of this section should be
6813 updated, though compatibility is maintained. Optionally, the
6814 client hostname may now be sent to the DHCP server.
6816 * networkd gained support for vxlan virtual networks as well
6817 as tun/tap and dummy devices.
6819 * networkd gained support for automatic allocation of address
6820 ranges for interfaces from a system-wide pool of
6821 addresses. This is useful for dynamically managing a large
6822 number of interfaces with a single network configuration
6823 file. In particular this is useful to easily assign
6824 appropriate IP addresses to the veth links of a large number
6825 of nspawn instances.
6827 * RPM macros for processing sysusers, sysctl and binfmt
6828 drop-in snippets at package installation time have been
6831 * The /etc/os-release file should now be placed in
6832 /usr/lib/os-release. The old location is automatically
6833 created as symlink. /usr/lib is the more appropriate
6834 location of this file, since it shall actually describe the
6835 vendor operating system shipped in /usr, and not the
6836 configuration stored in /etc.
6838 * .mount units gained a new boolean SloppyOptions= setting
6839 that maps to mount(8)'s -s option which enables permissive
6840 parsing of unknown mount options.
6842 * tmpfiles learnt a new "L+" directive which creates a symlink
6843 but (unlike "L") deletes a pre-existing file first, should
6844 it already exist and not already be the correct
6845 symlink. Similarly, "b+", "c+" and "p+" directives have been
6846 added as well, which create block and character devices, as
6847 well as fifos in the filesystem, possibly removing any
6848 pre-existing files of different types.
6850 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
6851 'argument' field (which so far specified the source to
6852 symlink/copy the files from) is now optional. If omitted the
6853 same file os copied from /usr/share/factory/ suffixed by the
6854 full destination path. This is useful for populating /etc
6855 with essential files, by copying them from vendor defaults
6856 shipped in /usr/share/factory/etc.
6858 * A new command "systemctl preset-all" has been added that
6859 applies the service preset settings to all installed unit
6860 files. A new switch --preset-mode= has been added that
6861 controls whether only enable or only disable operations
6864 * A new command "systemctl is-system-running" has been added
6865 that allows checking the overall state of the system, for
6866 example whether it is fully up and running.
6868 * When the system boots up with an empty /etc, the equivalent
6869 to "systemctl preset-all" is executed during early boot, to
6870 make sure all default services are enabled after a factory
6873 * systemd now contains a minimal preset file that enables the
6874 most basic services systemd ships by default.
6876 * Unit files' [Install] section gained a new DefaultInstance=
6877 field for defining the default instance to create if a
6878 template unit is enabled with no instance specified.
6880 * A new passive target cryptsetup-pre.target has been added
6881 that may be used by services that need to make they run and
6882 finish before the first LUKS cryptographic device is set up.
6884 * The /dev/loop-control and /dev/btrfs-control device nodes
6885 are now owned by the "disk" group by default, opening up
6886 access to this group.
6888 * systemd-coredump will now automatically generate a
6889 stack trace of all core dumps taking place on the system,
6890 based on elfutils' libdw library. This stack trace is logged
6893 * systemd-coredump may now optionally store coredumps directly
6894 on disk (in /var/lib/systemd/coredump, possibly compressed),
6895 instead of storing them unconditionally in the journal. This
6896 mode is the new default. A new configuration file
6897 /etc/systemd/coredump.conf has been added to configure this
6898 and other parameters of systemd-coredump.
6900 * coredumpctl gained a new "info" verb to show details about a
6901 specific coredump. A new switch "-1" has also been added
6902 that makes sure to only show information about the most
6903 recent entry instead of all entries. Also, as the tool is
6904 generally useful now the "systemd-" prefix of the binary
6905 name has been removed. Distributions that want to maintain
6906 compatibility with the old name should add a symlink from
6907 the old name to the new name.
6909 * journald's SplitMode= now defaults to "uid". This makes sure
6910 that unprivileged users can access their own coredumps with
6911 coredumpctl without restrictions.
6913 * New kernel command line options "systemd.wants=" (for
6914 pulling an additional unit during boot), "systemd.mask="
6915 (for masking a specific unit for the boot), and
6916 "systemd.debug-shell" (for enabling the debug shell on tty9)
6917 have been added. This is implemented in the new generator
6918 "systemd-debug-generator".
6920 * systemd-nspawn will now by default filter a couple of
6921 syscalls for containers, among them those required for
6922 kernel module loading, direct x86 IO port access, swap
6923 management, and kexec. Most importantly though
6924 open_by_handle_at() is now prohibited for containers,
6925 closing a hole similar to a recently discussed vulnerability
6926 in docker regarding access to files on file hierarchies the
6927 container should normally not have access to. Note that, for
6928 nspawn, we generally make no security claims anyway (and
6929 this is explicitly documented in the man page), so this is
6930 just a fix for one of the most obvious problems.
6932 * A new man page file-hierarchy(7) has been added that
6933 contains a minimized, modernized version of the file system
6934 layout systemd expects, similar in style to the FHS
6935 specification or hier(5). A new tool systemd-path(1) has
6936 been added to query many of these paths for the local
6939 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
6940 longer done. Since the directory now has a per-user size
6941 limit, and is cleaned on logout this appears unnecessary,
6942 in particular since this now brings the lifecycle of this
6943 directory closer in line with how IPC objects are handled.
6945 * systemd.pc now exports a number of additional directories,
6946 including $libdir (which is useful to identify the library
6947 path for the primary architecture of the system), and a
6948 couple of drop-in directories.
6950 * udev's predictable network interface names now use the dev_port
6951 sysfs attribute, introduced in linux 3.15 instead of dev_id to
6952 distinguish between ports of the same PCI function. dev_id should
6953 only be used for ports using the same HW address, hence the need
6956 * machined has been updated to export the OS version of a
6957 container (read from /etc/os-release and
6958 /usr/lib/os-release) on the bus. This is now shown in
6959 "machinectl status" for a machine.
6961 * A new service setting RestartForceExitStatus= has been
6962 added. If configured to a set of exit signals or process
6963 return values, the service will be restarted when the main
6964 daemon process exits with any of them, regardless of the
6967 * systemctl's -H switch for connecting to remote systemd
6968 machines has been extended so that it may be used to
6969 directly connect to a specific container on the
6970 host. "systemctl -H root@foobar:waldi" will now connect as
6971 user "root" to host "foobar", and then proceed directly to
6972 the container named "waldi". Note that currently you have to
6973 authenticate as user "root" for this to work, as entering
6974 containers is a privileged operation.
6976 Contributions from: Andreas Henriksson, Benjamin Steinwender,
6977 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
6978 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
6979 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
6980 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
6981 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
6982 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
6983 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
6984 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
6985 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
6986 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
6987 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
6989 — Berlin, 2014-07-03
6993 * As an experimental feature, udev now tries to lock the
6994 disk device node (flock(LOCK_SH|LOCK_NB)) while it
6995 executes events for the disk or any of its partitions.
6996 Applications like partitioning programs can lock the
6997 disk device node (flock(LOCK_EX)) and claim temporary
6998 device ownership that way; udev will entirely skip all event
6999 handling for this disk and its partitions. If the disk
7000 was opened for writing, the close will trigger a partition
7001 table rescan in udev's "watch" facility, and if needed
7002 synthesize "change" events for the disk and all its partitions.
7003 This is now unconditionally enabled, and if it turns out to
7004 cause major problems, we might turn it on only for specific
7005 devices, or might need to disable it entirely. Device Mapper
7006 devices are excluded from this logic.
7008 * We temporarily dropped the "-l" switch for fsck invocations,
7009 since they collide with the flock() logic above. util-linux
7010 upstream has been changed already to avoid this conflict,
7011 and we will re-add "-l" as soon as util-linux with this
7012 change has been released.
7014 * The dependency on libattr has been removed. Since a long
7015 time, the extended attribute calls have moved to glibc, and
7016 libattr is thus unnecessary.
7018 * Virtualization detection works without privileges now. This
7019 means the systemd-detect-virt binary no longer requires
7020 CAP_SYS_PTRACE file capabilities, and our daemons can run
7021 with fewer privileges.
7023 * systemd-networkd now runs under its own "systemd-network"
7024 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
7025 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
7026 loses the ability to write to files owned by root this way.
7028 * Similarly, systemd-resolved now runs under its own
7029 "systemd-resolve" user with no capabilities remaining.
7031 * Similarly, systemd-bus-proxyd now runs under its own
7032 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
7034 * systemd-networkd gained support for setting up "veth"
7035 virtual Ethernet devices for container connectivity, as well
7036 as GRE and VTI tunnels.
7038 * systemd-networkd will no longer automatically attempt to
7039 manually load kernel modules necessary for certain tunnel
7040 transports. Instead, it is assumed the kernel loads them
7041 automatically when required. This only works correctly on
7042 very new kernels. On older kernels, please consider adding
7043 the kernel modules to /etc/modules-load.d/ as a work-around.
7045 * The resolv.conf file systemd-resolved generates has been
7046 moved to /run/systemd/resolve/. If you have a symlink from
7047 /etc/resolv.conf, it might be necessary to correct it.
7049 * Two new service settings, ProtectHome= and ProtectSystem=,
7050 have been added. When enabled, they will make the user data
7051 (such as /home) inaccessible or read-only and the system
7052 (such as /usr) read-only, for specific services. This allows
7053 very light-weight per-service sandboxing to avoid
7054 modifications of user data or system files from
7055 services. These two new switches have been enabled for all
7056 of systemd's long-running services, where appropriate.
7058 * Socket units gained new SocketUser= and SocketGroup=
7059 settings to set the owner user and group of AF_UNIX sockets
7060 and FIFOs in the file system.
7062 * Socket units gained a new RemoveOnStop= setting. If enabled,
7063 all FIFOS and sockets in the file system will be removed
7064 when the specific socket unit is stopped.
7066 * Socket units gained a new Symlinks= setting. It takes a list
7067 of symlinks to create to file system sockets or FIFOs
7068 created by the specific Unix sockets. This is useful to
7069 manage symlinks to socket nodes with the same lifecycle as
7072 * The /dev/log socket and /dev/initctl FIFO have been moved to
7073 /run, and have been replaced by symlinks. This allows
7074 connecting to these facilities even if PrivateDevices=yes is
7075 used for a service (which makes /dev/log itself unavailable,
7076 but /run is left). This also has the benefit of ensuring
7077 that /dev only contains device nodes, directories and
7078 symlinks, and nothing else.
7080 * sd-daemon gained two new calls sd_pid_notify() and
7081 sd_pid_notifyf(). They are similar to sd_notify() and
7082 sd_notifyf(), but allow overriding of the source PID of
7083 notification messages if permissions permit this. This is
7084 useful to send notify messages on behalf of a different
7085 process (for example, the parent process). The
7086 systemd-notify tool has been updated to make use of this
7087 when sending messages (so that notification messages now
7088 originate from the shell script invoking systemd-notify and
7089 not the systemd-notify process itself. This should minimize
7090 a race where systemd fails to associate notification
7091 messages to services when the originating process already
7094 * A new "on-abnormal" setting for Restart= has been added. If
7095 set, it will result in automatic restarts on all "abnormal"
7096 reasons for a process to exit, which includes unclean
7097 signals, core dumps, timeouts and watchdog timeouts, but
7098 does not include clean and unclean exit codes or clean
7099 signals. Restart=on-abnormal is an alternative for
7100 Restart=on-failure for services that shall be able to
7101 terminate and avoid restarts on certain errors, by
7102 indicating so with an unclean exit code. Restart=on-failure
7103 or Restart=on-abnormal is now the recommended setting for
7104 all long-running services.
7106 * If the InaccessibleDirectories= service setting points to a
7107 mount point (or if there are any submounts contained within
7108 it), it is now attempted to completely unmount it, to make
7109 the file systems truly unavailable for the respective
7112 * The ReadOnlyDirectories= service setting and
7113 systemd-nspawn's --read-only parameter are now recursively
7114 applied to all submounts, too.
7116 * Mount units may now be created transiently via the bus APIs.
7118 * The support for SysV and LSB init scripts has been removed
7119 from the systemd daemon itself. Instead, it is now
7120 implemented as a generator that creates native systemd units
7121 from these scripts when needed. This enables us to remove a
7122 substantial amount of legacy code from PID 1, following the
7123 fact that many distributions only ship a very small number
7124 of LSB/SysV init scripts nowadays.
7126 * Privileged Xen (dom0) domains are not considered
7127 virtualization anymore by the virtualization detection
7128 logic. After all, they generally have unrestricted access to
7129 the hardware and usually are used to manage the unprivileged
7132 * systemd-tmpfiles gained a new "C" line type, for copying
7133 files or entire directories.
7135 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
7136 lines. So far, they have been non-globbing versions of the
7137 latter, and have thus been redundant. In future, it is
7138 recommended to only use "z". "m" has hence been removed
7139 from the documentation, even though it stays supported.
7141 * A tmpfiles snippet to recreate the most basic structure in
7142 /var has been added. This is enough to create the /var/run →
7143 /run symlink and create a couple of structural
7144 directories. This allows systems to boot up with an empty or
7145 volatile /var. Of course, while with this change, the core OS
7146 now is capable with dealing with a volatile /var, not all
7147 user services are ready for it. However, we hope that sooner
7148 or later, many service daemons will be changed upstream so
7149 that they are able to automatically create their necessary
7150 directories in /var at boot, should they be missing. This is
7151 the first step to allow state-less systems that only require
7152 the vendor image for /usr to boot.
7154 * systemd-nspawn has gained a new --tmpfs= switch to mount an
7155 empty tmpfs instance to a specific directory. This is
7156 particularly useful for making use of the automatic
7157 reconstruction of /var (see above), by passing --tmpfs=/var.
7159 * Access modes specified in tmpfiles snippets may now be
7160 prefixed with "~", which indicates that they shall be masked
7161 by whether the existing file or directory is currently
7162 writable, readable or executable at all. Also, if specified,
7163 the sgid/suid/sticky bits will be masked for all
7166 * A new passive target unit "network-pre.target" has been
7167 added which is useful for services that shall run before any
7168 network is configured, for example firewall scripts.
7170 * The "floppy" group that previously owned the /dev/fd*
7171 devices is no longer used. The "disk" group is now used
7172 instead. Distributions should probably deprecate usage of
7175 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
7176 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
7177 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
7178 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
7179 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
7180 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
7183 — Berlin, 2014-06-11
7187 * A new "systemd-timesyncd" daemon has been added for
7188 synchronizing the system clock across the network. It
7189 implements an SNTP client. In contrast to NTP
7190 implementations such as chrony or the NTP reference server,
7191 this only implements a client side, and does not bother with
7192 the full NTP complexity, focusing only on querying time from
7193 one remote server and synchronizing the local clock to
7194 it. Unless you intend to serve NTP to networked clients or
7195 want to connect to local hardware clocks, this simple NTP
7196 client should be more than appropriate for most
7197 installations. The daemon runs with minimal privileges, and
7198 has been hooked up with networkd to only operate when
7199 network connectivity is available. The daemon saves the
7200 current clock to disk every time a new NTP sync has been
7201 acquired, and uses this to possibly correct the system clock
7202 early at bootup, in order to accommodate for systems that
7203 lack an RTC such as the Raspberry Pi and embedded devices,
7204 and to make sure that time monotonically progresses on these
7205 systems, even if it is not always correct. To make use of
7206 this daemon, a new system user and group "systemd-timesync"
7207 needs to be created on installation of systemd.
7209 * The queue "seqnum" interface of libudev has been disabled, as
7210 it was generally incompatible with device namespacing as
7211 sequence numbers of devices go "missing" if the devices are
7212 part of a different namespace.
7214 * "systemctl list-timers" and "systemctl list-sockets" gained
7215 a --recursive switch for showing units of these types also
7216 for all local containers, similar in style to the already
7217 supported --recursive switch for "systemctl list-units".
7219 * A new RebootArgument= setting has been added for service
7220 units, which may be used to specify a kernel reboot argument
7221 to use when triggering reboots with StartLimitAction=.
7223 * A new FailureAction= setting has been added for service
7224 units which may be used to specify an operation to trigger
7225 when a service fails. This works similarly to
7226 StartLimitAction=, but unlike it, controls what is done
7227 immediately rather than only after several attempts to
7228 restart the service in question.
7230 * hostnamed got updated to also expose the kernel name,
7231 release, and version on the bus. This is useful for
7232 executing commands like hostnamectl with the -H switch.
7233 systemd-analyze makes use of this to properly display
7234 details when running non-locally.
7236 * The bootchart tool can now show cgroup information in the
7237 graphs it generates.
7239 * The CFS CPU quota cgroup attribute is now exposed for
7240 services. The new CPUQuota= switch has been added for this
7241 which takes a percentage value. Setting this will have the
7242 result that a service may never get more CPU time than the
7243 specified percentage, even if the machine is otherwise idle.
7245 * systemd-networkd learned IPIP and SIT tunnel support.
7247 * LSB init scripts exposing a dependency on $network will now
7248 get a dependency on network-online.target rather than simply
7249 network.target. This should bring LSB handling closer to
7250 what it was on SysV systems.
7252 * A new fsck.repair= kernel option has been added to control
7253 how fsck shall deal with unclean file systems at boot.
7255 * The (.ini) configuration file parser will now silently ignore
7256 sections whose names begin with "X-". This may be used to maintain
7257 application-specific extension sections in unit files.
7259 * machined gained a new API to query the IP addresses of
7260 registered containers. "machinectl status" has been updated
7261 to show these addresses in its output.
7263 * A new call sd_uid_get_display() has been added to the
7264 sd-login APIs for querying the "primary" session of a
7265 user. The "primary" session of the user is elected from the
7266 user's sessions and generally a graphical session is
7267 preferred over a text one.
7269 * A minimal systemd-resolved daemon has been added. It
7270 currently simply acts as a companion to systemd-networkd and
7271 manages resolv.conf based on per-interface DNS
7272 configuration, possibly supplied via DHCP. In the long run
7273 we hope to extend this into a local DNSSEC enabled DNS and
7276 * The systemd-networkd-wait-online tool is now enabled by
7277 default. It will delay network-online.target until a network
7278 connection has been configured. The tool primarily integrates
7279 with networkd, but will also make a best effort to make sense
7280 of network configuration performed in some other way.
7282 * Two new service options StartupCPUShares= and
7283 StartupBlockIOWeight= have been added that work similarly to
7284 CPUShares= and BlockIOWeight= however only apply during
7285 system startup. This is useful to prioritize certain services
7286 differently during bootup than during normal runtime.
7288 * hostnamed has been changed to prefer the statically
7289 configured hostname in /etc/hostname (unless set to
7290 'localhost' or empty) over any dynamic one supplied by
7291 dhcp. With this change, the rules for picking the hostname
7292 match more closely the rules of other configuration settings
7293 where the local administrator's configuration in /etc always
7294 overrides any other settings.
7296 Contributions from: Ali H. Caliskan, Alison Chaiken, Bas van
7297 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
7298 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
7299 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
7300 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
7301 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
7302 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
7303 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
7304 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
7305 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
7306 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
7307 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
7308 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
7309 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
7310 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
7311 Lindskog, WaLyong Cho, Will Woods, Zbigniew
7314 — Beijing, 2014-05-28
7318 * When restoring the screen brightness at boot, stay away from
7319 the darkest setting or from the lowest 5% of the available
7320 range, depending on which is the larger value of both. This
7321 should effectively protect the user from rebooting into a
7322 black screen, should the brightness have been set to minimum
7325 * sd-login gained a new sd_machine_get_class() call to
7326 determine the class ("vm" or "container") of a machine
7327 registered with machined.
7329 * sd-login gained new calls
7330 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
7331 to query the identity of the peer of a local AF_UNIX
7332 connection. They operate similarly to their sd_pid_get_xyz()
7335 * PID 1 will now maintain a system-wide system state engine
7336 with the states "starting", "running", "degraded",
7337 "maintenance", "stopping". These states are bound to system
7338 startup, normal runtime, runtime with at least one failed
7339 service, rescue/emergency mode and system shutdown. This
7340 state is shown in the "systemctl status" output when no unit
7341 name is passed. It is useful to determine system state, in
7342 particularly when doing so for many systems or containers at
7345 * A new command "list-machines" has been added to "systemctl"
7346 that lists all local OS containers and shows their system
7347 state (see above), if systemd runs inside of them.
7349 * systemctl gained a new "-r" switch to recursively enumerate
7350 units on all local containers, when used with the
7351 "list-unit" command (which is the default one that is
7352 executed when no parameters are specified).
7354 * The GPT automatic partition discovery logic will now honour
7355 two GPT partition flags: one may be set on a partition to
7356 cause it to be mounted read-only, and the other may be set
7357 on a partition to ignore it during automatic discovery.
7359 * Two new GPT type UUIDs have been added for automatic root
7360 partition discovery, for 32-bit and 64-bit ARM. This is not
7361 particularly useful for discovering the root directory on
7362 these architectures during bare-metal boots (since UEFI is
7363 not common there), but still very useful to allow booting of
7364 ARM disk images in nspawn with the -i option.
7366 * MAC addresses of interfaces created with nspawn's
7367 --network-interface= switch will now be generated from the
7368 machine name, and thus be stable between multiple invocations
7371 * logind will now automatically remove all IPC objects owned
7372 by a user if she or he fully logs out. This makes sure that
7373 users who are logged out cannot continue to consume IPC
7374 resources. This covers SysV memory, semaphores and message
7375 queues as well as POSIX shared memory and message
7376 queues. Traditionally, SysV and POSIX IPC had no lifecycle
7377 limits. With this functionality, that is corrected. This may
7378 be turned off by using the RemoveIPC= switch of logind.conf.
7380 * The systemd-machine-id-setup and tmpfiles tools gained a
7381 --root= switch to operate on a specific root directory,
7384 * journald can now forward logged messages to the TTYs of all
7385 logged in users ("wall"). This is the default for all
7386 emergency messages now.
7388 * A new tool systemd-journal-remote has been added to stream
7389 journal log messages across the network.
7391 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
7392 controller trees are mounted into it. Note that the
7393 directories mounted beneath it are not read-only. This is a
7394 security measure and is particularly useful because glibc
7395 actually includes a search logic to pick any tmpfs it can
7396 find to implement shm_open() if /dev/shm is not available
7397 (which it might very well be in namespaced setups).
7399 * machinectl gained a new "poweroff" command to cleanly power
7400 down a local OS container.
7402 * The PrivateDevices= unit file setting will now also drop the
7403 CAP_MKNOD capability from the capability bound set, and
7404 imply DevicePolicy=closed.
7406 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
7407 comprehensively on all long-running systemd services where
7408 this is appropriate.
7410 * systemd-udevd will now run in a disassociated mount
7411 namespace. To mount directories from udev rules, make sure to
7412 pull in mount units via SYSTEMD_WANTS properties.
7414 * The kdbus support gained support for uploading policy into
7415 the kernel. sd-bus gained support for creating "monitoring"
7416 connections that can eavesdrop into all bus communication
7417 for debugging purposes.
7419 * Timestamps may now be specified in seconds since the UNIX
7420 epoch Jan 1st, 1970 by specifying "@" followed by the value
7423 * Native tcpwrap support in systemd has been removed. tcpwrap
7424 is old code, not really maintained anymore and has serious
7425 shortcomings, and better options such as firewalls
7426 exist. For setups that require tcpwrap usage, please
7427 consider invoking your socket-activated service via tcpd,
7428 like on traditional inetd.
7430 * A new system.conf configuration option
7431 DefaultTimerAccuracySec= has been added that controls the
7432 default AccuracySec= setting of .timer units.
7434 * Timer units gained a new WakeSystem= switch. If enabled,
7435 timers configured this way will cause the system to resume
7436 from system suspend (if the system supports that, which most
7439 * Timer units gained a new Persistent= switch. If enabled,
7440 timers configured this way will save to disk when they have
7441 been last triggered. This information is then used on next
7442 reboot to possible execute overdue timer events, that
7443 could not take place because the system was powered off.
7444 This enables simple anacron-like behaviour for timer units.
7446 * systemctl's "list-timers" will now also list the time a
7447 timer unit was last triggered in addition to the next time
7448 it will be triggered.
7450 * systemd-networkd will now assign predictable IPv4LL
7451 addresses to its local interfaces.
7453 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
7454 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
7455 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
7456 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
7457 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
7458 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
7459 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
7460 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
7463 — Berlin, 2014-03-25
7467 * A new unit file setting RestrictAddressFamilies= has been
7468 added to restrict which socket address families unit
7469 processes gain access to. This takes address family names
7470 like "AF_INET" or "AF_UNIX", and is useful to minimize the
7471 attack surface of services via exotic protocol stacks. This
7472 is built on seccomp system call filters.
7474 * Two new unit file settings RuntimeDirectory= and
7475 RuntimeDirectoryMode= have been added that may be used to
7476 manage a per-daemon runtime directories below /run. This is
7477 an alternative for setting up directory permissions with
7478 tmpfiles snippets, and has the advantage that the runtime
7479 directory's lifetime is bound to the daemon runtime and that
7480 the daemon starts up with an empty directory each time. This
7481 is particularly useful when writing services that drop
7482 privileges using the User= or Group= setting.
7484 * The DeviceAllow= unit setting now supports globbing for
7485 matching against device group names.
7487 * The systemd configuration file system.conf gained new
7488 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
7489 DefaultMemoryAccounting= to globally turn on/off accounting
7490 for specific resources (cgroups) for all units. These
7491 settings may still be overridden individually in each unit
7494 * systemd-gpt-auto-generator is now able to discover /srv and
7495 root partitions in addition to /home and swap partitions. It
7496 also supports LUKS-encrypted partitions now. With this in
7497 place, automatic discovery of partitions to mount following
7498 the Discoverable Partitions Specification
7499 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
7500 is now a lot more complete. This allows booting without
7501 /etc/fstab and without root= on the kernel command line on
7502 systems prepared appropriately.
7504 * systemd-nspawn gained a new --image= switch which allows
7505 booting up disk images and Linux installations on any block
7506 device that follow the Discoverable Partitions Specification
7507 (see above). This means that installations made with
7508 appropriately updated installers may now be started and
7509 deployed using container managers, completely
7510 unmodified. (We hope that libvirt-lxc will add support for
7511 this feature soon, too.)
7513 * systemd-nspawn gained a new --network-macvlan= setting to
7514 set up a private macvlan interface for the
7515 container. Similarly, systemd-networkd gained a new
7516 Kind=macvlan setting in .netdev files.
7518 * systemd-networkd now supports configuring local addresses
7521 * A new tool systemd-network-wait-online has been added to
7522 synchronously wait for network connectivity using
7525 * The sd-bus.h bus API gained a new sd_bus_track object for
7526 tracking the lifecycle of bus peers. Note that sd-bus.h is
7527 still not a public API though (unless you specify
7528 --enable-kdbus on the configure command line, which however
7529 voids your warranty and you get no API stability guarantee).
7531 * The $XDG_RUNTIME_DIR runtime directories for each user are
7532 now individual tmpfs instances, which has the benefit of
7533 introducing separate pools for each user, with individual
7534 size limits, and thus making sure that unprivileged clients
7535 can no longer negatively impact the system or other users by
7536 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
7537 RuntimeDirectorySize= has been introduced that allows
7538 controlling the default size limit for all users. It
7539 defaults to 10% of the available physical memory. This is no
7540 replacement for quotas on tmpfs though (which the kernel
7541 still does not support), as /dev/shm and /tmp are still
7542 shared resources used by both the system and unprivileged
7545 * logind will now automatically turn off automatic suspending
7546 on laptop lid close when more than one display is
7547 connected. This was previously expected to be implemented
7548 individually in desktop environments (such as GNOME),
7549 however has been added to logind now, in order to fix a
7550 boot-time race where a desktop environment might not have
7551 been started yet and thus not been able to take an inhibitor
7552 lock at the time where logind already suspends the system
7553 due to a closed lid.
7555 * logind will now wait at least 30s after each system
7556 suspend/resume cycle, and 3min after system boot before
7557 suspending the system due to a closed laptop lid. This
7558 should give USB docking stations and similar enough time to
7559 be probed and configured after system resume and boot in
7560 order to then act as suspend blocker.
7562 * systemd-run gained a new --property= setting which allows
7563 initialization of resource control properties (and others)
7564 for the created scope or service unit. Example: "systemd-run
7565 --property=BlockIOWeight=10 updatedb" may be used to run
7566 updatedb at a low block IO scheduling weight.
7568 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
7569 now also work in --scope mode.
7571 * When systemd is compiled with kdbus support, basic support
7572 for enforced policies is now in place. (Note that enabling
7573 kdbus still voids your warranty and no API compatibility
7576 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
7577 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
7578 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
7579 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
7580 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
7581 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
7582 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
7583 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
7584 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
7585 Zbigniew Jędrzejewski-Szmek
7587 — Berlin, 2014-03-12
7591 * systemd will now relabel /dev after loading the SMACK policy
7592 according to SMACK rules.
7594 * A new unit file option AppArmorProfile= has been added to
7595 set the AppArmor profile for the processes of a unit.
7597 * A new condition check ConditionArchitecture= has been added
7598 to conditionalize units based on the system architecture, as
7599 reported by uname()'s "machine" field.
7601 * systemd-networkd now supports matching on the system
7602 virtualization, architecture, kernel command line, hostname
7605 * logind is now a lot more aggressive when suspending the
7606 machine due to a closed laptop lid. Instead of acting only
7607 on the lid close action, it will continuously watch the lid
7608 status and act on it. This is useful for laptops where the
7609 power button is on the outside of the chassis so that it can
7610 be reached without opening the lid (such as the Lenovo
7611 Yoga). On those machines, logind will now immediately
7612 re-suspend the machine if the power button has been
7613 accidentally pressed while the laptop was suspended and in a
7614 backpack or similar.
7616 * logind will now watch SW_DOCK switches and inhibit reaction
7617 to the lid switch if it is pressed. This means that logind
7618 will not suspend the machine anymore if the lid is closed
7619 and the system is docked, if the laptop supports SW_DOCK
7620 notifications via the input layer. Note that ACPI docking
7621 stations do not generate this currently. Also note that this
7622 logic is usually not fully sufficient and Desktop
7623 Environments should take a lid switch inhibitor lock when an
7624 external display is connected, as systemd will not watch
7627 * nspawn will now make use of the devices cgroup controller by
7628 default, and only permit creation of and access to the usual
7629 API device nodes like /dev/null or /dev/random, as well as
7630 access to (but not creation of) the pty devices.
7632 * We will now ship a default .network file for
7633 systemd-networkd that automatically configures DHCP for
7634 network interfaces created by nspawn's --network-veth or
7635 --network-bridge= switches.
7637 * systemd will now understand the usual M, K, G, T suffixes
7638 according to SI conventions (i.e. to the base 1000) when
7639 referring to throughput and hardware metrics. It will stay
7640 with IEC conventions (i.e. to the base 1024) for software
7641 metrics, according to what is customary according to
7642 Wikipedia. We explicitly document which base applies for
7643 each configuration option.
7645 * The DeviceAllow= setting in unit files now supports a syntax
7646 to whitelist an entire group of devices node majors at once,
7647 based on the /proc/devices listing. For example, with the
7648 string "char-pts", it is now possible to whitelist all
7649 current and future pseudo-TTYs at once.
7651 * sd-event learned a new "post" event source. Event sources of
7652 this type are triggered by the dispatching of any event
7653 source of a type that is not "post". This is useful for
7654 implementing clean-up and check event sources that are
7655 triggered by other work being done in the program.
7657 * systemd-networkd is no longer statically enabled, but uses
7658 the usual [Install] sections so that it can be
7659 enabled/disabled using systemctl. It still is enabled by
7662 * When creating a veth interface pair with systemd-nspawn, the
7663 host side will now be prefixed with "vb-" if
7664 --network-bridge= is used, and with "ve-" if --network-veth
7665 is used. This way, it is easy to distinguish these cases on
7666 the host, for example to apply different configuration to
7667 them with systemd-networkd.
7669 * The compatibility libraries for libsystemd-journal.so,
7670 libsystem-id128.so, libsystemd-login.so and
7671 libsystemd-daemon.so do not make use of IFUNC
7672 anymore. Instead, we now build libsystemd.so multiple times
7673 under these alternative names. This means that the footprint
7674 is drastically increased, but given that these are
7675 transitional compatibility libraries, this should not matter
7676 much. This change has been made necessary to support the ARM
7677 platform for these compatibility libraries, as the ARM
7678 toolchain is not really at the same level as the toolchain
7679 for other architectures like x86 and does not support
7680 IFUNC. Please make sure to use --enable-compat-libs only
7681 during a transitional period!
7683 * The .include syntax has been deprecated and is not documented
7684 anymore. Drop-in files in .d directories should be used instead.
7686 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
7687 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
7688 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
7689 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
7690 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
7691 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
7692 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
7693 Zbigniew Jędrzejewski-Szmek
7695 — Berlin, 2014-02-24
7699 * A new component "systemd-networkd" has been added that can
7700 be used to configure local network interfaces statically or
7701 via DHCP. It is capable of bringing up bridges, VLANs, and
7702 bonding. Currently, no hook-ups for interactive network
7703 configuration are provided. Use this for your initrd,
7704 container, embedded, or server setup if you need a simple,
7705 yet powerful, network configuration solution. This
7706 configuration subsystem is quite nifty, as it allows wildcard
7707 hotplug matching in interfaces. For example, with a single
7708 configuration snippet, you can configure that all Ethernet
7709 interfaces showing up are automatically added to a bridge,
7710 or similar. It supports link-sensing and more.
7712 * A new tool "systemd-socket-proxyd" has been added which can
7713 act as a bidirectional proxy for TCP sockets. This is
7714 useful for adding socket activation support to services that
7715 do not actually support socket activation, including virtual
7716 machines and the like.
7718 * Add a new tool to save/restore rfkill state on
7721 * Save/restore state of keyboard backlights in addition to
7722 display backlights on shutdown/boot.
7724 * udev learned a new SECLABEL{} construct to label device
7725 nodes with a specific security label when they appear. For
7726 now, only SECLABEL{selinux} is supported, but the syntax is
7727 prepared for additional security frameworks.
7729 * udev gained a new scheme to configure link-level attributes
7730 from files in /etc/systemd/network/*.link. These files can
7731 match against MAC address, device path, driver name and type,
7732 and will apply attributes like the naming policy, link speed,
7733 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
7734 address assignment policy (randomized, ...).
7736 * The configuration of network interface naming rules for
7737 "permanent interface names" has changed: a new NamePolicy=
7738 setting in the [Link] section of .link files determines the
7739 priority of possible naming schemes (onboard, slot, MAC,
7740 path). The default value of this setting is determined by
7741 /usr/lib/net/links/99-default.link. Old
7742 80-net-name-slot.rules udev configuration file has been
7743 removed, so local configuration overriding this file should
7744 be adapted to override 99-default.link instead.
7746 * When the User= switch is used in a unit file, also
7747 initialize $SHELL= based on the user database entry.
7749 * systemd no longer depends on libdbus. All communication is
7750 now done with sd-bus, systemd's low-level bus library
7753 * kdbus support has been added to PID 1 itself. When kdbus is
7754 enabled, this causes PID 1 to set up the system bus and
7755 enable support for a new ".busname" unit type that
7756 encapsulates bus name activation on kdbus. It works a little
7757 bit like ".socket" units, except for bus names. A new
7758 generator has been added that converts classic dbus1 service
7759 activation files automatically into native systemd .busname
7762 * sd-bus: add a light-weight vtable implementation that allows
7763 defining objects on the bus with a simple static const
7764 vtable array of its methods, signals and properties.
7766 * systemd will not generate or install static dbus
7767 introspection data anymore to /usr/share/dbus-1/interfaces,
7768 as the precise format of these files is unclear, and
7769 nothing makes use of it.
7771 * A proxy daemon is now provided to proxy clients connecting
7772 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
7773 compatibility with classic D-Bus.
7775 * A bus driver implementation has been added that supports the
7776 classic D-Bus bus driver calls on kdbus, also for
7777 compatibility purposes.
7779 * A new API "sd-event.h" has been added that implements a
7780 minimal event loop API built around epoll. It provides a
7781 couple of features that direct epoll usage is lacking:
7782 prioritization of events, scales to large numbers of timer
7783 events, per-event timer slack (accuracy), system-wide
7784 coalescing of timer events, exit handlers, watchdog
7785 supervision support using systemd's sd_notify() API, child
7788 * A new API "sd-rntl.h" has been added that provides an API
7789 around the route netlink interface of the kernel, similar in
7790 style to "sd-bus.h".
7792 * A new API "sd-dhcp-client.h" has been added that provides a
7793 small DHCPv4 client-side implementation. This is used by
7796 * There is a new kernel command line option
7797 "systemd.restore_state=0|1". When set to "0", none of the
7798 systemd tools will restore saved runtime state to hardware
7799 devices. More specifically, the rfkill and backlight states
7802 * The FsckPassNo= compatibility option in mount/service units
7803 has been removed. The fstab generator will now add the
7804 necessary dependencies automatically, and does not require
7805 PID1's support for that anymore.
7807 * journalctl gained a new switch, --list-boots, that lists
7808 recent boots with their times and boot IDs.
7810 * The various tools like systemctl, loginctl, timedatectl,
7811 busctl, systemd-run, ... have gained a new switch "-M" to
7812 connect to a specific, local OS container (as direct
7813 connection, without requiring SSH). This works on any
7814 container that is registered with machined, such as those
7815 created by libvirt-lxc or nspawn.
7817 * systemd-run and systemd-analyze also gained support for "-H"
7818 to connect to remote hosts via SSH. This is particularly
7819 useful for systemd-run because it enables queuing of jobs
7820 onto remote systems.
7822 * machinectl gained a new command "login" to open a getty
7823 login in any local container. This works with any container
7824 that is registered with machined (such as those created by
7825 libvirt-lxc or nspawn), and which runs systemd inside.
7827 * machinectl gained a new "reboot" command that may be used to
7828 trigger a reboot on a specific container that is registered
7829 with machined. This works on any container that runs an init
7830 system of some kind.
7832 * systemctl gained a new "list-timers" command to print a nice
7833 listing of installed timer units with the times they elapse
7836 * Alternative reboot() parameters may now be specified on the
7837 "systemctl reboot" command line and are passed to the
7838 reboot() system call.
7840 * systemctl gained a new --job-mode= switch to configure the
7841 mode to queue a job with. This is a more generic version of
7842 --fail, --irreversible, and --ignore-dependencies, which are
7843 still available but not advertised anymore.
7845 * /etc/systemd/system.conf gained new settings to configure
7846 various default timeouts of units, as well as the default
7847 start limit interval and burst. These may still be overridden
7850 * PID1 will now export on the bus profile data of the security
7851 policy upload process (such as the SELinux policy upload to
7854 * journald: when forwarding logs to the console, include
7855 timestamps (following the setting in
7856 /sys/module/printk/parameters/time).
7858 * OnCalendar= in timer units now understands the special
7859 strings "yearly" and "annually". (Both are equivalent)
7861 * The accuracy of timer units is now configurable with the new
7862 AccuracySec= setting. It defaults to 1min.
7864 * A new dependency type JoinsNamespaceOf= has been added that
7865 allows running two services within the same /tmp and network
7866 namespace, if PrivateNetwork= or PrivateTmp= are used.
7868 * A new command "cat" has been added to systemctl. It outputs
7869 the original unit file of a unit, and concatenates the
7870 contents of additional "drop-in" unit file snippets, so that
7871 the full configuration is shown.
7873 * systemctl now supports globbing on the various "list-xyz"
7874 commands, like "list-units" or "list-sockets", as well as on
7875 those commands which take multiple unit names.
7877 * journalctl's --unit= switch gained support for globbing.
7879 * All systemd daemons now make use of the watchdog logic so
7880 that systemd automatically notices when they hang.
7882 * If the $container_ttys environment variable is set,
7883 getty-generator will automatically spawn a getty for each
7884 listed tty. This is useful for container managers to request
7885 login gettys to be spawned on as many ttys as needed.
7887 * %h, %s, %U specifier support is not available anymore when
7888 used in unit files for PID 1. This is because NSS calls are
7889 not safe from PID 1. They stay available for --user
7890 instances of systemd, and as special case for the root user.
7892 * loginctl gained a new "--no-legend" switch to turn off output
7895 * The "sd-login.h" API gained three new calls:
7896 sd_session_is_remote(), sd_session_get_remote_user(),
7897 sd_session_get_remote_host() to query information about
7900 * The udev hardware database now also carries vendor/product
7901 information of SDIO devices.
7903 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
7904 determine whether watchdog notifications are requested by
7907 * Socket-activated per-connection services now include a
7908 short description of the connection parameters in the
7911 * tmpfiles gained a new "--boot" option. When this is not used,
7912 only lines where the command character is not suffixed with
7913 "!" are executed. When this option is specified, those
7914 options are executed too. This partitions tmpfiles
7915 directives into those that can be safely executed at any
7916 time, and those which should be run only at boot (for
7917 example, a line that creates /run/nologin).
7919 * A new API "sd-resolve.h" has been added which provides a simple
7920 asynchronous wrapper around glibc NSS hostname resolution
7921 calls, such as getaddrinfo(). In contrast to glibc's
7922 getaddrinfo_a(), it does not use signals. In contrast to most
7923 other asynchronous name resolution libraries, this one does
7924 not reimplement DNS, but reuses NSS, so that alternate
7925 hostname resolution systems continue to work, such as mDNS,
7926 LDAP, etc. This API is based on libasyncns, but it has been
7927 cleaned up for inclusion in systemd.
7929 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
7930 "sd-daemon.h" are no longer found in individual libraries
7931 libsystemd-journal.so, libsystemd-login.so,
7932 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
7933 merged them into a single library, libsystemd.so, which
7934 provides all symbols. The reason for this is cyclic
7935 dependencies, as these libraries tend to use each other's
7936 symbols. So far, we have managed to workaround that by linking
7937 a copy of a good part of our code into each of these
7938 libraries again and again, which, however, makes certain
7939 things hard to do, like sharing static variables. Also, it
7940 substantially increases footprint. With this change, there
7941 is only one library for the basic APIs systemd
7942 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
7943 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
7944 library as well, however are subject to the --enable-kdbus
7945 switch (see below). Note that "sd-dhcp-client.h" is not part
7946 of this library (this is because it only consumes, never
7947 provides, services of/to other APIs). To make the transition
7948 easy from the separate libraries to the unified one, we
7949 provide the --enable-compat-libs compile-time switch which
7950 will generate stub libraries that are compatible with the
7951 old ones but redirect all calls to the new one.
7953 * All of the kdbus logic and the new APIs "sd-bus.h",
7954 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
7955 and "sd-utf8.h" are compile-time optional via the
7956 "--enable-kdbus" switch, and they are not compiled in by
7957 default. To make use of kdbus, you have to explicitly enable
7958 the switch. Note however, that neither the kernel nor the
7959 userspace API for all of this is considered stable yet. We
7960 want to maintain the freedom to still change the APIs for
7961 now. By specifying this build-time switch, you acknowledge
7962 that you are aware of the instability of the current
7965 * Also, note that while kdbus is pretty much complete,
7966 it lacks one thing: proper policy support. This means you
7967 can build a fully working system with all features; however,
7968 it will be highly insecure. Policy support will be added in
7969 one of the next releases, at the same time that we will
7970 declare the APIs stable.
7972 * When the kernel command line argument "kdbus" is specified,
7973 systemd will automatically load the kdbus.ko kernel module. At
7974 this stage of development, it is only useful for testing kdbus
7975 and should not be used in production. Note: if "--enable-kdbus"
7976 is specified, and the kdbus.ko kernel module is available, and
7977 "kdbus" is added to the kernel command line, the entire system
7978 runs with kdbus instead of dbus-daemon, with the above mentioned
7979 problem of missing the system policy enforcement. Also a future
7980 version of kdbus.ko or a newer systemd will not be compatible with
7981 each other, and will unlikely be able to boot the machine if only
7982 one of them is updated.
7984 * systemctl gained a new "import-environment" command which
7985 uploads the caller's environment (or parts thereof) into the
7986 service manager so that it is inherited by services started
7987 by the manager. This is useful to upload variables like
7988 $DISPLAY into the user service manager.
7990 * A new PrivateDevices= switch has been added to service units
7991 which allows running a service with a namespaced /dev
7992 directory that does not contain any device nodes for
7993 physical devices. More specifically, it only includes devices
7994 such as /dev/null, /dev/urandom, and /dev/zero which are API
7997 * logind has been extended to support behaviour like VT
7998 switching on seats that do not support a VT. This makes
7999 multi-session available on seats that are not the first seat
8000 (seat0), and on systems where kernel support for VTs has
8001 been disabled at compile-time.
8003 * If a process holds a delay lock for system sleep or shutdown
8004 and fails to release it in time, we will now log its
8005 identity. This makes it easier to identify processes that
8006 cause slow suspends or power-offs.
8008 * When parsing /etc/crypttab, support for a new key-slot=
8009 option as supported by Debian is added. It allows indicating
8010 which LUKS slot to use on disk, speeding up key loading.
8012 * The sd_journal_sendv() API call has been checked and
8013 officially declared to be async-signal-safe so that it may
8014 be invoked from signal handlers for logging purposes.
8016 * Boot-time status output is now enabled automatically after a
8017 short timeout if boot does not progress, in order to give
8018 the user an indication what she or he is waiting for.
8020 * The boot-time output has been improved to show how much time
8021 remains until jobs expire.
8023 * The KillMode= switch in service units gained a new possible
8024 value "mixed". If set, and the unit is shut down, then the
8025 initial SIGTERM signal is sent only to the main daemon
8026 process, while the following SIGKILL signal is sent to
8027 all remaining processes of the service.
8029 * When a scope unit is registered, a new property "Controller"
8030 may be set. If set to a valid bus name, systemd will send a
8031 RequestStop() signal to this name when it would like to shut
8032 down the scope. This may be used to hook manager logic into
8033 the shutdown logic of scope units. Also, scope units may now
8034 be put in a special "abandoned" state, in which case the
8035 manager process which created them takes no further
8036 responsibilities for it.
8038 * When reading unit files, systemd will now verify
8039 the access mode of these files, and warn about certain
8040 suspicious combinations. This has been added to make it
8041 easier to track down packaging bugs where unit files are
8042 marked executable or world-writable.
8044 * systemd-nspawn gained a new "--setenv=" switch to set
8045 container-wide environment variables. The similar option in
8046 systemd-activate was renamed from "--environment=" to
8047 "--setenv=" for consistency.
8049 * systemd-nspawn has been updated to create a new kdbus domain
8050 for each container that is invoked, thus allowing each
8051 container to have its own set of system and user buses,
8052 independent of the host.
8054 * systemd-nspawn gained a new --drop-capability= switch to run
8055 the container with less capabilities than the default. Both
8056 --drop-capability= and --capability= now take the special
8057 string "all" for dropping or keeping all capabilities.
8059 * systemd-nspawn gained new switches for executing containers
8060 with specific SELinux labels set.
8062 * systemd-nspawn gained a new --quiet switch to not generate
8063 any additional output but the container's own console
8066 * systemd-nspawn gained a new --share-system switch to run a
8067 container without PID namespacing enabled.
8069 * systemd-nspawn gained a new --register= switch to control
8070 whether the container is registered with systemd-machined or
8071 not. This is useful for containers that do not run full
8072 OS images, but only specific apps.
8074 * systemd-nspawn gained a new --keep-unit which may be used
8075 when invoked as the only program from a service unit, and
8076 results in registration of the unit service itself in
8077 systemd-machined, instead of a newly opened scope unit.
8079 * systemd-nspawn gained a new --network-interface= switch for
8080 moving arbitrary interfaces to the container. The new
8081 --network-veth switch creates a virtual Ethernet connection
8082 between host and container. The new --network-bridge=
8083 switch then allows assigning the host side of this virtual
8084 Ethernet connection to a bridge device.
8086 * systemd-nspawn gained a new --personality= switch for
8087 setting the kernel personality for the container. This is
8088 useful when running a 32-bit container on a 64-bit host. A
8089 similar option Personality= is now also available for service
8092 * logind will now also track a "Desktop" identifier for each
8093 session which encodes the desktop environment of it. This is
8094 useful for desktop environments that want to identify
8095 multiple running sessions of itself easily.
8097 * A new SELinuxContext= setting for service units has been
8098 added that allows setting a specific SELinux execution
8099 context for a service.
8101 * Most systemd client tools will now honour $SYSTEMD_LESS for
8102 settings of the "less" pager. By default, these tools will
8103 override $LESS to allow certain operations to work, such as
8104 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
8105 influence this logic.
8107 * systemd's "seccomp" hook-up has been changed to make use of
8108 the libseccomp library instead of using its own
8109 implementation. This has benefits for portability among
8112 * For usage together with SystemCallFilter=, a new
8113 SystemCallErrorNumber= setting has been introduced that
8114 allows configuration of a system error number to be returned
8115 on filtered system calls, instead of immediately killing the
8116 process. Also, SystemCallArchitectures= has been added to
8117 limit access to system calls of a particular architecture
8118 (in order to turn off support for unused secondary
8119 architectures). There is also a global
8120 SystemCallArchitectures= setting in system.conf now to turn
8121 off support for non-native system calls system-wide.
8123 * systemd requires a kernel with a working name_to_handle_at(),
8124 please see the kernel config requirements in the README file.
8126 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
8127 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
8128 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
8129 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
8130 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
8131 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
8132 Elia Pinto, Florian Weimer, George McCollister, Goffredo
8133 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
8134 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
8135 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
8136 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
8137 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
8138 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
8139 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
8140 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
8141 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
8142 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
8143 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
8144 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
8145 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
8146 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
8147 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
8148 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
8149 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
8151 — Berlin, 2014-02-20
8155 * logind has gained support for facilitating privileged input
8156 and drm device access for unprivileged clients. This work is
8157 useful to allow Wayland display servers (and similar
8158 programs, such as kmscon) to run under the user's ID and
8159 access input and drm devices which are normally
8160 protected. When this is used (and the kernel is new enough)
8161 logind will "mute" IO on the file descriptors passed to
8162 Wayland as long as it is in the background and "unmute" it
8163 if it returns into the foreground. This allows secure
8164 session switching without allowing background sessions to
8165 eavesdrop on input and display data. This also introduces
8166 session switching support if VT support is turned off in the
8167 kernel, and on seats that are not seat0.
8169 * A new kernel command line option luks.options= is understood
8170 now which allows specifying LUKS options for usage for LUKS
8171 encrypted partitions specified with luks.uuid=.
8173 * tmpfiles.d(5) snippets may now use specifier expansion in
8174 path names. More specifically %m, %b, %H, %v, are now
8175 replaced by the local machine id, boot id, hostname, and
8176 kernel version number.
8178 * A new tmpfiles.d(5) command "m" has been introduced which
8179 may be used to change the owner/group/access mode of a file
8180 or directory if it exists, but do nothing if it does not.
8182 * This release removes high-level support for the
8183 MemorySoftLimit= cgroup setting. The underlying kernel
8184 cgroup attribute memory.soft_limit= is currently badly
8185 designed and likely to be removed from the kernel API in its
8186 current form, hence we should not expose it for now.
8188 * The memory.use_hierarchy cgroup attribute is now enabled for
8189 all cgroups systemd creates in the memory cgroup
8190 hierarchy. This option is likely to be come the built-in
8191 default in the kernel anyway, and the non-hierarchical mode
8192 never made much sense in the intrinsically hierarchical
8195 * A new field _SYSTEMD_SLICE= is logged along with all journal
8196 messages containing the slice a message was generated
8197 from. This is useful to allow easy per-customer filtering of
8198 logs among other things.
8200 * systemd-journald will no longer adjust the group of journal
8201 files it creates to the "systemd-journal" group. Instead we
8202 rely on the journal directory to be owned by the
8203 "systemd-journal" group, and its setgid bit set, so that the
8204 kernel file system layer will automatically enforce that
8205 journal files inherit this group assignment. The reason for
8206 this change is that we cannot allow NSS look-ups from
8207 journald which would be necessary to resolve
8208 "systemd-journal" to a numeric GID, because this might
8209 create deadlocks if NSS involves synchronous queries to
8210 other daemons (such as nscd, or sssd) which in turn are
8211 logging clients of journald and might block on it, which
8212 would then dead lock. A tmpfiles.d(5) snippet included in
8213 systemd will make sure the setgid bit and group are
8214 properly set on the journal directory if it exists on every
8215 boot. However, we recommend adjusting it manually after
8216 upgrades too (or from RPM scriptlets), so that the change is
8217 not delayed until next reboot.
8219 * Backlight and random seed files in /var/lib/ have moved into
8220 the /var/lib/systemd/ directory, in order to centralize all
8221 systemd generated files in one directory.
8223 * Boot time performance measurements (as displayed by
8224 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
8225 performance information if that's available to determine how
8226 much time BIOS and boot loader initialization required. With
8227 a sufficiently new BIOS you hence no longer need to boot
8228 with Gummiboot to get access to such information.
8230 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
8231 Cristian Rodríguez, Dave Reisner, David Herrmann, David
8232 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
8233 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
8234 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
8235 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
8236 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
8238 — Berlin, 2013-10-02
8242 * The Restart= option for services now understands a new
8243 on-watchdog setting, which will restart the service
8244 automatically if the service stops sending out watchdog keep
8245 alive messages (as configured with WatchdogSec=).
8247 * The getty generator (which is responsible for bringing up a
8248 getty on configured serial consoles) will no longer only
8249 start a getty on the primary kernel console but on all
8250 others, too. This makes the order in which console= is
8251 specified on the kernel command line less important.
8253 * libsystemd-logind gained a new sd_session_get_vt() call to
8254 retrieve the VT number of a session.
8256 * If the option "tries=0" is set for an entry of /etc/crypttab
8257 its passphrase is queried indefinitely instead of any
8258 maximum number of tries.
8260 * If a service with a configure PID file terminates its PID
8261 file will now be removed automatically if it still exists
8262 afterwards. This should put an end to stale PID files.
8264 * systemd-run will now also take relative binary path names
8265 for execution and no longer insists on absolute paths.
8267 * InaccessibleDirectories= and ReadOnlyDirectories= now take
8268 paths that are optionally prefixed with "-" to indicate that
8269 it should not be considered a failure if they do not exist.
8271 * journalctl -o (and similar commands) now understands a new
8272 output mode "short-precise", it is similar to "short" but
8273 shows timestamps with usec accuracy.
8275 * The option "discard" (as known from Debian) is now
8276 synonymous to "allow-discards" in /etc/crypttab. In fact,
8277 "discard" is preferred now (since it is easier to remember
8280 * Some licensing clean-ups were made, so that more code is now
8281 LGPL-2.1 licensed than before.
8283 * A minimal tool to save/restore the display backlight
8284 brightness across reboots has been added. It will store the
8285 backlight setting as late as possible at shutdown, and
8286 restore it as early as possible during reboot.
8288 * A logic to automatically discover and enable home and swap
8289 partitions on GPT disks has been added. With this in place
8290 /etc/fstab becomes optional for many setups as systemd can
8291 discover certain partitions located on the root disk
8292 automatically. Home partitions are recognized under their
8293 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
8294 partitions are recognized under their GPT type ID
8295 0657fd6da4ab43c484e50933c84b4f4f.
8297 * systemd will no longer pass any environment from the kernel
8298 or initrd to system services. If you want to set an
8299 environment for all services, do so via the kernel command
8300 line systemd.setenv= assignment.
8302 * The systemd-sysctl tool no longer natively reads the file
8303 /etc/sysctl.conf. If desired, the file should be symlinked
8304 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
8305 legacy support by a symlink rather than built-in code, it
8306 also makes the otherwise hidden order of application of the
8307 different files visible. (Note that this partly reverts to a
8308 pre-198 application order of sysctl knobs!)
8310 * The "systemctl set-log-level" and "systemctl dump" commands
8311 have been moved to systemd-analyze.
8313 * systemd-run learned the new --remain-after-exit switch,
8314 which causes the scope unit not to be cleaned up
8315 automatically after the process terminated.
8317 * tmpfiles learned a new --exclude-prefix= switch to exclude
8318 certain paths from operation.
8320 * journald will now automatically flush all messages to disk
8321 as soon as a message at the log level CRIT, ALERT or EMERG
8324 Contributions from: Andrew Cook, Brandon Philips, Christian
8325 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
8326 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
8327 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
8328 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
8329 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
8330 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
8331 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
8332 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
8333 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
8334 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
8335 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
8336 William Giokas, Zbigniew Jędrzejewski-Szmek
8338 — Berlin, 2013-09-13
8342 * The documentation has been updated to cover the various new
8343 concepts introduced with 205.
8345 * Unit files now understand the new %v specifier which
8346 resolves to the kernel version string as returned by "uname
8349 * systemctl now supports filtering the unit list output by
8350 load state, active state and sub state, using the new
8353 * "systemctl status" will now show the results of the
8354 condition checks (like ConditionPathExists= and similar) of
8355 the last start attempts of the unit. They are also logged to
8358 * "journalctl -b" may now be used to look for boot output of a
8359 specific boot. Try "journalctl -b -1" for the previous boot,
8360 but the syntax is substantially more powerful.
8362 * "journalctl --show-cursor" has been added which prints the
8363 cursor string the last shown log line. This may then be used
8364 with the new "journalctl --after-cursor=" switch to continue
8365 browsing logs from that point on.
8367 * "journalctl --force" may now be used to force regeneration
8370 * Creation of "dead" device nodes has been moved from udev
8371 into kmod and tmpfiles. Previously, udev would read the kmod
8372 databases to pre-generate dead device nodes based on meta
8373 information contained in kernel modules, so that these would
8374 be auto-loaded on access rather then at boot. As this
8375 does not really have much to do with the exposing actual
8376 kernel devices to userspace this has always been slightly
8377 alien in the udev codebase. Following the new scheme kmod
8378 will now generate a runtime snippet for tmpfiles from the
8379 module meta information and it now is tmpfiles' job to the
8380 create the nodes. This also allows overriding access and
8381 other parameters for the nodes using the usual tmpfiles
8382 facilities. As side effect this allows us to remove the
8383 CAP_SYS_MKNOD capability bit from udevd entirely.
8385 * logind's device ACLs may now be applied to these "dead"
8386 devices nodes too, thus finally allowing managed access to
8387 devices such as /dev/snd/sequencer without loading the
8388 backing module right-away.
8390 * A new RPM macro has been added that may be used to apply
8391 tmpfiles configuration during package installation.
8393 * systemd-detect-virt and ConditionVirtualization= now can
8394 detect User-Mode-Linux machines (UML).
8396 * journald will now implicitly log the effective capabilities
8397 set of processes in the message metadata.
8399 * systemd-cryptsetup has gained support for TrueCrypt volumes.
8401 * The initrd interface has been simplified (more specifically,
8402 support for passing performance data via environment
8403 variables and fsck results via files in /run has been
8404 removed). These features were non-essential, and are
8405 nowadays available in a much nicer way by having systemd in
8406 the initrd serialize its state and have the hosts systemd
8407 deserialize it again.
8409 * The udev "keymap" data files and tools to apply keyboard
8410 specific mappings of scan to key codes, and force-release
8411 scan code lists have been entirely replaced by a udev
8412 "keyboard" builtin and a hwdb data file.
8414 * systemd will now honour the kernel's "quiet" command line
8415 argument also during late shutdown, resulting in a
8416 completely silent shutdown when used.
8418 * There's now an option to control the SO_REUSEPORT socket
8419 option in .socket units.
8421 * Instance units will now automatically get a per-template
8422 subslice of system.slice unless something else is explicitly
8423 configured. For example, instances of sshd@.service will now
8424 implicitly be placed in system-sshd.slice rather than
8425 system.slice as before.
8427 * Test coverage support may now be enabled at build time.
8429 Contributions from: Dave Reisner, Frederic Crozat, Harald
8430 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
8431 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
8432 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
8433 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
8434 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
8435 Giokas, Zbigniew Jędrzejewski-Szmek
8437 — Berlin, 2013-07-23
8441 * Two new unit types have been introduced:
8443 Scope units are very similar to service units, however, are
8444 created out of pre-existing processes — instead of PID 1
8445 forking off the processes. By using scope units it is
8446 possible for system services and applications to group their
8447 own child processes (worker processes) in a powerful way
8448 which then maybe used to organize them, or kill them
8449 together, or apply resource limits on them.
8451 Slice units may be used to partition system resources in an
8452 hierarchical fashion and then assign other units to them. By
8453 default there are now three slices: system.slice (for all
8454 system services), user.slice (for all user sessions),
8455 machine.slice (for VMs and containers).
8457 Slices and scopes have been introduced primarily in
8458 context of the work to move cgroup handling to a
8459 single-writer scheme, where only PID 1
8460 creates/removes/manages cgroups.
8462 * There's a new concept of "transient" units. In contrast to
8463 normal units these units are created via an API at runtime,
8464 not from configuration from disk. More specifically this
8465 means it is now possible to run arbitrary programs as
8466 independent services, with all execution parameters passed
8467 in via bus APIs rather than read from disk. Transient units
8468 make systemd substantially more dynamic then it ever was,
8469 and useful as a general batch manager.
8471 * logind has been updated to make use of scope and slice units
8472 for managing user sessions. As a user logs in he will get
8473 his own private slice unit, to which all sessions are added
8474 as scope units. We also added support for automatically
8475 adding an instance of user@.service for the user into the
8476 slice. Effectively logind will no longer create cgroup
8477 hierarchies on its own now, it will defer entirely to PID 1
8478 for this by means of scope, service and slice units. Since
8479 user sessions this way become entities managed by PID 1
8480 the output of "systemctl" is now a lot more comprehensive.
8482 * A new mini-daemon "systemd-machined" has been added which
8483 may be used by virtualization managers to register local
8484 VMs/containers. nspawn has been updated accordingly, and
8485 libvirt will be updated shortly. machined will collect a bit
8486 of meta information about the VMs/containers, and assign
8487 them their own scope unit (see above). The collected
8488 meta-data is then made available via the "machinectl" tool,
8489 and exposed in "ps" and similar tools. machined/machinectl
8490 is compile-time optional.
8492 * As discussed earlier, the low-level cgroup configuration
8493 options ControlGroup=, ControlGroupModify=,
8494 ControlGroupPersistent=, ControlGroupAttribute= have been
8495 removed. Please use high-level attribute settings instead as
8496 well as slice units.
8498 * A new bus call SetUnitProperties() has been added to alter
8499 various runtime parameters of a unit. This is primarily
8500 useful to alter cgroup parameters dynamically in a nice way,
8501 but will be extended later on to make more properties
8502 modifiable at runtime. systemctl gained a new set-properties
8503 command that wraps this call.
8505 * A new tool "systemd-run" has been added which can be used to
8506 run arbitrary command lines as transient services or scopes,
8507 while configuring a number of settings via the command
8508 line. This tool is currently very basic, however already
8509 very useful. We plan to extend this tool to even allow
8510 queuing of execution jobs with time triggers from the
8511 command line, similar in fashion to "at".
8513 * nspawn will now inform the user explicitly that kernels with
8514 audit enabled break containers, and suggest the user to turn
8517 * Support for detecting the IMA and AppArmor security
8518 frameworks with ConditionSecurity= has been added.
8520 * journalctl gained a new "-k" switch for showing only kernel
8521 messages, mimicking dmesg output; in addition to "--user"
8522 and "--system" switches for showing only user's own logs
8525 * systemd-delta can now show information about drop-in
8526 snippets extending unit files.
8528 * libsystemd-bus has been substantially updated but is still
8529 not available as public API.
8531 * systemd will now look for the "debug" argument on the kernel
8532 command line and enable debug logging, similar to what
8533 "systemd.log_level=debug" already did before.
8535 * "systemctl set-default", "systemctl get-default" has been
8536 added to configure the default.target symlink, which
8537 controls what to boot into by default.
8539 * "systemctl set-log-level" has been added as a convenient
8540 way to raise and lower systemd logging threshold.
8542 * "systemd-analyze plot" will now show the time the various
8543 generators needed for execution, as well as information
8544 about the unit file loading.
8546 * libsystemd-journal gained a new sd_journal_open_files() call
8547 for opening specific journal files. journactl also gained a
8548 new switch to expose this new functionality. Previously we
8549 only supported opening all files from a directory, or all
8550 files from the system, as opening individual files only is
8551 racy due to journal file rotation.
8553 * systemd gained the new DefaultEnvironment= setting in
8554 /etc/systemd/system.conf to set environment variables for
8557 * If a privileged process logs a journal message with the
8558 OBJECT_PID= field set, then journald will automatically
8559 augment this with additional OBJECT_UID=, OBJECT_GID=,
8560 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
8561 system services want to log events about specific client
8562 processes. journactl/systemctl has been updated to make use
8563 of this information if all log messages regarding a specific
8566 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
8567 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
8568 Reisner, David Coppa, David King, David Strauss, Eelco
8569 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
8570 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
8571 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
8572 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
8573 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
8574 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
8575 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
8576 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
8577 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
8578 Łukasz Stelmach, 장동준
8582 * The Python bindings gained some minimal support for the APIs
8583 exposed by libsystemd-logind.
8585 * ConditionSecurity= gained support for detecting SMACK. Since
8586 this condition already supports SELinux and AppArmor we only
8587 miss IMA for this. Patches welcome!
8589 Contributions from: Karol Lewandowski, Lennart Poettering,
8590 Zbigniew Jędrzejewski-Szmek
8594 * systemd-nspawn will now create /etc/resolv.conf if
8595 necessary, before bind-mounting the host's file onto it.
8597 * systemd-nspawn will now store meta information about a
8598 container on the container's cgroup as extended attribute
8599 fields, including the root directory.
8601 * The cgroup hierarchy has been reworked in many ways. All
8602 objects any of the components systemd creates in the cgroup
8603 tree are now suffixed. More specifically, user sessions are
8604 now placed in cgroups suffixed with ".session", users in
8605 cgroups suffixed with ".user", and nspawn containers in
8606 cgroups suffixed with ".nspawn". Furthermore, all cgroup
8607 names are now escaped in a simple scheme to avoid collision
8608 of userspace object names with kernel filenames. This work
8609 is preparation for making these objects relocatable in the
8610 cgroup tree, in order to allow easy resource partitioning of
8611 these objects without causing naming conflicts.
8613 * systemctl list-dependencies gained the new switches
8614 --plain, --reverse, --after and --before.
8616 * systemd-inhibit now shows the process name of processes that
8617 have taken an inhibitor lock.
8619 * nss-myhostname will now also resolve "localhost"
8620 implicitly. This makes /etc/hosts an optional file and
8621 nicely handles that on IPv6 ::1 maps to both "localhost" and
8624 * libsystemd-logind.so gained a new call
8625 sd_get_machine_names() to enumerate running containers and
8626 VMs (currently only supported by very new libvirt and
8627 nspawn). sd_login_monitor can now be used to watch
8628 VMs/containers coming and going.
8630 * .include is not allowed recursively anymore, and only in
8631 unit files. Usually it is better to use drop-in snippets in
8632 .d/*.conf anyway, as introduced with systemd 198.
8634 * systemd-analyze gained a new "critical-chain" command that
8635 determines the slowest chain of units run during system
8636 boot-up. It is very useful for tracking down where
8637 optimizing boot time is the most beneficial.
8639 * systemd will no longer allow manipulating service paths in
8640 the name=systemd:/system cgroup tree using ControlGroup= in
8641 units. (But is still fine with it in all other dirs.)
8643 * There's a new systemd-nspawn@.service service file that may
8644 be used to easily run nspawn containers as system
8645 services. With the container's root directory in
8646 /var/lib/container/foobar it is now sufficient to run
8647 "systemctl start systemd-nspawn@foobar.service" to boot it.
8649 * systemd-cgls gained a new parameter "--machine" to list only
8650 the processes within a certain container.
8652 * ConditionSecurity= now can check for "apparmor". We still
8653 are lacking checks for SMACK and IMA for this condition
8654 check though. Patches welcome!
8656 * A new configuration file /etc/systemd/sleep.conf has been
8657 added that may be used to configure which kernel operation
8658 systemd is supposed to execute when "suspend", "hibernate"
8659 or "hybrid-sleep" is requested. This makes the new kernel
8660 "freeze" state accessible to the user.
8662 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
8663 the passed argument if applicable.
8665 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
8666 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
8667 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
8668 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
8669 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
8670 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
8671 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
8676 * The output of 'systemctl list-jobs' got some polishing. The
8677 '--type=' argument may now be passed more than once. A new
8678 command 'systemctl list-sockets' has been added which shows
8679 a list of kernel sockets systemd is listening on with the
8680 socket units they belong to, plus the units these socket
8683 * The experimental libsystemd-bus library got substantial
8684 updates to work in conjunction with the (also experimental)
8685 kdbus kernel project. It works well enough to exchange
8686 messages with some sophistication. Note that kdbus is not
8687 ready yet, and the library is mostly an elaborate test case
8688 for now, and not installable.
8690 * systemd gained a new unit 'systemd-static-nodes.service'
8691 that generates static device nodes earlier during boot, and
8692 can run in conjunction with udev.
8694 * libsystemd-login gained a new call sd_pid_get_user_unit()
8695 to retrieve the user systemd unit a process is running
8696 in. This is useful for systems where systemd is used as
8699 * systemd-nspawn now places all containers in the new /machine
8700 top-level cgroup directory in the name=systemd
8701 hierarchy. libvirt will soon do the same, so that we get a
8702 uniform separation of /system, /user and /machine for system
8703 services, user processes and containers/virtual
8704 machines. This new cgroup hierarchy is also useful to stick
8705 stable names to specific container instances, which can be
8706 recognized later this way (this name may be controlled
8707 via systemd-nspawn's new -M switch). libsystemd-login also
8708 gained a new call sd_pid_get_machine_name() to retrieve the
8709 name of the container/VM a specific process belongs to.
8711 * bootchart can now store its data in the journal.
8713 * libsystemd-journal gained a new call
8714 sd_journal_add_conjunction() for AND expressions to the
8715 matching logic. This can be used to express more complex
8716 logical expressions.
8718 * journactl can now take multiple --unit= and --user-unit=
8721 * The cryptsetup logic now understands the "luks.key=" kernel
8722 command line switch for specifying a file to read the
8723 decryption key from. Also, if a configured key file is not
8724 found the tool will now automatically fall back to prompting
8727 * Python systemd.journal module was updated to wrap recently
8728 added functions from libsystemd-journal. The interface was
8729 changed to bring the low level interface in s.j._Reader
8730 closer to the C API, and the high level interface in
8731 s.j.Reader was updated to wrap and convert all data about
8734 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
8735 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
8736 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
8737 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
8738 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
8739 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
8743 * journalctl --update-catalog now understands a new --root=
8744 option to operate on catalogs found in a different root
8747 * During shutdown after systemd has terminated all running
8748 services a final killing loop kills all remaining left-over
8749 processes. We will now print the name of these processes
8750 when we send SIGKILL to them, since this usually indicates a
8753 * If /etc/crypttab refers to password files stored on
8754 configured mount points automatic dependencies will now be
8755 generated to ensure the specific mount is established first
8756 before the key file is attempted to be read.
8758 * 'systemctl status' will now show information about the
8759 network sockets a socket unit is listening on.
8761 * 'systemctl status' will also shown information about any
8762 drop-in configuration file for units. (Drop-In configuration
8763 files in this context are files such as
8764 /etc/systemd/systemd/foobar.service.d/*.conf)
8766 * systemd-cgtop now optionally shows summed up CPU times of
8767 cgroups. Press '%' while running cgtop to switch between
8768 percentage and absolute mode. This is useful to determine
8769 which cgroups use up the most CPU time over the entire
8770 runtime of the system. systemd-cgtop has also been updated
8771 to be 'pipeable' for processing with further shell tools.
8773 * 'hostnamectl set-hostname' will now allow setting of FQDN
8776 * The formatting and parsing of time span values has been
8777 changed. The parser now understands fractional expressions
8778 such as "5.5h". The formatter will now output fractional
8779 expressions for all time spans under 1min, i.e. "5.123456s"
8780 rather than "5s 123ms 456us". For time spans under 1s
8781 millisecond values are shown, for those under 1ms
8782 microsecond values are shown. This should greatly improve
8783 all time-related output of systemd.
8785 * libsystemd-login and libsystemd-journal gained new
8786 functions for querying the poll() events mask and poll()
8787 timeout value for integration into arbitrary event
8790 * localectl gained the ability to list available X11 keymaps
8791 (models, layouts, variants, options).
8793 * 'systemd-analyze dot' gained the ability to filter for
8794 specific units via shell-style globs, to create smaller,
8795 more useful graphs. I.e. it is now possible to create simple
8796 graphs of all the dependencies between only target units, or
8797 of all units that Avahi has dependencies with.
8799 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
8800 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
8801 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
8802 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
8803 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
8804 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
8805 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
8809 * The boot-time readahead implementation for rotating media
8810 will now read the read-ahead data in multiple passes which
8811 consist of all read requests made in equidistant time
8812 intervals. This means instead of strictly reading read-ahead
8813 data in its physical order on disk we now try to find a
8814 middle ground between physical and access time order.
8816 * /etc/os-release files gained a new BUILD_ID= field for usage
8817 on operating systems that provide continuous builds of OS
8820 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
8821 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
8822 William Douglas, Zbigniew Jędrzejewski-Szmek
8826 * systemd-python gained an API exposing libsystemd-daemon.
8828 * The SMACK setup logic gained support for uploading CIPSO
8831 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
8832 ReadOnlyDirectories= and InaccessibleDirectories= has
8833 changed. The private /tmp and /var/tmp directories are now
8834 shared by all processes of a service (which means
8835 ExecStartPre= may now leave data in /tmp that ExecStart= of
8836 the same service can still access). When a service is
8837 stopped its temporary directories are immediately deleted
8838 (normal clean-up with tmpfiles is still done in addition to
8841 * By default, systemd will now set a couple of sysctl
8842 variables in the kernel: the safe sysrq options are turned
8843 on, IP route verification is turned on, and source routing
8844 disabled. The recently added hardlink and softlink
8845 protection of the kernel is turned on. These settings should
8846 be reasonably safe, and good defaults for all new systems.
8848 * The predictable network naming logic may now be turned off
8849 with a new kernel command line switch: net.ifnames=0.
8851 * A new libsystemd-bus module has been added that implements a
8852 pretty complete D-Bus client library. For details see:
8854 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
8856 * journald will now explicitly flush the journal files to disk
8857 at the latest 5min after each write. The file will then also
8858 be marked offline until the next write. This should increase
8859 reliability in case of a crash. The synchronization delay
8860 can be configured via SyncIntervalSec= in journald.conf.
8862 * There's a new remote-fs-setup.target unit that can be used
8863 to pull in specific services when at least one remote file
8864 system is to be mounted.
8866 * There are new targets timers.target and paths.target as
8867 canonical targets to pull user timer and path units in
8868 from. This complements sockets.target with a similar
8869 purpose for socket units.
8871 * libudev gained a new call udev_device_set_attribute_value()
8872 to set sysfs attributes of a device.
8874 * The udev daemon now sets the default number of worker
8875 processes executed in parallel based on the number of available
8876 CPUs instead of the amount of available RAM. This is supposed
8877 to provide a more reliable default and limit a too aggressive
8878 parallelism for setups with 1000s of devices connected.
8880 Contributions from: Auke Kok, Colin Walters, Cristian
8881 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
8882 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
8883 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
8884 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
8885 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
8886 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
8887 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
8888 Zbigniew Jędrzejewski-Szmek
8892 * Configuration of unit files may now be extended via drop-in
8893 files without having to edit/override the unit files
8894 themselves. More specifically, if the administrator wants to
8895 change one value for a service file foobar.service he can
8896 now do so by dropping in a configuration snippet into
8897 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
8898 will load all these snippets and apply them on top of the
8899 main unit configuration file, possibly extending or
8900 overriding its settings. Using these drop-in snippets is
8901 generally nicer than the two earlier options for changing
8902 unit files locally: copying the files from
8903 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
8904 them there; or creating a new file in /etc/systemd/system/
8905 that incorporates the original one via ".include". Drop-in
8906 snippets into these .d/ directories can be placed in any
8907 directory systemd looks for units in, and the usual
8908 overriding semantics between /usr/lib, /etc and /run apply
8911 * Most unit file settings which take lists of items can now be
8912 reset by assigning the empty string to them. For example,
8913 normally, settings such as Environment=FOO=BAR append a new
8914 environment variable assignment to the environment block,
8915 each time they are used. By assigning Environment= the empty
8916 string the environment block can be reset to empty. This is
8917 particularly useful with the .d/*.conf drop-in snippets
8918 mentioned above, since this adds the ability to reset list
8919 settings from vendor unit files via these drop-ins.
8921 * systemctl gained a new "list-dependencies" command for
8922 listing the dependencies of a unit recursively.
8924 * Inhibitors are now honored and listed by "systemctl
8925 suspend", "systemctl poweroff" (and similar) too, not only
8926 GNOME. These commands will also list active sessions by
8929 * Resource limits (as exposed by the various control group
8930 controllers) can now be controlled dynamically at runtime
8931 for all units. More specifically, you can now use a command
8932 like "systemctl set-cgroup-attr foobar.service cpu.shares
8933 2000" to alter the CPU shares a specific service gets. These
8934 settings are stored persistently on disk, and thus allow the
8935 administrator to easily adjust the resource usage of
8936 services with a few simple commands. This dynamic resource
8937 management logic is also available to other programs via the
8938 bus. Almost any kernel cgroup attribute and controller is
8941 * systemd-vconsole-setup will now copy all font settings to
8942 all allocated VTs, where it previously applied them only to
8945 * libsystemd-login gained the new sd_session_get_tty() API
8948 * This release drops support for a few legacy or
8949 distribution-specific LSB facility names when parsing init
8950 scripts: $x-display-manager, $mail-transfer-agent,
8951 $mail-transport-agent, $mail-transfer-agent, $smtp,
8952 $null. Also, the mail-transfer-agent.target unit backing
8953 this has been removed. Distributions which want to retain
8954 compatibility with this should carry the burden for
8955 supporting this themselves and patch support for these back
8956 in, if they really need to. Also, the facilities $syslog and
8957 $local_fs are now ignored, since systemd does not support
8958 early-boot LSB init scripts anymore, and these facilities
8959 are implied anyway for normal services. syslog.target has
8962 * There are new bus calls on PID1's Manager object for
8963 cancelling jobs, and removing snapshot units. Previously,
8964 both calls were only available on the Job and Snapshot
8967 * systemd-journal-gatewayd gained SSL support.
8969 * The various "environment" files, such as /etc/locale.conf
8970 now support continuation lines with a backslash ("\") as
8971 last character in the line, similarly in style (but different)
8972 to how this is supported in shells.
8974 * For normal user processes the _SYSTEMD_USER_UNIT= field is
8975 now implicitly appended to every log entry logged. systemctl
8976 has been updated to filter by this field when operating on a
8977 user systemd instance.
8979 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
8980 CAP_AUDIT_CONTROL capabilities to the capabilities set for
8981 the container. This makes it easier to boot unmodified
8982 Fedora systems in a container, which however still requires
8983 audit=0 to be passed on the kernel command line. Auditing in
8984 kernel and userspace is unfortunately still too broken in
8985 context of containers, hence we recommend compiling it out
8986 of the kernel or using audit=0. Hopefully this will be fixed
8987 one day for good in the kernel.
8989 * nspawn gained the new --bind= and --bind-ro= parameters to
8990 bind mount specific directories from the host into the
8993 * nspawn will now mount its own devpts file system instance
8994 into the container, in order not to leak pty devices from
8995 the host into the container.
8997 * systemd will now read the firmware boot time performance
8998 information from the EFI variables, if the used boot loader
8999 supports this, and takes it into account for boot performance
9000 analysis via "systemd-analyze". This is currently supported
9001 only in conjunction with Gummiboot, but could be supported
9002 by other boot loaders too. For details see:
9004 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
9006 * A new generator has been added that automatically mounts the
9007 EFI System Partition (ESP) to /boot, if that directory
9008 exists, is empty, and no other file system has been
9009 configured to be mounted there.
9011 * logind will now send out PrepareForSleep(false) out
9012 unconditionally, after coming back from suspend. This may be
9013 used by applications as asynchronous notification for
9014 system resume events.
9016 * "systemctl unlock-sessions" has been added, that allows
9017 unlocking the screens of all user sessions at once, similar
9018 to how "systemctl lock-sessions" already locked all users
9019 sessions. This is backed by a new D-Bus call UnlockSessions().
9021 * "loginctl seat-status" will now show the master device of a
9022 seat. (i.e. the device of a seat that needs to be around for
9023 the seat to be considered available, usually the graphics
9026 * tmpfiles gained a new "X" line type, that allows
9027 configuration of files and directories (with wildcards) that
9028 shall be excluded from automatic cleanup ("aging").
9030 * udev default rules set the device node permissions now only
9031 at "add" events, and do not change them any longer with a
9032 later "change" event.
9034 * The log messages for lid events and power/sleep keypresses
9035 now carry a message ID.
9037 * We now have a substantially larger unit test suite, but this
9038 continues to be work in progress.
9040 * udevadm hwdb gained a new --root= parameter to change the
9041 root directory to operate relative to.
9043 * logind will now issue a background sync() request to the kernel
9044 early at shutdown, so that dirty buffers are flushed to disk early
9045 instead of at the last moment, in order to optimize shutdown
9048 * A new bootctl tool has been added that is an interface for
9049 certain boot loader operations. This is currently a preview
9050 and is likely to be extended into a small mechanism daemon
9051 like timedated, localed, hostnamed, and can be used by
9052 graphical UIs to enumerate available boot options, and
9053 request boot into firmware operations.
9055 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
9056 the rest of the package. It also has been updated to work
9057 correctly in initrds.
9059 * polkit previously has been runtime optional, and is now also
9060 compile time optional via a configure switch.
9062 * systemd-analyze has been reimplemented in C. Also "systemctl
9063 dot" has moved into systemd-analyze.
9065 * "systemctl status" with no further parameters will now print
9066 the status of all active or failed units.
9068 * Operations such as "systemctl start" can now be executed
9069 with a new mode "--irreversible" which may be used to queue
9070 operations that cannot accidentally be reversed by a later
9071 job queuing. This is by default used to make shutdown
9072 requests more robust.
9074 * The Python API of systemd now gained a new module for
9075 reading journal files.
9077 * A new tool kernel-install has been added that can install
9078 kernel images according to the Boot Loader Specification:
9080 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
9082 * Boot time console output has been improved to provide
9083 animated boot time output for hanging jobs.
9085 * A new tool systemd-activate has been added which can be used
9086 to test socket activation with, directly from the command
9087 line. This should make it much easier to test and debug
9088 socket activation in daemons.
9090 * journalctl gained a new "--reverse" (or -r) option to show
9091 journal output in reverse order (i.e. newest line first).
9093 * journalctl gained a new "--pager-end" (or -e) option to jump
9094 to immediately jump to the end of the journal in the
9095 pager. This is only supported in conjunction with "less".
9097 * journalctl gained a new "--user-unit=" option, that works
9098 similarly to "--unit=" but filters for user units rather than
9101 * A number of unit files to ease adoption of systemd in
9102 initrds has been added. This moves some minimal logic from
9103 the various initrd implementations into systemd proper.
9105 * The journal files are now owned by a new group
9106 "systemd-journal", which exists specifically to allow access
9107 to the journal, and nothing else. Previously, we used the
9108 "adm" group for that, which however possibly covers more
9109 than just journal/log file access. This new group is now
9110 already used by systemd-journal-gatewayd to ensure this
9111 daemon gets access to the journal files and as little else
9112 as possible. Note that "make install" will also set FS ACLs
9113 up for /var/log/journal to give "adm" and "wheel" read
9114 access to it, in addition to "systemd-journal" which owns
9115 the journal files. We recommend that packaging scripts also
9116 add read access to "adm" + "wheel" to /var/log/journal, and
9117 all existing/future journal files. To normal users and
9118 administrators little changes, however packagers need to
9119 ensure to create the "systemd-journal" system group at
9120 package installation time.
9122 * The systemd-journal-gatewayd now runs as unprivileged user
9123 systemd-journal-gateway:systemd-journal-gateway. Packaging
9124 scripts need to create these system user/group at
9127 * timedated now exposes a new boolean property CanNTP that
9128 indicates whether a local NTP service is available or not.
9130 * systemd-detect-virt will now also detect xen PVs
9132 * The pstore file system is now mounted by default, if it is
9135 * In addition to the SELinux and IMA policies we will now also
9136 load SMACK policies at early boot.
9138 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
9139 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
9140 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
9141 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
9142 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
9143 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
9144 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
9145 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
9146 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
9147 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
9148 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
9149 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
9150 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
9151 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
9155 * Timer units now support calendar time events in addition to
9156 monotonic time events. That means you can now trigger a unit
9157 based on a calendar time specification such as "Thu,Fri
9158 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
9159 or fifth day of any month of the year 2013, given that it is
9160 a thursday or friday. This brings timer event support
9161 considerably closer to cron's capabilities. For details on
9162 the supported calendar time specification language see
9165 * udev now supports a number of different naming policies for
9166 network interfaces for predictable names, and a combination
9167 of these policies is now the default. Please see this wiki
9168 document for details:
9170 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
9172 * Auke Kok's bootchart implementation has been added to the
9173 systemd tree. It is an optional component that can graph the
9174 boot in quite some detail. It is one of the best bootchart
9175 implementations around and minimal in its code and
9178 * nss-myhostname has been integrated into the systemd source
9179 tree. nss-myhostname guarantees that the local hostname
9180 always stays resolvable via NSS. It has been a weak
9181 requirement of systemd-hostnamed since a long time, and
9182 since its code is actually trivial we decided to just
9183 include it in systemd's source tree. It can be turned off
9184 with a configure switch.
9186 * The read-ahead logic is now capable of properly detecting
9187 whether a btrfs file system is on SSD or rotating media, in
9188 order to optimize the read-ahead scheme. Previously, it was
9189 only capable of detecting this on traditional file systems
9192 * In udev, additional device properties are now read from the
9193 IAB in addition to the OUI database. Also, Bluetooth company
9194 identities are attached to the devices as well.
9196 * In service files %U may be used as specifier that is
9197 replaced by the configured user name of the service.
9199 * nspawn may now be invoked without a controlling TTY. This
9200 makes it suitable for invocation as its own service. This
9201 may be used to set up a simple containerized server system
9202 using only core OS tools.
9204 * systemd and nspawn can now accept socket file descriptors
9205 when they are started for socket activation. This enables
9206 implementation of socket activated nspawn
9207 containers. i.e. think about autospawning an entire OS image
9208 when the first SSH or HTTP connection is received. We expect
9209 that similar functionality will also be added to libvirt-lxc
9212 * journalctl will now suppress ANSI color codes when
9213 presenting log data.
9215 * systemctl will no longer show control group information for
9216 a unit if the control group is empty anyway.
9218 * logind can now automatically suspend/hibernate/shutdown the
9221 * /etc/machine-info and hostnamed now also expose the chassis
9222 type of the system. This can be used to determine whether
9223 the local system is a laptop, desktop, handset or
9224 tablet. This information may either be configured by the
9225 user/vendor or is automatically determined from ACPI and DMI
9226 information if possible.
9228 * A number of polkit actions are now bound together with "imply"
9229 rules. This should simplify creating UIs because many actions
9230 will now authenticate similar ones as well.
9232 * Unit files learnt a new condition ConditionACPower= which
9233 may be used to conditionalize a unit depending on whether an
9234 AC power source is connected or not, of whether the system
9235 is running on battery power.
9237 * systemctl gained a new "is-failed" verb that may be used in
9238 shell scripts and suchlike to check whether a specific unit
9239 is in the "failed" state.
9241 * The EnvironmentFile= setting in unit files now supports file
9242 globbing, and can hence be used to easily read a number of
9243 environment files at once.
9245 * systemd will no longer detect and recognize specific
9246 distributions. All distribution-specific #ifdeffery has been
9247 removed, systemd is now fully generic and
9248 distribution-agnostic. Effectively, not too much is lost as
9249 a lot of the code is still accessible via explicit configure
9250 switches. However, support for some distribution specific
9251 legacy configuration file formats has been dropped. We
9252 recommend distributions to simply adopt the configuration
9253 files everybody else uses now and convert the old
9254 configuration from packaging scripts. Most distributions
9255 already did that. If that's not possible or desirable,
9256 distributions are welcome to forward port the specific
9257 pieces of code locally from the git history.
9259 * When logging a message about a unit systemd will now always
9260 log the unit name in the message meta data.
9262 * localectl will now also discover system locale data that is
9263 not stored in locale archives, but directly unpacked.
9265 * logind will no longer unconditionally use framebuffer
9266 devices as seat masters, i.e. as devices that are required
9267 to be existing before a seat is considered preset. Instead,
9268 it will now look for all devices that are tagged as
9269 "seat-master" in udev. By default, framebuffer devices will
9270 be marked as such, but depending on local systems, other
9271 devices might be marked as well. This may be used to
9272 integrate graphics cards using closed source drivers (such
9273 as NVidia ones) more nicely into logind. Note however, that
9274 we recommend using the open source NVidia drivers instead,
9275 and no udev rules for the closed-source drivers will be
9276 shipped from us upstream.
9278 Contributions from: Adam Williamson, Alessandro Crismani, Auke
9279 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
9280 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
9281 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
9282 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
9283 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
9284 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
9285 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
9286 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
9287 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
9288 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
9289 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
9294 * udev gained support for loading additional device properties
9295 from an indexed database that is keyed by vendor/product IDs
9296 and similar device identifiers. For the beginning this
9297 "hwdb" is populated with data from the well-known PCI and
9298 USB database, but also includes PNP, ACPI and OID data. In
9299 the longer run this indexed database shall grow into
9300 becoming the one central database for non-essential
9301 userspace device metadata. Previously, data from the PCI/USB
9302 database was only attached to select devices, since the
9303 lookup was a relatively expensive operation due to O(n) time
9304 complexity (with n being the number of entries in the
9305 database). Since this is now O(1), we decided to add in this
9306 data for all devices where this is available, by
9307 default. Note that the indexed database needs to be rebuilt
9308 when new data files are installed. To achieve this you need
9309 to update your packaging scripts to invoke "udevadm hwdb
9310 --update" after installation of hwdb data files. For
9311 RPM-based distributions we introduced the new
9312 %udev_hwdb_update macro for this purpose.
9314 * The Journal gained support for the "Message Catalog", an
9315 indexed database to link up additional information with
9316 journal entries. For further details please check:
9318 https://www.freedesktop.org/wiki/Software/systemd/catalog
9320 The indexed message catalog database also needs to be
9321 rebuilt after installation of message catalog files. Use
9322 "journalctl --update-catalog" for this. For RPM-based
9323 distributions we introduced the %journal_catalog_update
9324 macro for this purpose.
9326 * The Python Journal bindings gained support for the standard
9327 Python logging framework.
9329 * The Journal API gained new functions for checking whether
9330 the underlying file system of a journal file is capable of
9331 properly reporting file change notifications, or whether
9332 applications that want to reflect journal changes "live"
9333 need to recheck journal files continuously in appropriate
9336 * It is now possible to set the "age" field for tmpfiles
9337 entries to 0, indicating that files matching this entry
9338 shall always be removed when the directories are cleaned up.
9340 * coredumpctl gained a new "gdb" verb which invokes gdb
9341 right-away on the selected coredump.
9343 * There's now support for "hybrid sleep" on kernels that
9344 support this, in addition to "suspend" and "hibernate". Use
9345 "systemctl hybrid-sleep" to make use of this.
9347 * logind's HandleSuspendKey= setting (and related settings)
9348 now gained support for a new "lock" setting to simply
9349 request the screen lock on all local sessions, instead of
9350 actually executing a suspend or hibernation.
9352 * systemd will now mount the EFI variables file system by
9355 * Socket units now gained support for configuration of the
9356 SMACK security label.
9358 * timedatectl will now output the time of the last and next
9359 daylight saving change.
9361 * We dropped support for various legacy and distro-specific
9362 concepts, such as insserv, early-boot SysV services
9363 (i.e. those for non-standard runlevels such as 'b' or 'S')
9364 or ArchLinux /etc/rc.conf support. We recommend the
9365 distributions who still need support this to either continue
9366 to maintain the necessary patches downstream, or find a
9367 different solution. (Talk to us if you have questions!)
9369 * Various systemd components will now bypass polkit checks for
9370 root and otherwise handle properly if polkit is not found to
9371 be around. This should fix most issues for polkit-less
9372 systems. Quite frankly this should have been this way since
9373 day one. It is absolutely our intention to make systemd work
9374 fine on polkit-less systems, and we consider it a bug if
9375 something does not work as it should if polkit is not around.
9377 * For embedded systems it is now possible to build udev and
9378 systemd without blkid and/or kmod support.
9380 * "systemctl switch-root" is now capable of switching root
9381 more than once. I.e. in addition to transitions from the
9382 initrd to the host OS it is now possible to transition to
9383 further OS images from the host. This is useful to implement
9384 offline updating tools.
9386 * Various other additions have been made to the RPM macros
9387 shipped with systemd. Use %udev_rules_update() after
9388 installing new udev rules files. %_udevhwdbdir,
9389 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
9390 %_sysctldir are now available which resolve to the right
9391 directories for packages to place various data files in.
9393 * journalctl gained the new --full switch (in addition to
9394 --all, to disable ellipsation for long messages.
9396 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
9397 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
9398 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
9399 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
9400 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
9401 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
9402 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
9403 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
9404 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
9408 * journalctl gained new --since= and --until= switches to
9409 filter by time. It also now supports nice filtering for
9410 units via --unit=/-u.
9412 * Type=oneshot services may use ExecReload= and do the
9415 * The journal daemon now supports time-based rotation and
9416 vacuuming, in addition to the usual disk-space based
9419 * The journal will now index the available field values for
9420 each field name. This enables clients to show pretty drop
9421 downs of available match values when filtering. The bash
9422 completion of journalctl has been updated
9423 accordingly. journalctl gained a new switch -F to list all
9424 values a certain field takes in the journal database.
9426 * More service events are now written as structured messages
9427 to the journal, and made recognizable via message IDs.
9429 * The timedated, localed and hostnamed mini-services which
9430 previously only provided support for changing time, locale
9431 and hostname settings from graphical DEs such as GNOME now
9432 also have a minimal (but very useful) text-based client
9433 utility each. This is probably the nicest way to changing
9434 these settings from the command line now, especially since
9435 it lists available options and is fully integrated with bash
9438 * There's now a new tool "systemd-coredumpctl" to list and
9439 extract coredumps from the journal.
9441 * We now install a README each in /var/log/ and
9442 /etc/rc.d/init.d explaining where the system logs and init
9443 scripts went. This hopefully should help folks who go to
9444 that dirs and look into the otherwise now empty void and
9445 scratch their heads.
9447 * When user-services are invoked (by systemd --user) the
9448 $MANAGERPID env var is set to the PID of systemd.
9450 * SIGRTMIN+24 when sent to a --user instance will now result
9451 in immediate termination of systemd.
9453 * gatewayd received numerous feature additions such as a
9454 "follow" mode, for live syncing and filtering.
9456 * browse.html now allows filtering and showing detailed
9457 information on specific entries. Keyboard navigation and
9458 mouse screen support has been added.
9460 * gatewayd/journalctl now supports HTML5/JSON
9461 Server-Sent-Events as output.
9463 * The SysV init script compatibility logic will now
9464 heuristically determine whether a script supports the
9465 "reload" verb, and only then make this available as
9468 * "systemctl status --follow" has been removed, use "journalctl
9471 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
9472 have been removed since they are hardly useful to be
9475 * And I'd like to take the opportunity to specifically mention
9476 Zbigniew for his great contributions. Zbigniew, you rock!
9478 Contributions from: Andrew Eikum, Christian Hesse, Colin
9479 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
9480 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
9481 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
9482 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
9483 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
9484 Jędrzejewski-Szmek, Сковорода Никита Андреевич
9488 * If /etc/vconsole.conf is non-existent or empty we will no
9489 longer load any console font or key map at boot by
9490 default. Instead the kernel defaults will be left
9491 intact. This is definitely the right thing to do, as no
9492 configuration should mean no configuration, and hard-coding
9493 font names that are different on all archs is probably a bad
9494 idea. Also, the kernel default key map and font should be
9495 good enough for most cases anyway, and mostly identical to
9496 the userspace fonts/key maps we previously overloaded them
9497 with. If distributions want to continue to default to a
9498 non-kernel font or key map they should ship a default
9499 /etc/vconsole.conf with the appropriate contents.
9501 Contributions from: Colin Walters, Daniel J Walsh, Dave
9502 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
9503 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
9507 * journalctl gained a new --cursor= switch to show entries
9508 starting from the specified location in the journal.
9510 * We now enforce a size limit on journal entry fields exported
9511 with "-o json" in journalctl. Fields larger than 4K will be
9512 assigned null. This can be turned off with --all.
9514 * An (optional) journal gateway daemon is now available as
9515 "systemd-journal-gatewayd.service". This service provides
9516 access to the journal via HTTP and JSON. This functionality
9517 will be used to implement live log synchronization in both
9518 pull and push modes, but has various other users too, such
9519 as easy log access for debugging of embedded devices. Right
9520 now it is already useful to retrieve the journal via HTTP:
9522 # systemctl start systemd-journal-gatewayd.service
9523 # wget http://localhost:19531/entries
9525 This will download the journal contents in a
9526 /var/log/messages compatible format. The same as JSON:
9528 # curl -H"Accept: application/json" http://localhost:19531/entries
9530 This service is also accessible via a web browser where a
9531 single static HTML5 app is served that uses the JSON logic
9532 to enable the user to do some basic browsing of the
9533 journal. This will be extended later on. Here's an example
9534 screenshot of this app in its current state:
9536 http://0pointer.de/public/journal-gatewayd
9538 Contributions from: Kay Sievers, Lennart Poettering, Robert
9539 Milasan, Tom Gundersen
9543 * The bash completion logic is now available for journalctl
9546 * We do not mount the "cpuset" controller anymore together with
9547 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
9548 started if no parameters are assigned to it. "cpuset" hence
9549 broke code that assumed it could create "cpu" groups and
9552 * journalctl -f will now subscribe to terminal size changes,
9553 and line break accordingly.
9555 Contributions from: Dave Reisner, Kay Sievers, Lennart
9556 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
9560 * nspawn will now create a symlink /etc/localtime in the
9561 container environment, copying the host's timezone
9562 setting. Previously this has been done via a bind mount, but
9563 since symlinks cannot be bind mounted this has now been
9564 changed to create/update the appropriate symlink.
9566 * journalctl -n's line number argument is now optional, and
9567 will default to 10 if omitted.
9569 * journald will now log the maximum size the journal files may
9570 take up on disk. This is particularly useful if the default
9571 built-in logic of determining this parameter from the file
9572 system size is used. Use "systemctl status
9573 systemd-journald.service" to see this information.
9575 * The multi-seat X wrapper tool has been stripped down. As X
9576 is now capable of enumerating graphics devices via udev in a
9577 seat-aware way the wrapper is not strictly necessary
9578 anymore. A stripped down temporary stop-gap is still shipped
9579 until the upstream display managers have been updated to
9580 fully support the new X logic. Expect this wrapper to be
9581 removed entirely in one of the next releases.
9583 * HandleSleepKey= in logind.conf has been split up into
9584 HandleSuspendKey= and HandleHibernateKey=. The old setting
9585 is not available anymore. X11 and the kernel are
9586 distinguishing between these keys and we should too. This
9587 also means the inhibition lock for these keys has been split
9590 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
9591 Poettering, Lukas Nykryn, Václav Pavlín
9595 * Whenever a unit changes state we will now log this to the
9596 journal and show along the unit's own log output in
9599 * ConditionPathIsMountPoint= can now properly detect bind
9600 mount points too. (Previously, a bind mount of one file
9601 system to another place in the same file system could not be
9602 detected as mount, since they shared struct stat's st_dev
9605 * We will now mount the cgroup controllers cpu, cpuacct,
9606 cpuset and the controllers net_cls, net_prio together by
9609 * nspawn containers will now have a virtualized boot
9610 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
9611 over with a randomized ID at container initialization). This
9612 has the effect of making "journalctl -b" do the right thing
9615 * The JSON output journal serialization has been updated not
9616 to generate "endless" list objects anymore, but rather one
9617 JSON object per line. This is more in line how most JSON
9618 parsers expect JSON objects. The new output mode
9619 "json-pretty" has been added to provide similar output, but
9620 neatly aligned for readability by humans.
9622 * We dropped all explicit sync() invocations in the shutdown
9623 code. The kernel does this implicitly anyway in the kernel
9624 reboot() syscall. halt(8)'s -n option is now a compatibility
9627 * We now support virtualized reboot() in containers, as
9628 supported by newer kernels. We will fall back to exit() if
9629 CAP_SYS_REBOOT is not available to the container. Also,
9630 nspawn makes use of this now and will actually reboot the
9631 container if the containerized OS asks for that.
9633 * journalctl will only show local log output by default
9634 now. Use --merge (-m) to show remote log output, too.
9636 * libsystemd-journal gained the new sd_journal_get_usage()
9637 call to determine the current disk usage of all journal
9638 files. This is exposed in the new "journalctl --disk-usage"
9641 * journald gained a new configuration setting SplitMode= in
9642 journald.conf which may be used to control how user journals
9643 are split off. See journald.conf(5) for details.
9645 * A new condition type ConditionFileNotEmpty= has been added.
9647 * tmpfiles' "w" lines now support file globbing, to write
9648 multiple files at once.
9650 * We added Python bindings for the journal submission
9651 APIs. More Python APIs for a number of selected APIs will
9652 likely follow. Note that we intend to add native bindings
9653 only for the Python language, as we consider it common
9654 enough to deserve bindings shipped within systemd. There are
9655 various projects outside of systemd that provide bindings
9656 for languages such as PHP or Lua.
9658 * Many conditions will now resolve specifiers such as %i. In
9659 addition, PathChanged= and related directives of .path units
9660 now support specifiers as well.
9662 * There's now a new RPM macro definition for the system preset
9665 * journald will now warn if it ca not forward a message to the
9666 syslog daemon because its socket is full.
9668 * timedated will no longer write or process /etc/timezone,
9669 except on Debian. As we do not support late mounted /usr
9670 anymore /etc/localtime always being a symlink is now safe,
9671 and hence the information in /etc/timezone is not necessary
9674 * logind will now always reserve one VT for a text getty (VT6
9675 by default). Previously if more than 6 X sessions where
9676 started they took up all the VTs with auto-spawned gettys,
9677 so that no text gettys were available anymore.
9679 * udev will now automatically inform the btrfs kernel logic
9680 about btrfs RAID components showing up. This should make
9681 simple hotplug based btrfs RAID assembly work.
9683 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
9684 (but not for its children which will stay at the kernel
9685 default). This should allow setups with a lot more listening
9688 * systemd will now always pass the configured timezone to the
9689 kernel at boot. timedated will do the same when the timezone
9692 * logind's inhibition logic has been updated. By default,
9693 logind will now handle the lid switch, the power and sleep
9694 keys all the time, even in graphical sessions. If DEs want
9695 to handle these events on their own they should take the new
9696 handle-power-key, handle-sleep-key and handle-lid-switch
9697 inhibitors during their runtime. A simple way to achieve
9698 that is to invoke the DE wrapped in an invocation of:
9700 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
9702 * Access to unit operations is now checked via SELinux taking
9703 the unit file label and client process label into account.
9705 * systemd will now notify the administrator in the journal
9706 when he over-mounts a non-empty directory.
9708 * There are new specifiers that are resolved in unit files,
9709 for the hostname (%H), the machine ID (%m) and the boot ID
9712 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
9713 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
9714 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
9715 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
9716 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
9717 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
9718 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
9722 * Support for reading structured kernel messages from
9723 /dev/kmsg has now been added and is enabled by default.
9725 * Support for reading kernel messages from /proc/kmsg has now
9726 been removed. If you want kernel messages in the journal
9727 make sure to run a recent kernel (>= 3.5) that supports
9728 reading structured messages from /dev/kmsg (see
9729 above). /proc/kmsg is now exclusive property of classic
9730 syslog daemons again.
9732 * The libudev API gained the new
9733 udev_device_new_from_device_id() call.
9735 * The logic for file system namespace (ReadOnlyDirectory=,
9736 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
9737 require pivot_root() anymore. This means fewer temporary
9738 directories are created below /tmp for this feature.
9740 * nspawn containers will now see and receive all submounts
9741 made on the host OS below the root file system of the
9744 * Forward Secure Sealing is now supported for Journal files,
9745 which provide cryptographical sealing of journal files so
9746 that attackers cannot alter log history anymore without this
9747 being detectable. Lennart will soon post a blog story about
9748 this explaining it in more detail.
9750 * There are two new service settings RestartPreventExitStatus=
9751 and SuccessExitStatus= which allow configuration of exit
9752 status (exit code or signal) which will be excepted from the
9753 restart logic, resp. consider successful.
9755 * journalctl gained the new --verify switch that can be used
9756 to check the integrity of the structure of journal files and
9757 (if Forward Secure Sealing is enabled) the contents of
9760 * nspawn containers will now be run with /dev/stdin, /dev/fd/
9761 and similar symlinks pre-created. This makes running shells
9762 as container init process a lot more fun.
9764 * The fstab support can now handle PARTUUID= and PARTLABEL=
9767 * A new ConditionHost= condition has been added to match
9768 against the hostname (with globs) and machine ID. This is
9769 useful for clusters where a single OS image is used to
9770 provision a large number of hosts which shall run slightly
9771 different sets of services.
9773 * Services which hit the restart limit will now be placed in a
9776 Contributions from: Bertram Poettering, Dave Reisner, Huang
9777 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
9778 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
9782 * When running in --user mode systemd will now become a
9783 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
9784 tree a lot more organized.
9786 * A new PartOf= unit dependency type has been introduced that
9787 may be used to group services in a natural way.
9789 * "systemctl enable" may now be used to enable instances of
9792 * journalctl now prints error log levels in red, and
9793 warning/notice log levels in bright white. It also supports
9794 filtering by log level now.
9796 * cgtop gained a new -n switch (similar to top), to configure
9797 the maximum number of iterations to run for. It also gained
9798 -b, to run in batch mode (accepting no input).
9800 * The suffix ".service" may now be omitted on most systemctl
9801 command lines involving service unit names.
9803 * There's a new bus call in logind to lock all sessions, as
9804 well as a loginctl verb for it "lock-sessions".
9806 * libsystemd-logind.so gained a new call sd_journal_perror()
9807 that works similar to libc perror() but logs to the journal
9808 and encodes structured information about the error number.
9810 * /etc/crypttab entries now understand the new keyfile-size=
9813 * shutdown(8) now can send a (configurable) wall message when
9814 a shutdown is cancelled.
9816 * The mount propagation mode for the root file system will now
9817 default to "shared", which is useful to make containers work
9818 nicely out-of-the-box so that they receive new mounts from
9819 the host. This can be undone locally by running "mount
9820 --make-rprivate /" if needed.
9822 * The prefdm.service file has been removed. Distributions
9823 should maintain this unit downstream if they intend to keep
9824 it around. However, we recommend writing normal unit files
9825 for display managers instead.
9827 * Since systemd is a crucial part of the OS we will now
9828 default to a number of compiler switches that improve
9829 security (hardening) such as read-only relocations, stack
9830 protection, and suchlike.
9832 * The TimeoutSec= setting for services is now split into
9833 TimeoutStartSec= and TimeoutStopSec= to allow configuration
9834 of individual time outs for the start and the stop phase of
9837 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
9838 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
9839 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
9840 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
9841 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
9842 Gundersen, Zbigniew Jędrzejewski-Szmek
9846 * The journal and id128 C APIs are now fully documented as man
9849 * Extra safety checks have been added when transitioning from
9850 the initial RAM disk to the main system to avoid accidental
9853 * /etc/crypttab entries now understand the new keyfile-offset=
9856 * systemctl -t can now be used to filter by unit load state.
9858 * The journal C API gained the new sd_journal_wait() call to
9859 make writing synchronous journal clients easier.
9861 * journalctl gained the new -D switch to show journals from a
9864 * journalctl now displays a special marker between log
9865 messages of two different boots.
9867 * The journal is now explicitly flushed to /var via a service
9868 systemd-journal-flush.service, rather than implicitly simply
9869 by seeing /var/log/journal to be writable.
9871 * journalctl (and the journal C APIs) can now match for much
9872 more complex expressions, with alternatives and
9875 * When transitioning from the initial RAM disk to the main
9876 system we will now kill all processes in a killing spree to
9877 ensure no processes stay around by accident.
9879 * Three new specifiers may be used in unit files: %u, %h, %s
9880 resolve to the user name, user home directory resp. user
9881 shell. This is useful for running systemd user instances.
9883 * We now automatically rotate journal files if their data
9884 object hash table gets a fill level > 75%. We also size the
9885 hash table based on the configured maximum file size. This
9886 together should lower hash collisions drastically and thus
9887 speed things up a bit.
9889 * journalctl gained the new "--header" switch to introspect
9890 header data of journal files.
9892 * A new setting SystemCallFilters= has been added to services
9893 which may be used to apply blacklists or whitelists to
9894 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
9896 * nspawn gained a new --link-journal= switch (and quicker: -j)
9897 to link the container journal with the host. This makes it
9898 very easy to centralize log viewing on the host for all
9899 guests while still keeping the journal files separated.
9901 * Many bugfixes and optimizations
9903 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
9904 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
9905 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
9910 * Several tools now understand kernel command line arguments,
9911 which are only read when run in an initial RAM disk. They
9912 usually follow closely their normal counterparts, but are
9915 * There's a new tool to analyze the readahead files that are
9916 automatically generated at boot. Use:
9918 /usr/lib/systemd/systemd-readahead analyze /.readahead
9920 * We now provide an early debug shell on tty9 if this enabled. Use:
9922 systemctl enable debug-shell.service
9924 * All plymouth related units have been moved into the Plymouth
9925 package. Please make sure to upgrade your Plymouth version
9928 * systemd-tmpfiles now supports getting passed the basename of
9929 a configuration file only, in which case it will look for it
9930 in all appropriate directories automatically.
9932 * udevadm info now takes a /dev or /sys path as argument, and
9933 does the right thing. Example:
9935 udevadm info /dev/sda
9936 udevadm info /sys/class/block/sda
9938 * systemctl now prints a warning if a unit is stopped but a
9939 unit that might trigger it continues to run. Example: a
9940 service is stopped but the socket that activates it is left
9943 * "systemctl status" will now mention if the log output was
9944 shortened due to rotation since a service has been started.
9946 * The journal API now exposes functions to determine the
9947 "cutoff" times due to rotation.
9949 * journald now understands SIGUSR1 and SIGUSR2 for triggering
9950 immediately flushing of runtime logs to /var if possible,
9951 resp. for triggering immediate rotation of the journal
9954 * It is now considered an error if a service is attempted to
9955 be stopped that is not loaded.
9957 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
9959 * systemd-analyze now supports Python 3
9961 * tmpfiles now supports cleaning up directories via aging
9962 where the first level dirs are always kept around but
9963 directories beneath it automatically aged. This is enabled
9964 by prefixing the age field with '~'.
9966 * Seat objects now expose CanGraphical, CanTTY properties
9967 which is required to deal with very fast bootups where the
9968 display manager might be running before the graphics drivers
9969 completed initialization.
9971 * Seat objects now expose a State property.
9973 * We now include RPM macros for service enabling/disabling
9974 based on the preset logic. We recommend RPM based
9975 distributions to make use of these macros if possible. This
9976 makes it simpler to reuse RPM spec files across
9979 * We now make sure that the collected systemd unit name is
9980 always valid when services log to the journal via
9983 * There's a new man page kernel-command-line(7) detailing all
9984 command line options we understand.
9986 * The fstab generator may now be disabled at boot by passing
9987 fstab=0 on the kernel command line.
9989 * A new kernel command line option modules-load= is now understood
9990 to load a specific kernel module statically, early at boot.
9992 * Unit names specified on the systemctl command line are now
9993 automatically escaped as needed. Also, if file system or
9994 device paths are specified they are automatically turned
9995 into the appropriate mount or device unit names. Example:
9997 systemctl status /home
9998 systemctl status /dev/sda
10000 * The SysVConsole= configuration option has been removed from
10001 system.conf parsing.
10003 * The SysV search path is no longer exported on the D-Bus
10006 * The Names= option has been removed from unit file parsing.
10008 * There's a new man page bootup(7) detailing the boot process.
10010 * Every unit and every generator we ship with systemd now
10011 comes with full documentation. The self-explanatory boot is
10014 * A couple of services gained "systemd-" prefixes in their
10015 name if they wrap systemd code, rather than only external
10016 code. Among them fsck@.service which is now
10017 systemd-fsck@.service.
10019 * The HaveWatchdog property has been removed from the D-Bus
10022 * systemd.confirm_spawn= on the kernel command line should now
10025 * There's a new man page crypttab(5) which details all options
10026 we actually understand.
10028 * systemd-nspawn gained a new --capability= switch to pass
10029 additional capabilities to the container.
10031 * timedated will now read known NTP implementation unit names
10032 from /usr/lib/systemd/ntp-units.d/*.list,
10033 systemd-timedated-ntp.target has been removed.
10035 * journalctl gained a new switch "-b" that lists log data of
10036 the current boot only.
10038 * The notify socket is in the abstract namespace again, in
10039 order to support daemons which chroot() at start-up.
10041 * There is a new Storage= configuration option for journald
10042 which allows configuration of where log data should go. This
10043 also provides a way to disable journal logging entirely, so
10044 that data collected is only forwarded to the console, the
10045 kernel log buffer or another syslog implementation.
10047 * Many bugfixes and optimizations
10049 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
10050 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
10051 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
10052 Shawn Landden, Tom Gundersen
10056 * "systemctl help <unit>" now shows the man page if one is
10059 * Several new man pages have been added.
10061 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
10062 MaxLevelConsole= can now be specified in
10063 journald.conf. These options allow reducing the amount of
10064 data stored on disk or forwarded by the log level.
10066 * TimerSlackNSec= can now be specified in system.conf for
10067 PID1. This allows system-wide power savings.
10069 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
10070 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
10075 * logind is now capable of (optionally) handling power and
10076 sleep keys as well as the lid switch.
10078 * journalctl now understands the syntax "journalctl
10079 /usr/bin/avahi-daemon" to get all log output of a specific
10082 * CapabilityBoundingSet= in system.conf now also influences
10083 the capability bound set of usermode helpers of the kernel.
10085 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
10086 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
10087 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
10088 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
10092 * Note that we skipped 139 releases here in order to set the
10093 new version to something that is greater than both udev's
10094 and systemd's most recent version number.
10096 * udev: all udev sources are merged into the systemd source tree now.
10097 All future udev development will happen in the systemd tree. It
10098 is still fully supported to use the udev daemon and tools without
10099 systemd running, like in initramfs or other init systems. Building
10100 udev though, will require the *build* of the systemd tree, but
10101 udev can be properly *run* without systemd.
10103 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
10104 should be used to create dead device nodes as workarounds for broken
10107 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
10108 no longer supported. udev_monitor_new_from_netlink() needs to be
10109 used to subscribe to events.
10111 * udev: when udevd is started by systemd, processes which are left
10112 behind by forking them off of udev rules, are unconditionally cleaned
10113 up and killed now after the event handling has finished. Services or
10114 daemons must be started as systemd services. Services can be
10115 pulled-in by udev to get started, but they can no longer be directly
10116 forked by udev rules.
10118 * udev: the daemon binary is called systemd-udevd now and installed
10119 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
10120 to adapt to that, create symlink, or rename the binary after building
10123 * libudev no longer provides these symbols:
10124 udev_monitor_from_socket()
10125 udev_queue_get_failed_list_entry()
10126 udev_get_{dev,sys,run}_path()
10127 The versions number was bumped and symbol versioning introduced.
10129 * systemd-loginctl and systemd-journalctl have been renamed
10130 to loginctl and journalctl to match systemctl.
10132 * The config files: /etc/systemd/systemd-logind.conf and
10133 /etc/systemd/systemd-journald.conf have been renamed to
10134 logind.conf and journald.conf. Package updates should rename
10135 the files to the new names on upgrade.
10137 * For almost all files the license is now LGPL2.1+, changed
10138 from the previous GPL2.0+. Exceptions are some minor stuff
10139 of udev (which will be changed to LGPL2.1 eventually, too),
10140 and the MIT licensed sd-daemon.[ch] library that is suitable
10141 to be used as drop-in files.
10143 * systemd and logind now handle system sleep states, in
10144 particular suspending and hibernating.
10146 * logind now implements a sleep/shutdown/idle inhibiting logic
10147 suitable for a variety of uses. Soonishly Lennart will blog
10148 about this in more detail.
10150 * var-run.mount and var-lock.mount are no longer provided
10151 (which previously bind mounted these directories to their new
10152 places). Distributions which have not converted these
10153 directories to symlinks should consider stealing these files
10154 from git history and add them downstream.
10156 * We introduced the Documentation= field for units and added
10157 this to all our shipped units. This is useful to make it
10158 easier to explore the boot and the purpose of the various
10161 * All smaller setup units (such as
10162 systemd-vconsole-setup.service) now detect properly if they
10163 are run in a container and are skipped when
10164 appropriate. This guarantees an entirely noise-free boot in
10165 Linux container environments such as systemd-nspawn.
10167 * A framework for implementing offline system updates is now
10168 integrated, for details see:
10169 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
10171 * A new service type Type=idle is available now which helps us
10172 avoiding ugly interleaving of getty output and boot status
10175 * There's now a system-wide CapabilityBoundingSet= option to
10176 globally reduce the set of capabilities for the
10177 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
10178 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
10179 even CAP_NET_ADMIN system-wide for secure systems.
10181 * There are now system-wide DefaultLimitXXX= options to
10182 globally change the defaults of the various resource limits
10183 for all units started by PID 1.
10185 * Harald Hoyer's systemd test suite has been integrated into
10186 systemd which allows easy testing of systemd builds in qemu
10187 and nspawn. (This is really awesome! Ask us for details!)
10189 * The fstab parser is now implemented as generator, not inside
10192 * systemctl will now warn you if .mount units generated from
10193 /etc/fstab are out of date due to changes in fstab that
10194 have not been read by systemd yet.
10196 * systemd is now suitable for usage in initrds. Dracut has
10197 already been updated to make use of this. With this in place
10198 initrds get a slight bit faster but primarily are much
10199 easier to introspect and debug since "systemctl status" in
10200 the host system can be used to introspect initrd services,
10201 and the journal from the initrd is kept around too.
10203 * systemd-delta has been added, a tool to explore differences
10204 between user/admin configuration and vendor defaults.
10206 * PrivateTmp= now affects both /tmp and /var/tmp.
10208 * Boot time status messages are now much prettier and feature
10209 proper english language. Booting up systemd has never been
10212 * Read-ahead pack files now include the inode number of all
10213 files to pre-cache. When the inode changes the pre-caching
10214 is not attempted. This should be nicer to deal with updated
10215 packages which might result in changes of read-ahead
10218 * We now temporaritly lower the kernel's read_ahead_kb variable
10219 when collecting read-ahead data to ensure the kernel's
10220 built-in read-ahead does not add noise to our measurements
10221 of necessary blocks to pre-cache.
10223 * There's now RequiresMountsFor= to add automatic dependencies
10224 for all mounts necessary for a specific file system path.
10226 * MountAuto= and SwapAuto= have been removed from
10227 system.conf. Mounting file systems at boot has to take place
10230 * nspawn now learned a new switch --uuid= to set the machine
10231 ID on the command line.
10233 * nspawn now learned the -b switch to automatically search
10234 for an init system.
10236 * vt102 is now the default TERM for serial TTYs, upgraded from
10239 * systemd-logind now works on VT-less systems.
10241 * The build tree has been reorganized. The individual
10242 components now have directories of their own.
10244 * A new condition type ConditionPathIsReadWrite= is now available.
10246 * nspawn learned the new -C switch to create cgroups for the
10247 container in other hierarchies.
10249 * We now have support for hardware watchdogs, configurable in
10252 * The scheduled shutdown logic now has a public API.
10254 * We now mount /tmp as tmpfs by default, but this can be
10255 masked and /etc/fstab can override it.
10257 * Since udisks does not make use of /media anymore we are not
10258 mounting a tmpfs on it anymore.
10260 * journalctl gained a new --local switch to only interleave
10261 locally generated journal files.
10263 * We can now load the IMA policy at boot automatically.
10265 * The GTK tools have been split off into a systemd-ui.
10267 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
10268 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
10269 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
10270 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
10271 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
10272 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
10273 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
10274 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
10275 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
10280 * This is mostly a bugfix release
10282 * Support optional initialization of the machine ID from the
10283 KVM or container configured UUID.
10285 * Support immediate reboots with "systemctl reboot -ff"
10287 * Show /etc/os-release data in systemd-analyze output
10289 * Many bugfixes for the journal, including endianness fixes and
10290 ensuring that disk space enforcement works
10292 * sd-login.h is C++ compatible again
10294 * Extend the /etc/os-release format on request of the Debian
10297 * We now refuse non-UTF8 strings used in various configuration
10298 and unit files. This is done to ensure we do not pass invalid
10299 data over D-Bus or expose it elsewhere.
10301 * Register Mimo USB Screens as suitable for automatic seat
10304 * Read SELinux client context from journal clients in a race
10307 * Reorder configuration file lookup order. /etc now always
10308 overrides /run in order to allow the administrator to always
10309 and unconditionally override vendor-supplied or
10310 automatically generated data.
10312 * The various user visible bits of the journal now have man
10313 pages. We still lack man pages for the journal API calls
10316 * We now ship all man pages in HTML format again in the
10319 Contributions from: Dave Reisner, Dirk Eibach, Frederic
10320 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
10321 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
10326 * This is mostly a bugfix release
10328 * systems lacking /etc/os-release are no longer supported.
10330 * Various functionality updates to libsystemd-login.so
10332 * Track class of PAM logins to distinguish greeters from
10333 normal user logins.
10335 Contributions from: Kay Sievers, Lennart Poettering, Michael
10340 * This is an important bugfix release for v41.
10342 * Building man pages is now optional which should be useful
10343 for those building systemd from git but unwilling to install
10346 * Watchdog support for supervising services is now usable. In
10347 a future release support for hardware watchdogs
10348 (i.e. /dev/watchdog) will be added building on this.
10350 * Service start rate limiting is now configurable and can be
10351 turned off per service. When a start rate limit is hit a
10352 reboot can automatically be triggered.
10354 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
10356 Contributions from: Benjamin Franzke, Bill Nottingham,
10357 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
10358 Schmidt, Michał Górny, Piotr Drąg
10362 * The systemd binary is installed /usr/lib/systemd/systemd now;
10363 An existing /sbin/init symlink needs to be adapted with the
10366 * The code that loads kernel modules has been ported to invoke
10367 libkmod directly, instead of modprobe. This means we do not
10368 support systems with module-init-tools anymore.
10370 * Watchdog support is now already useful, but still not
10373 * A new kernel command line option systemd.setenv= is
10374 understood to set system wide environment variables
10375 dynamically at boot.
10377 * We now limit the set of capabilities of systemd-journald.
10379 * We now set SIGPIPE to ignore by default, since it only is
10380 useful in shell pipelines, and has little use in general
10381 code. This can be disabled with IgnoreSIPIPE=no in unit
10384 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
10385 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
10390 * This is mostly a bugfix release
10392 * We now expose the reason why a service failed in the
10393 "Result" D-Bus property.
10395 * Rudimentary service watchdog support (will be completed over
10396 the next few releases.)
10398 * When systemd forks off in order execute some service we will
10399 now immediately changes its argv[0] to reflect which process
10400 it will execute. This is useful to minimize the time window
10401 with a generic argv[0], which makes bootcharts more useful
10403 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
10404 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
10405 Mike Kazantsev, Ray Strode
10409 * This is mostly a test release, but incorporates many
10412 * New systemd-cgtop tool to show control groups by their
10415 * Linking against libacl for ACLs is optional again. If
10416 disabled, support tracking device access for active logins
10417 goes becomes unavailable, and so does access to the user
10418 journals by the respective users.
10420 * If a group "adm" exists, journal files are automatically
10421 owned by them, thus allow members of this group full access
10422 to the system journal as well as all user journals.
10424 * The journal now stores the SELinux context of the logging
10425 client for all entries.
10427 * Add C++ inclusion guards to all public headers
10429 * New output mode "cat" in the journal to print only text
10430 messages, without any meta data like date or time.
10432 * Include tiny X server wrapper as a temporary stop-gap to
10433 teach XOrg udev display enumeration. This is used by display
10434 managers such as gdm, and will go away as soon as XOrg
10435 learned native udev hotplugging for display devices.
10437 * Add new systemd-cat tool for executing arbitrary programs
10438 with STDERR/STDOUT connected to the journal. Can also act as
10439 BSD logger replacement, and does so by default.
10441 * Optionally store all locally generated coredumps in the
10442 journal along with meta data.
10444 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
10445 writing short strings to files (for usage for /sys), and for
10446 creating symlinks, character and block device nodes.
10448 * New unit file option ControlGroupPersistent= to make cgroups
10449 persistent, following the mechanisms outlined in
10450 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
10452 * Support multiple local RTCs in a sane way
10454 * No longer monopolize IO when replaying readahead data on
10455 rotating disks, since we might starve non-file-system IO to
10456 death, since fanotify() will not see accesses done by blkid,
10459 * Do not show kernel threads in systemd-cgls anymore, unless
10460 requested with new -k switch.
10462 Contributions from: Dan Horák, Kay Sievers, Lennart
10463 Poettering, Michal Schmidt
10467 * This is mostly a test release, but incorporates many
10470 * The git repository moved to:
10471 git://anongit.freedesktop.org/systemd/systemd
10472 ssh://git.freedesktop.org/git/systemd/systemd
10474 * First release with the journal
10475 http://0pointer.de/blog/projects/the-journal.html
10477 * The journal replaces both systemd-kmsg-syslogd and
10478 systemd-stdout-bridge.
10480 * New sd_pid_get_unit() API call in libsystemd-logind
10482 * Many systemadm clean-ups
10484 * Introduce remote-fs-pre.target which is ordered before all
10485 remote mounts and may be used to start services before all
10488 * Added Mageia support
10490 * Add bash completion for systemd-loginctl
10492 * Actively monitor PID file creation for daemons which exit in
10493 the parent process before having finished writing the PID
10494 file in the daemon process. Daemons which do this need to be
10495 fixed (i.e. PID file creation must have finished before the
10496 parent exits), but we now react a bit more gracefully to them.
10498 * Add colourful boot output, mimicking the well-known output
10499 of existing distributions.
10501 * New option PassCredentials= for socket units, for
10502 compatibility with a recent kernel ABI breakage.
10504 * /etc/rc.local is now hooked in via a generator binary, and
10505 thus will no longer act as synchronization point during
10508 * systemctl list-unit-files now supports --root=.
10510 * systemd-tmpfiles now understands two new commands: z, Z for
10511 relabelling files according to the SELinux database. This is
10512 useful to apply SELinux labels to specific files in /sys,
10513 among other things.
10515 * Output of SysV services is now forwarded to both the console
10516 and the journal by default, not only just the console.
10518 * New man pages for all APIs from libsystemd-login.
10520 * The build tree got reorganized and the build system is a
10521 lot more modular allowing embedded setups to specifically
10522 select the components of systemd they are interested in.
10524 * Support for Linux systems lacking the kernel VT subsystem is
10527 * configure's --with-rootdir= got renamed to
10528 --with-rootprefix= to follow the naming used by udev and
10531 * Unless specified otherwise we will now install to /usr instead
10532 of /usr/local by default.
10534 * Processes with '@' in argv[0][0] are now excluded from the
10535 final shut-down killing spree, following the logic explained
10537 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
10539 * All processes remaining in a service cgroup when we enter
10540 the START or START_PRE states are now killed with
10541 SIGKILL. That means it is no longer possible to spawn
10542 background processes from ExecStart= lines (which was never
10543 supported anyway, and bad style).
10545 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
10546 reloading of units together.
10548 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
10549 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
10550 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
10551 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
10552 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek