1 systemd System and Service Manager
5 * A new tool "systemd-repart" has been added, that operates as an
6 idempotent declarative repartitioner for GPT partition tables.
7 Specifically, a set of partitions that must or may exist can be
8 configured via drop-in files, and during every boot the partition
9 table on disk is compared with these files, creating missing
10 partitions or growing existing ones based on configurable relative
11 and absolute size constraints. The tool is strictly incremental,
12 i.e. does not delete, shrink or move partitions, but only adds and
13 grows them. The primary use-case is OS images that ship in minimized
14 form, that on first boot are grown to the size of the underlying
15 block device or augmented with additional partitions. For example,
16 the root partition could be extended to cover the whole disk, or a
17 swap or /home partitions could be added on first boot. It can also be
18 used for systems that use an A/B update scheme but ship images with
19 just the A partition, with B added on first boot. The tool is
20 primarily intended to be run in the initrd, shortly before
21 transitioning into the host OS, but can also be run after the
22 transition took place. It automatically discovers the disk backing
23 the root file system, and should hence not require any additional
24 configuration besides the partition definition drop-ins. If no
25 configuration drop-ins are present, no action is taken.
27 * A new component "userdb" has been added, along with a small daemon
28 "systemd-userdb.service" and a client tool "userdbctl". The framework
29 allows defining rich user and group records in a JSON format,
30 extending on the classic "struct passwd" and "struct group"
31 structures. Various components in systemd have been updated to
32 process records in this format, including systemd-logind and
33 pam-systemd. The user records are intended to be extensible, and
34 allow setting various resource management, security and runtime
35 parameters that shall be applied to processes and sessions of the
36 user as they log in. This facility is intended to allow associating
37 such metadata directly with user/group records so that they can be
38 produced, extended and consumed in unified form. We hope that
39 eventually frameworks such as sssd will generate records this way, so
40 that for the first time resource management and various other
41 per-user settings can be configured in LDAP directories and then
42 provided to systemd (specifically to systemd-logind and pam-system)
43 to apply on login. For further details see:
45 https://systemd.io/USER_RECORD
46 https://systemd.io/GROUP_RECORD
47 https://systemd.io/USER_GROUP_API
49 * A small new service systemd-homed.service has been added, that may be
50 used to securely manage home directories with built-in encryption.
51 The complete user record data is unified with the home directory,
52 thus making home directories naturally migratable. Its primary
53 back-end is based on LUKS volumes, but fscrypt, plain directories,
54 and other storage schemes are also supported. This solves a couple of
55 problems we saw with traditional ways to manage home directories, in
56 particular when it comes to encryption. For further discussion of
57 this, see the video of Lennart's talk at AllSystemsGo! 2019:
59 https://media.ccc.de/v/ASG2019-164-reinventing-home-directories
61 For further details about the format and expectations on home
62 directories this new daemon makes, see:
64 https://systemd.io/HOME_DIRECTORY
66 * systemd-journald is now multi-instantiable. In addition to the main
67 instance systemd-journald.service there's now a template unit
68 systemd-journald@.service, with each instance defining a new named
69 log 'namespace' (whose name is specified via the instance part of the
70 unit name). A new unit file setting LogNamespace= has been added,
71 taking such a namespace name, that assigns services to the specified
72 log namespaces. As each log namespace is serviced by its own
73 independent journal daemon, this functionality may be used to improve
74 performance and increase isolation of applications, at the price of
75 losing global message ordering. Each instance of journald has a
76 separate set of configuration files, with possibly different disk
77 usage limitations and other settings.
79 journalctl now takes a new option --namespace= to show logs from a
80 specific log namespace. The sd-journal.h API gained
81 sd_journal_open_namespace() for opening the log stream of a specific
82 log namespace. systemd-journald also gained the ability to exit on
83 idle, which is useful in the context of log namespaces, as this means
84 log daemons for log namespaces can be activated automatically on
85 demand and will stop automatically when no longer used, minimizing
88 * When systemd-tmpfiles copies a file tree using the 'C' line type it
89 will now label every copied file according to the SELinux database.
91 * When systemd/PID 1 detects it is used in the initrd it will now boot
92 into initrd.target rather than default.target by default. This should
93 make it simpler to build initrds with systemd as for many cases the
94 only difference between a host OS image and an initrd image now is
95 the presence of the /etc/initrd-release file.
97 * A new kernel command line option systemd.cpu_affinity= is now
98 understood. It's equivalent to the CPUAffinity= option in
99 /etc/systemd/system.conf and allows setting the CPU mask for PID 1
100 itself and the default for all other processes.
102 * When systemd/PID 1 is reloaded (with systemctl daemon-reload or
103 equivalent), the SELinux database is now reloaded, ensuring that
104 sockets and other file system objects are generated taking the new
105 database into account.
107 * systemd/PID 1 accepts a new "systemd.show-status=error" setting, and
108 "quiet" has been changed to imply that instead of
109 "systemd.show-status=auto". In this mode, only messages about errors
110 and significant delays in boot are shown on the console.
112 * The sd-event.h API gained native support for the new Linux "pidfd"
113 concept. This permits watching processes using file descriptors
114 instead of PID numbers, which fixes a number of races and makes
115 process supervision more robust and efficient. All of systemd's
116 components will now use pidfds if the kernel supports it for process
117 watching, with the exception of PID 1 itself, unfortunately. We hope
118 to move PID 1 to exclusively using pidfds too eventually, but this
119 requires some more kernel work first. (Background: PID 1 watches
120 processes using waitid() with the P_ALL flag, and that does not play
121 together nicely with pidfds yet.)
123 * Closely related to this, the sd-event.h API gained two new calls
124 sd_event_source_send_child_signal() (for sending a signal to a
125 watched process) and sd_event_source_get_child_process_own() (for
126 marking a process so that it is killed automatically whenever the
127 event source watching it is freed).
129 * systemd-networkd gained support for configuring Token Bucket Filter
130 (TBF) parameters in its qdisc configuration support. Similarly,
131 support for Stochastic Fairness Queuing (SFQ), Controlled-Delay
132 Active Queue Management (CoDel), and Fair Queue (FQ) has been added.
134 * systemd-networkd gained support for Intermediate Functional Block
135 (IFB) network devices.
137 * systemd-networkd gained support for configuring multi-path IP routes,
138 using the new MultiPathRoute= setting in the [Route] section.
140 * systemd-networkd's DHCPv4 client has been updated to support a new
141 SendDecline= option. If enabled, duplicate address detection is done
142 after a DHCP offer is received from the server. If a conflict is
143 detected, the address is declined. The DHCPv4 client also gained
144 support for a new RouteMTUBytes= setting that allows to configure the
145 MTU size to be used for routes generated from DHCPv4 leases.
147 * The PrefixRoute= setting in systemd-networkd's [Address] section of
148 .network files has been deprecated, and replaced by AddPrefixRoute=,
149 with its sense inverted.
151 * The Gateway= setting of [Route] sections of .network files gained
152 support for a special new value "_dhcp". If set, the configured
153 static route uses the gateway host configured via DHCP.
155 * New User= and SuppressPrefixLength= settings have been implemented
156 for the [RoutingPolicyRule] section of .network files to configure
157 source routing based on UID ranges and prefix length, respectively.
159 * sd-bus gained a new API call sd_bus_message_sensitive() that marks a
160 D-Bus message object as "sensitive". Those objects are erased from
161 memory when they are freed. This concept is intended to be used for
162 messages that contain security sensitive data. A new flag
163 SD_BUS_VTABLE_SENSITIVE has been introduced as well to mark methods
164 in sd-bus vtables, causing any incoming and outgoing messages of
165 those methods to be implicitly marked as "sensitive".
167 * sd-bus gained a new API call sd_bus_message_dump() for dumping the
168 contents of a message (or parts thereof) to standard output for
171 * systemd-sysusers gained support for creating users with the primary
172 group named differently than the user.
174 * systemd-resolved's DNS-over-TLS support gained SNI validation.
176 * systemd-growfs (i.e. the x-systemd.growfs mount option in /etc/fstab)
177 gained support for growing XFS partitions. Previously it supported
178 only ext4 and btrfs partitions.
180 * The support for /etc/crypttab gained a new x-initrd.attach option. If
181 set, the specified encrypted volume is unlocked already in the
182 initrd. This concept corresponds to the x-initrd.mount option in
185 * systemd-cryptsetup gained native support for unlocking encrypted
186 volumes utilizing PKCS#11 smartcards, i.e. for example to bind
187 encryption of volumes to YubiKeys. This is exposed in the new
188 pkcs11-uri= option in /etc/crypttab.
190 * The /etc/fstab support in systemd now supports two new mount options
191 x-systemd.{required,wanted}-by=, for explicitly configuring the units
192 that the specified mount shall be pulled in by, in place of
193 the usual local-fs.target/remote-fs.target.
195 * The https://systemd.io/ web site has been relaunched, directly
196 populated with most of the documentation included in the systemd
197 repository. systemd also acquired a new logo, thanks to Tobias
200 * systemd-udevd gained support for managing "alternative" network
201 interface names, as supported by new Linux kernels. For the first
202 time this permits assigning multiple (and longer!) names to a network
203 interface. systemd-udevd will now by default assign the names
204 generated via all supported naming schemes to each interface. This
205 may be further tweaked with .link files and the AlternativeName= and
206 AlternativeNamesPolicy= settings. Other components of systemd have
207 been updated to support the new alternative names wherever
208 appropriate. For example, systemd-nspawn will now generate
209 alternative interface names for the host-facing side of container
210 veth links based on the full container name without truncation.
212 * systemd-nspawn interface naming logic has been updated in another way
213 too: if the main interface name (i.e. as opposed to new-style
214 "alternative" names) based on the container name is truncated, a
215 simple hashing scheme is used to give different interface names to
216 multiple containers whose names all begin with the same prefix. Since
217 this changes the primary interface names pointing to containers if
218 truncation happens, the old scheme may still be requested by
219 selecting an older naming scheme, via the net.naming-scheme= kernel
222 * PrivateUsers= in service files now works in services run by the
223 systemd --user per-user instance of the service manager.
225 * A new per-service sandboxing option ProtectClock= has been added that
226 locks down write access to the system clock. It takes away device
227 node access to /dev/rtc as well as the system calls that set the
228 system clock and the CAP_SYS_TIME and CAP_WAKE_ALARM capabilities.
229 Note that this option does not affect access to auxiliary services
230 that allow changing the clock, for example access to
233 * The systemd-id128 tool gained a new "show" verb for listing or
234 resolving a number of well-known UUIDs/128bit IDs, currently mostly
235 GPT partition table types.
237 * The Discoverable Partitions Specification has been updated to support
238 /var and /var/tmp partition discovery. Support for this has been
239 added to systemd-gpt-auto-generator. For details see:
241 https://systemd.io/DISCOVERABLE_PARTITIONS
243 * "systemctl list-unit-files" has been updated to show a new column
244 with the suggested enablement state based on the vendor preset files
245 for the respective units.
247 * "systemctl" gained a new option "--with-dependencies". If specified
248 commands such as "systemctl status" or "systemctl cat" will now show
249 all specified units along with all units they depend on.
251 * networkctl gained support for showing per-interface logs in its
254 * systemd-networkd-wait-online gained support for specifying the maximum
255 operational state to wait for, and to wait for interfaces to
258 * The [Match] section of .link and .network files now supports a new
259 option PermanentMACAddress= which may be used to check against the
260 permanent MAC address of a network device even if a randomized MAC
263 * The [TrafficControlQueueingDiscipline] section in .network files has
264 been renamed to [NetworkEmulator] with the "NetworkEmulator" prefix
265 dropped from the individual setting names.
267 * Any .link and .network files that have an empty [Match] section (this
268 also includes empty and commented-out files) will now be
269 rejected. systemd-udev and systemd-networkd started warning about
270 such files in version 243.
272 * systemd-logind will now validate access to the operation of changing
273 the virtual terminal via a PolicyKit action. By default, only users
274 with at least one session on a local VT are granted permission.
276 * When systemd sets up PAM sessions that invoked service processes
277 shall run in, the pam_setcred() API is now invoked, thus permitting
278 PAM modules to set additional credentials for the processes.
280 * portablectl attach/detach verbs now accept --now and --enable options
281 to combine attachment with enablement and invocation, or detachment
282 with stopping and disablement.
284 Contributions from: AJ Bagwell, Alin Popa, Andreas Rammhold, Anita
285 Zhang, Ansgar Burchardt, Antonio Russo, Arian van Putten, Ashley Davis,
286 Balint Reczey, Bart Willems, Bastien Nocera, Benjamin Dahlhoff, Charles
287 (Chas) Williams, cheese1, Chris Down, Chris Murphy, Christian Ehrhardt,
288 Christian Göttsche, cvoinf, Daan De Meyer, Daniele Medri, Daniel Rusek,
289 Daniel Shahaf, Dann Frazier, Dan Streetman, Dariusz Gadomski, David
290 Michael, Dimitri John Ledkov, Emmanuel Bourg, Evgeny Vereshchagin,
291 ezst036, Felipe Sateler, Filipe Brandenburger, Florian Klink, Franck
292 Bui, Fran Dieguez, Frantisek Sumsal, Greg "GothAck" Miell, Guilhem
293 Lettron, Guillaume Douézan-Grard, Hans de Goede, HATAYAMA Daisuke, Iain
294 Lane, James Buren, Jan Alexander Steffens (heftig), Jérémy Rosen, Jin
295 Park, Jun'ichi Nomura, Kai Krakow, Kevin Kuehler, Kevin P. Fleming,
296 Lennart Poettering, Leonid Bloch, Leonid Evdokimov, lothrond, Luca
297 Boccassi, Lukas K, Lynn Kirby, Mario Limonciello, Mark Deneen, Matthew
298 Leeds, Michael Biebl, Michal Koutný, Michal Sekletár, Mike Auty, Mike
299 Gilbert, mtron, nabijaczleweli, Naïm Favier, Nate Jones, Norbert Lange,
300 Oliver Giles, Paul Davey, Paul Menzel, Peter Hutterer, Piotr Drąg, Rafa
301 Couto, Raphael, rhn, Robert Scheck, Rocka, Romain Naour, Ryan Attard,
302 Sascha Dewald, Shengjing Zhu, Slava Kardakov, Spencer Michaels, Sylvain
303 Plantefeve, Stanislav Angelovič, Susant Sahani, Thomas Haller, Thomas
304 Schmitt, Timo Schlüßler, Timo Wilken, Tobias Bernard, Tobias Klauser,
305 Tobias Stoeckmann, Topi Miettinen, tsia, WataruMatsuoka, Wieland
306 Hoffmann, Wilhelm Schuster, Will Fleming, xduugu, Yong Cong Sin, Yuri
307 Chornoivan, Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek, Zeyu
314 * Support for the cpuset cgroups v2 controller has been added.
315 Processes may be restricted to specific CPUs using the new
316 AllowedCPUs= setting, and to specific memory NUMA nodes using the new
317 AllowedMemoryNodes= setting.
319 * The signal used in restart jobs (as opposed to e.g. stop jobs) may
320 now be configured using a new RestartKillSignal= setting. This
321 allows units which signals to request termination to implement
322 different behaviour when stopping in preparation for a restart.
324 * "systemctl clean" may now be used also for socket, mount, and swap
327 * systemd will also read configuration options from the EFI variable
328 SystemdOptions. This may be used to configure systemd behaviour when
329 modifying the kernel command line is inconvenient, but configuration
330 on disk is read too late, for example for the options related to
331 cgroup hierarchy setup. 'bootctl systemd-efi-options' may be used to
332 set the EFI variable.
334 * systemd will now disable printk ratelimits in early boot. This should
335 allow us to capture more logs from the early boot phase where normal
336 storage is not available and the kernel ring buffer is used for
337 logging. Configuration on the kernel command line has higher priority
338 and overrides the systemd setting.
340 systemd programs which log to /dev/kmsg directly use internal
341 ratelimits to prevent runaway logging. (Normally this is only used
342 during early boot, so in practice this change has very little
345 * Unit files now support top level dropin directories of the form
346 <unit_type>.d/ (e.g. service.d/) that may be used to add configuration
347 that affects all corresponding unit files.
349 * systemctl gained support for 'stop --job-mode=triggering' which will
350 stop the specified unit and any units which could trigger it.
352 * Unit status display now includes units triggering and triggered by
353 the unit being shown.
355 * The RuntimeMaxSec= setting is now supported by scopes, not just
356 .service units. This is particularly useful for PAM sessions which
357 create a scope unit for the user login. systemd.runtime_max_sec=
358 setting may used with the pam_systemd module to limit the duration
359 of the PAM session, for example for time-limited logins.
361 * A new @pkey system call group is now defined to make it easier to
362 whitelist memory protection syscalls for containers and services
363 which need to use them.
365 * systemd-udevd: removed the 30s timeout for killing stale workers on
366 exit. systemd-udevd now waits for workers to finish. The hard-coded
367 exit timeout of 30s was too short for some large installations, where
368 driver initialization could be prematurely interrupted during initrd
369 processing if the root file system had been mounted and init was
370 preparing to switch root. If udevd is run without systemd and workers
371 are hanging while udevd receives an exit signal, udevd will now exit
372 when udev.event_timeout is reached for the last hanging worker. With
373 systemd, the exit timeout can additionally be configured using
374 TimeoutStopSec= in systemd-udevd.service.
376 * udev now provides a program (fido_id) that identifies FIDO CTAP1
377 ("U2F")/CTAP2 security tokens based on the usage declared in their
378 report and descriptor and outputs suitable environment variables.
379 This replaces the externally maintained whitelists of all known
380 security tokens that were used previously.
382 * Automatically generated autosuspend udev rules for whitelisted
383 devices have been imported from the Chromium OS project. This should
384 improve power saving with many more devices.
386 * udev gained a new "CONST{key}=value" setting that allows matching
387 against system-wide constants without forking a helper binary.
388 Currently "arch" and "virt" keys are supported.
390 * udev now opens CDROMs in non-exclusive mode when querying their
391 capabilities. This should fix issues where other programs trying to
392 use the CDROM cannot gain access to it, but carries a risk of
393 interfering with programs writing to the disk, if they did not open
394 the device in exclusive mode as they should.
396 * systemd-networkd does not create a default route for IPv4 link local
397 addressing anymore. The creation of the route was unexpected and was
398 breaking routing in various cases, but people who rely on it being
399 created implicitly will need to adjust. Such a route may be requested
400 with DefaultRouteOnDevice=yes.
402 Similarly, systemd-networkd will not assign a link-local IPv6 address
403 when IPv6 link-local routing is not enabled.
405 * Receive and transmit buffers may now be configured on links with
406 the new RxBufferSize= and TxBufferSize= settings.
408 * systemd-networkd may now advertise additional IPv6 routes. A new
409 [IPv6RoutePrefix] section with Route= and LifetimeSec= options is
412 * systemd-networkd may now configure "next hop" routes using the
413 [NextHop] section and Gateway= and Id= settings.
415 * systemd-networkd will now retain DHCP config on restarts by default
416 (but this may be overridden using the KeepConfiguration= setting).
417 The default for SendRelease= has been changed to true.
419 * The DHCPv4 client now uses the OPTION_INFORMATION_REFRESH_TIME option
420 received from the server.
422 The client will use the received SIP server list if UseSIP=yes is
425 The client may be configured to request specific options from the
426 server using a new RequestOptions= setting.
428 The client may be configured to send arbitrary options to the server
429 using a new SendOption= setting.
431 A new IPServiceType= setting has been added to configure the "IP
432 service type" value used by the client.
434 * The DHCPv6 client learnt a new PrefixDelegationHint= option to
435 request prefix hints in the DHCPv6 solicitation.
437 * The DHCPv4 server may be configured to send arbitrary options using
438 a new SendOption= setting.
440 * The DHCPv4 server may now be configured to emit SIP server list using
441 the new EmitSIP= and SIP= settings.
443 * systemd-networkd and networkctl may now renew DHCP leases on demand.
444 networkctl has a new 'networkctl renew' verb.
446 * systemd-networkd may now reconfigure links on demand. networkctl
447 gained two new verbs: "reload" will reload the configuration, and
448 "reconfigure DEVICE…" will reconfigure one or more devices.
450 * .network files may now match on SSID and BSSID of a wireless network,
451 i.e. the access point name and hardware address using the new SSID=
452 and BSSID= options. networkctl will display the current SSID and
453 BSSID for wireless links.
455 .network files may also match on the wireless network type using the
456 new WLANInterfaceType= option.
458 * systemd-networkd now includes default configuration that enables
459 link-local addressing when connected to an ad-hoc wireless network.
461 * systemd-networkd may configure the Traffic Control queueing
462 disciplines in the kernel using the new
463 [TrafficControlQueueingDiscipline] section and Parent=,
464 NetworkEmulatorDelaySec=, NetworkEmulatorDelayJitterSec=,
465 NetworkEmulatorPacketLimit=, NetworkEmulatorLossRate=,
466 NetworkEmulatorDuplicateRate= settings.
468 * systemd-tmpfiles gained a new w+ setting to append to files.
470 * systemd-analyze dump will now report when the memory configuration in
471 the kernel does not match what systemd has configured (usually,
472 because some external program has modified the kernel configuration
475 * systemd-analyze gained a new --base-time= switch instructs the
476 'calendar' verb to resolve times relative to that timestamp instead
479 * journalctl --update-catalog now produces deterministic output (making
480 reproducible image builds easier).
482 * A new devicetree-overlay setting is now documented in the Boot Loader
485 * The default value of the WatchdogSec= setting used in systemd
486 services (the ones bundled with the project itself) may be set at
487 configuration time using the -Dservice-watchdog= setting. If set to
488 empty, the watchdogs will be disabled.
490 * systemd-resolved validates IP addresses in certificates now when GnuTLS
493 * libcryptsetup >= 2.0.1 is now required.
495 * A configuration option -Duser-path= may be used to override the $PATH
496 used by the user service manager. The default is again to use the same
497 path as the system manager.
499 * The systemd-id128 tool gained a new switch "-u" (or "--uuid") for
500 outputting the 128bit IDs in UUID format (i.e. in the "canonical
503 * Service units gained a new sandboxing option ProtectKernelLogs= which
504 makes sure the program cannot get direct access to the kernel log
505 buffer anymore, i.e. the syslog() system call (not to be confused
506 with the API of the same name in libc, which is not affected), the
507 /proc/kmsg and /dev/kmsg nodes and the CAP_SYSLOG capability are made
508 inaccessible to the service. It's recommended to enable this setting
509 for all services that should not be able to read from or write to the
510 kernel log buffer, which are probably almost all.
512 Contributions from: Aaron Plattner, Alcaro, Anita Zhang, Balint Reczey,
513 Bastien Nocera, Baybal Ni, Benjamin Bouvier, Benjamin Gilbert, Carlo
514 Teubner, cbzxt, Chen Qi, Chris Down, Christian Rebischke, Claudio
515 Zumbo, ClydeByrdIII, crashfistfight, Cyprien Laplace, Daniel Edgecumbe,
516 Daniel Gorbea, Daniel Rusek, Daniel Stuart, Dan Streetman, David
517 Pedersen, David Tardon, Dimitri John Ledkov, Dominique Martinet, Donald
518 A. Cupp Jr, Evgeny Vereshchagin, Fabian Henneke, Filipe Brandenburger,
519 Franck Bui, Frantisek Sumsal, Georg Müller, Hans de Goede, Haochen
520 Tong, HATAYAMA Daisuke, Iwan Timmer, Jan Janssen, Jan Kundrát, Jan
521 Synacek, Jan Tojnar, Jay Strict, Jérémy Rosen, Jóhann B. Guðmundsson,
522 Jonas Jelten, Jonas Thelemann, Justin Trudell, J. Xing, Kai-Heng Feng,
523 Kenneth D'souza, Kevin Becker, Kevin Kuehler, Lennart Poettering,
524 Léonard Gérard, Lorenz Bauer, Luca Boccassi, Maciej Stanczew, Mario
525 Limonciello, Marko Myllynen, Mark Stosberg, Martin Wilck, matthiasroos,
526 Michael Biebl, Michael Olbrich, Michael Tretter, Michal Sekletar,
527 Michal Sekletár, Michal Suchanek, Mike Gilbert, Mike Kazantsev, Nicolas
528 Douma, nikolas, Norbert Lange, pan93412, Pascal de Bruijn, Paul Menzel,
529 Pavel Hrdina, Peter Wu, Philip Withnall, Piotr Drąg, Rafael Fontenelle,
530 Renaud Métrich, Riccardo Schirone, RoadrunnerWMC, Ronan Pigott, Ryan
531 Attard, Sebastian Wick, Serge, Siddharth Chandrasekara, Steve Ramage,
532 Steve Traylen, Susant Sahani, Thibault Nélis, Tim Teichmann, Tom
533 Fitzhenry, Tommy J, Torsten Hilbrich, Vito Caputo, ypf791, Yu Watanabe,
534 Zach Smith, Zbigniew Jędrzejewski-Szmek
540 * This release enables unprivileged programs (i.e. requiring neither
541 setuid nor file capabilities) to send ICMP Echo (i.e. ping) requests
542 by turning on the "net.ipv4.ping_group_range" sysctl of the Linux
543 kernel for the whole UNIX group range, i.e. all processes. This
544 change should be reasonably safe, as the kernel support for it was
545 specifically implemented to allow safe access to ICMP Echo for
546 processes lacking any privileges. If this is not desirable, it can be
547 disabled again by setting the parameter to "1 0".
549 * Previously, filters defined with SystemCallFilter= would have the
550 effect that any calling of an offending system call would terminate
551 the calling thread. This behaviour never made much sense, since
552 killing individual threads of unsuspecting processes is likely to
553 create more problems than it solves. With this release the default
554 action changed from killing the thread to killing the whole
555 process. For this to work correctly both a kernel version (>= 4.14)
556 and a libseccomp version (>= 2.4.0) supporting this new seccomp
557 action is required. If an older kernel or libseccomp is used the old
558 behaviour continues to be used. This change does not affect any
559 services that have no system call filters defined, or that use
560 SystemCallErrorNumber= (and thus see EPERM or another error instead
561 of being killed when calling an offending system call). Note that
562 systemd documentation always claimed that the whole process is
563 killed. With this change behaviour is thus adjusted to match the
566 * On 64 bit systems, the "kernel.pid_max" sysctl is now bumped to
567 4194304 by default, i.e. the full 22bit range the kernel allows, up
568 from the old 16bit range. This should improve security and
569 robustness, as PID collisions are made less likely (though certainly
570 still possible). There are rumours this might create compatibility
571 problems, though at this moment no practical ones are known to
572 us. Downstream distributions are hence advised to undo this change in
573 their builds if they are concerned about maximum compatibility, but
574 for everybody else we recommend leaving the value bumped. Besides
575 improving security and robustness this should also simplify things as
576 the maximum number of allowed concurrent tasks was previously bounded
577 by both "kernel.pid_max" and "kernel.threads-max" and now effectively
578 only a single knob is left ("kernel.threads-max"). There have been
579 concerns that usability is affected by this change because larger PID
580 numbers are harder to type, but we believe the change from 5 digits
581 to 7 digits doesn't hamper usability.
583 * MemoryLow= and MemoryMin= gained hierarchy-aware counterparts,
584 DefaultMemoryLow= and DefaultMemoryMin=, which can be used to
585 hierarchically set default memory protection values for a particular
586 subtree of the unit hierarchy.
588 * Memory protection directives can now take a value of zero, allowing
589 explicit opting out of a default value propagated by an ancestor.
591 * systemd now defaults to the "unified" cgroup hierarchy setup during
592 build-time, i.e. -Ddefault-hierarchy=unified is now the build-time
593 default. Previously, -Ddefault-hierarchy=hybrid was the default. This
594 change reflects the fact that cgroupsv2 support has matured
595 substantially in both systemd and in the kernel, and is clearly the
596 way forward. Downstream production distributions might want to
597 continue to use -Ddefault-hierarchy=hybrid (or even =legacy) for
598 their builds as unfortunately the popular container managers have not
599 caught up with the kernel API changes.
601 * Man pages are not built by default anymore (html pages were already
602 disabled by default), to make development builds quicker. When
603 building systemd for a full installation with documentation, meson
604 should be called with -Dman=true and/or -Dhtml=true as appropriate.
605 The default was changed based on the assumption that quick one-off or
606 repeated development builds are much more common than full optimized
607 builds for installation, and people need to pass various other
608 options to when doing "proper" builds anyway, so the gain from making
609 development builds quicker is bigger than the one time disruption for
612 Two scripts are created in the *build* directory to generate and
613 preview man and html pages on demand, e.g.:
615 build/man/man systemctl
616 build/man/html systemd.index
618 * libidn2 is used by default if both libidn2 and libidn are installed.
619 Please use -Dlibidn=true if libidn is preferred.
621 * The D-Bus "wire format" of the CPUAffinity= attribute is changed on
622 big-endian machines. Before, bytes were written and read in native
623 machine order as exposed by the native libc __cpu_mask interface.
624 Now, little-endian order is always used (CPUs 0–7 are described by
625 bits 0–7 in byte 0, CPUs 8–15 are described by byte 1, and so on).
626 This change fixes D-Bus calls that cross endianness boundary.
628 The presentation format used for CPUAffinity= by "systemctl show" and
629 "systemd-analyze dump" is changed to present CPU indices instead of
630 the raw __cpu_mask bitmask. For example, CPUAffinity=0-1 would be
631 shown as CPUAffinity=03000000000000000000000000000… (on
632 little-endian) or CPUAffinity=00000000000000300000000000000… (on
633 64-bit big-endian), and is now shown as CPUAffinity=0-1, matching the
634 input format. The maximum integer that will be printed in the new
635 format is 8191 (four digits), while the old format always used a very
636 long number (with the length varying by architecture), so they can be
637 unambiguously distinguished.
639 * /usr/sbin/halt.local is no longer supported. Implementation in
640 distributions was inconsistent and it seems this functionality was
643 To replace this functionality, users should:
644 - either define a new unit and make it a dependency of final.target
645 (systemctl add-wants final.target my-halt-local.service)
646 - or move the shutdown script to /usr/lib/systemd/system-shutdown/
647 and ensure that it accepts "halt", "poweroff", "reboot", and
648 "kexec" as an argument, see the description in systemd-shutdown(8).
650 * When a [Match] section in .link or .network file is empty (contains
651 no match patterns), a warning will be emitted. Please add any "match
652 all" pattern instead, e.g. OriginalName=* or Name=* in case all
653 interfaces should really be matched.
655 * A new setting NUMAPolicy= may be used to set process memory
656 allocation policy. This setting can be specified in
657 /etc/systemd/system.conf and hence will set the default policy for
658 PID1. The default policy can be overridden on a per-service
659 basis. The related setting NUMAMask= is used to specify NUMA node
660 mask that should be associated with the selected policy.
662 * PID 1 will now listen to Out-Of-Memory (OOM) events the kernel
663 generates when processes it manages are reaching their memory limits,
664 and will place their units in a special state, and optionally kill or
667 * The service manager will now expose bus properties for the IO
668 resources used by units. This information is also shown in "systemctl
669 status" now (for services that have IOAccounting=yes set). Moreover,
670 the IO accounting data is included in the resource log message
671 generated whenever a unit stops.
673 * Units may now configure an explicit time-out to wait for when killed
674 with SIGABRT, for example when a service watchdog is hit. Previously,
675 the regular TimeoutStopSec= time-out was applied in this case too —
676 now a separate time-out may be set using TimeoutAbortSec=.
678 * Services may now send a special WATCHDOG=trigger message with
679 sd_notify() to trigger an immediate "watchdog missed" event, and thus
680 trigger service termination. This is useful both for testing watchdog
681 handling, but also for defining error paths in services, that shall
682 be handled the same way as watchdog events.
684 * There are two new per-unit settings IPIngressFilterPath= and
685 IPEgressFilterPath= which allow configuration of a BPF program
686 (usually by specifying a path to a program uploaded to /sys/fs/bpf/)
687 to apply to the IP packet ingress/egress path of all processes of a
688 unit. This is useful to allow running systemd services with BPF
689 programs set up externally.
691 * systemctl gained a new "clean" verb for removing the state, cache,
692 runtime or logs directories of a service while it is terminated. The
693 new verb may also be used to remove the state maintained on disk for
694 timer units that have Persistent= configured.
696 * During the last phase of shutdown systemd will now automatically
697 increase the log level configured in the "kernel.printk" sysctl so
698 that any relevant loggable events happening during late shutdown are
699 made visible. Previously, loggable events happening so late during
700 shutdown were generally lost if the "kernel.printk" sysctl was set to
701 high thresholds, as regular logging daemons are terminated at that
702 time and thus nothing is written to disk.
704 * If processes terminated during the last phase of shutdown do not exit
705 quickly systemd will now show their names after a short time, to make
706 debugging easier. After a longer time-out they are forcibly killed,
709 * journalctl (and the other tools that display logs) will now highlight
710 warnings in yellow (previously, both LOG_NOTICE and LOG_WARNING where
711 shown in bright bold, now only LOG_NOTICE is). Moreover, audit logs
712 are now shown in blue color, to separate them visually from regular
713 logs. References to configuration files are now turned into clickable
714 links on terminals that support that.
716 * systemd-journald will now stop logging to /var/log/journal during
717 shutdown when /var/ is on a separate mount, so that it can be
718 unmounted safely during shutdown.
720 * systemd-resolved gained support for a new 'strict' DNS-over-TLS mode.
722 * systemd-resolved "Cache=" configuration option in resolved.conf has
723 been extended to also accept the 'no-negative' value. Previously,
724 only a boolean option was allowed (yes/no), having yes as the
725 default. If this option is set to 'no-negative', negative answers are
726 not cached while the old cache heuristics are used positive answers.
727 The default remains unchanged.
729 * The predictable naming scheme for network devices now supports
730 generating predictable names for "netdevsim" devices.
732 Moreover, the "en" prefix was dropped from the ID_NET_NAME_ONBOARD
735 Those two changes form a new net.naming-policy-scheme= entry.
736 Distributions which want to preserve naming stability may want to set
737 the -Ddefault-net-naming-scheme= configuration option.
739 * systemd-networkd now supports MACsec, nlmon, IPVTAP and Xfrm
742 * systemd-networkd's bridge FDB support now allows configuration of a
743 destination address for each entry (Destination=), as well as the
744 VXLAN VNI (VNI=), as well as an option to declare what an entry is
745 associated with (AssociatedWith=).
747 * systemd-networkd's DHCPv4 support now understands a new MaxAttempts=
748 option for configuring the maximum number of DHCP lease requests. It
749 also learnt a new BlackList= option for blacklisting DHCP servers (a
750 similar setting has also been added to the IPv6 RA client), as well
751 as a SendRelease= option for configuring whether to send a DHCP
752 RELEASE message when terminating.
754 * systemd-networkd's DHCPv4 and DHCPv6 stacks can now be configured
755 separately in the [DHCPv4] and [DHCPv6] sections.
757 * systemd-networkd's DHCP support will now optionally create an
758 implicit host route to the DNS server specified in the DHCP lease, in
759 addition to the routes listed explicitly in the lease. This should
760 ensure that in multi-homed systems DNS traffic leaves the systems on
761 the interface that acquired the DNS server information even if other
762 routes such as default routes exist. This behaviour may be turned on
763 with the new RoutesToDNS= option.
765 * systemd-networkd's VXLAN support gained a new option
766 GenericProtocolExtension= for enabling VXLAN Generic Protocol
767 Extension support, as well as IPDoNotFragment= for setting the IP
768 "Don't fragment" bit on outgoing packets. A similar option has been
769 added to the GENEVE support.
771 * In systemd-networkd's [Route] section you may now configure
772 FastOpenNoCookie= for configuring per-route TCP fast-open support, as
773 well as TTLPropagate= for configuring Label Switched Path (LSP) TTL
774 propagation. The Type= setting now supports local, broadcast,
775 anycast, multicast, any, xresolve routes, too.
777 * systemd-networkd's [Network] section learnt a new option
778 DefaultRouteOnDevice= for automatically configuring a default route
779 onto the network device.
781 * systemd-networkd's bridging support gained two new options ProxyARP=
782 and ProxyARPWifi= for configuring proxy ARP behaviour as well as
783 MulticastRouter= for configuring multicast routing behaviour. A new
784 option MulticastIGMPVersion= may be used to change bridge's multicast
785 Internet Group Management Protocol (IGMP) version.
787 * systemd-networkd's FooOverUDP support gained the ability to configure
788 local and peer IP addresses via Local= and Peer=. A new option
789 PeerPort= may be used to configure the peer's IP port.
791 * systemd-networkd's TUN support gained a new setting VnetHeader= for
792 tweaking Generic Segment Offload support.
794 * networkctl gained a new "delete" command for removing virtual network
795 devices, as well as a new "--stats" switch for showing device
798 * networkd.conf gained a new setting SpeedMeter= and
799 SpeedMeterIntervalSec=, to measure bitrate of network interfaces. The
800 measured speed may be shown by 'networkctl status'.
802 * "networkctl status" now displays MTU and queue lengths, and more
803 detailed information about VXLAN and bridge devices.
805 * systemd-networkd's .network and .link files gained a new Property=
806 setting in the [Match] section, to match against devices with
807 specific udev properties.
809 * systemd-networkd's tunnel support gained a new option
810 AssignToLoopback= for selecting whether to use the loopback device
811 "lo" as underlying device.
813 * systemd-networkd's MACAddress= setting in the [Neighbor] section has
814 been renamed to LinkLayerAddress=, and it now allows configuration of
817 * systemd-networkd's handling of the kernel's disable_ipv6 sysctl is
818 simplified: systemd-networkd will disable the sysctl (enable IPv6) if
819 IPv6 configuration (static or DHCPv6) was found for a given
820 interface. It will not touch the sysctl otherwise.
822 * The order of entries is $PATH used by the user manager instance was
823 changed to put bin/ entries before the corresponding sbin/ entries.
824 It is recommended to not rely on this order, and only ever have one
825 binary with a given name in the system paths under /usr.
827 * A new tool systemd-network-generator has been added that may generate
828 .network, .netdev and .link files from IP configuration specified on
829 the kernel command line in the format used by Dracut.
831 * The CriticalConnection= setting in .network files is now deprecated,
832 and replaced by a new KeepConfiguration= setting which allows more
833 detailed configuration of the IP configuration to keep in place.
835 * systemd-analyze gained a few new verbs:
837 - "systemd-analyze timestamp" parses and converts timestamps. This is
838 similar to the existing "systemd-analyze calendar" command which
839 does the same for recurring calendar events.
841 - "systemd-analyze timespan" parses and converts timespans (i.e.
842 durations as opposed to points in time).
844 - "systemd-analyze condition" will parse and test ConditionXYZ=
847 - "systemd-analyze exit-status" will parse and convert exit status
848 codes to their names and back.
850 - "systemd-analyze unit-files" will print a list of all unit
851 file paths and unit aliases.
853 * SuccessExitStatus=, RestartPreventExitStatus=, and
854 RestartForceExitStatus= now accept exit status names (e.g. "DATAERR"
855 is equivalent to "65"). Those exit status name mappings may be
856 displayed with the systemd-analyze exit-status verb describe above.
858 * systemd-logind now exposes a per-session SetBrightness() bus call,
859 which may be used to securely change the brightness of a kernel
860 brightness device, if it belongs to the session's seat. By using this
861 call unprivileged clients can make changes to "backlight" and "leds"
862 devices securely with strict requirements on session membership.
863 Desktop environments may use this to generically make brightness
864 changes to such devices without shipping private SUID binaries or
865 udev rules for that purpose.
867 * "udevadm info" gained a --wait-for-initialization switch to wait for
868 a device to be initialized.
870 * systemd-hibernate-resume-generator will now look for resumeflags= on
871 the kernel command line, which is similar to rootflags= and may be
872 used to configure device timeout for the hibernation device.
874 * sd-event learnt a new API call sd_event_source_disable_unref() for
875 disabling and unref'ing an event source in a single function. A
876 related call sd_event_source_disable_unrefp() has been added for use
877 with gcc's cleanup extension.
879 * The sd-id128.h public API gained a new definition
880 SD_ID128_UUID_FORMAT_STR for formatting a 128bit ID in UUID format
883 * "busctl introspect" gained a new switch --xml-interface for dumping
884 XML introspection data unmodified.
886 * PID 1 may now show the unit name instead of the unit description
887 string in its status output during boot. This may be configured in
888 the StatusUnitFormat= setting in /etc/systemd/system.conf or the
889 kernel command line option systemd.status_unit_format=.
891 * PID 1 now understands a new option KExecWatchdogSec= in
892 /etc/systemd/system.conf to set a watchdog timeout for kexec reboots.
893 Previously watchdog functionality was only available for regular
894 reboots. The new setting defaults to off, because we don't know in
895 the general case if the watchdog will be reset after kexec (some
896 drivers do reset it, but not all), and the new userspace might not be
897 configured to handle the watchdog.
899 Moreover, the old ShutdownWatchdogSec= setting has been renamed to
900 RebootWatchdogSec= to more clearly communicate what it is about. The
901 old name is still accepted for compatibility.
903 * The systemd.debug_shell kernel command line option now optionally
904 takes a tty name to spawn the debug shell on, which allows a
905 different tty to be selected than the built-in default.
907 * Service units gained a new ExecCondition= setting which will run
908 before ExecStartPre= and either continue execution of the unit (for
909 clean exit codes), stop execution without marking the unit failed
910 (for exit codes 1 through 254), or stop execution and fail the unit
911 (for exit code 255 or abnormal termination).
913 * A new service systemd-pstore.service has been added that pulls data
914 from /sys/fs/pstore/ and saves it to /var/lib/pstore for later
917 * timedatectl gained new verbs for configuring per-interface NTP
918 service configuration for systemd-timesyncd.
920 * "localectl list-locales" won't list non-UTF-8 locales anymore. It's
921 2019. (You can set non-UTF-8 locales though, if you know their name.)
923 * If variable assignments in sysctl.d/ files are prefixed with "-" any
924 failures to apply them are now ignored.
926 * systemd-random-seed.service now optionally credits entropy when
927 applying the seed to the system. Set $SYSTEMD_RANDOM_SEED_CREDIT to
928 true for the service to enable this behaviour, but please consult the
929 documentation first, since this comes with a couple of caveats.
931 * systemd-random-seed.service is now a synchronization point for full
932 initialization of the kernel's entropy pool. Services that require
933 /dev/urandom to be correctly initialized should be ordered after this
936 * The systemd-boot boot loader has been updated to optionally maintain
937 a random seed file in the EFI System Partition (ESP). During the boot
938 phase, this random seed is read and updated with a new seed
939 cryptographically derived from it. Another derived seed is passed to
940 the OS. The latter seed is then credited to the kernel's entropy pool
941 very early during userspace initialization (from PID 1). This allows
942 systems to boot up with a fully initialized kernel entropy pool from
943 earliest boot on, and thus entirely removes all entropy pool
944 initialization delays from systems using systemd-boot. Special care
945 is taken to ensure different seeds are derived on system images
946 replicated to multiple systems. "bootctl status" will show whether
947 a seed was received from the boot loader.
949 * bootctl gained two new verbs:
951 - "bootctl random-seed" will generate the file in ESP and an EFI
952 variable to allow a random seed to be passed to the OS as described
955 - "bootctl is-installed" checks whether systemd-boot is currently
958 * bootctl will warn if it detects that boot entries are misconfigured
959 (for example if the kernel image was removed without purging the
962 * A new document has been added describing systemd's use and support
963 for the kernel's entropy pool subsystem:
965 https://systemd.io/RANDOM_SEEDS
967 * When the system is hibernated the swap device to write the
968 hibernation image to is now automatically picked from all available
969 swap devices, preferring the swap device with the highest configured
970 priority over all others, and picking the device with the most free
971 space if there are multiple devices with the highest priority.
973 * /etc/crypttab support has learnt a new keyfile-timeout= per-device
974 option that permits selecting the timout how long to wait for a
975 device with an encryption key before asking for the password.
977 * IOWeight= has learnt to properly set the IO weight when using the
978 BFQ scheduler officially found in kernels 5.0+.
980 * A new mailing list has been created for reporting of security issues:
981 systemd-security@redhat.com. For mode details, see
982 https://systemd.io/CONTRIBUTING#security-vulnerability-reports.
984 Contributions from: Aaron Barany, Adrian Bunk, Alan Jenkins, Albrecht
985 Lohofener, Andrej Valek, Anita Zhang, Arian van Putten, Balint Reczey,
986 Bastien Nocera, Ben Boeckel, Benjamin Robin, camoz, Chen Qi, Chris
987 Chiu, Chris Down, Christian Göttsche, Christian Kellner, Clinton Roy,
988 Connor Reeder, Daniel Black, Daniel Lublin, Daniele Medri, Dan
989 Streetman, Dave Reisner, Dave Ross, David Art, David Tardon, Debarshi
990 Ray, Dimitri John Ledkov, Dominick Grift, Donald Buczek, Douglas
991 Christman, Eric DeVolder, EtherGraf, Evgeny Vereshchagin, Feldwor,
992 Felix Riemann, Florian Dollinger, Francesco Pennica, Franck Bui,
993 Frantisek Sumsal, Franz Pletz, frederik, Hans de Goede, Iago López
994 Galeiras, Insun Pyo, Ivan Shapovalov, Iwan Timmer, Jack, Jakob
995 Unterwurzacher, Jan Chren, Jan Klötzke, Jan Losinski, Jan Pokorný, Jan
996 Synacek, Jan-Michael Brummer, Jeka Pats, Jeremy Soller, Jérémy Rosen,
997 Jiri Pirko, Joe Lin, Joerg Behrmann, Joe Richey, Jóhann B. Guðmundsson,
998 Johannes Christ, Johannes Schmitz, Jonathan Rouleau, Jorge Niedbalski,
999 Jörg Thalheim, Kai Krakow, Kai Lüke, Karel Zak, Kashyap Chamarthy,
1000 Krayushkin Konstantin, Lennart Poettering, Lubomir Rintel, Luca
1001 Boccassi, Luís Ferreira, Marc-André Lureau, Markus Felten, Martin Pitt,
1002 Matthew Leeds, Mattias Jernberg, Michael Biebl, Michael Olbrich,
1003 Michael Prokop, Michael Stapelberg, Michael Zhivich, Michal Koutný,
1004 Michal Sekletar, Mike Gilbert, Milan Broz, Miroslav Lichvar, mpe85,
1005 Mr-Foo, Network Silence, Oliver Harley, pan93412, Paul Menzel, pEJipE,
1006 Peter A. Bigot, Philip Withnall, Piotr Drąg, Rafael Fontenelle, Robert
1007 Scheck, Roberto Santalla, Ronan Pigott, root, RussianNeuroMancer,
1008 Sebastian Jennen, shinygold, Shreyas Behera, Simon Schricker, Susant
1009 Sahani, Thadeu Lima de Souza Cascardo, Theo Ouzhinski, Thiebaud
1010 Weksteen, Thomas Haller, Thomas Weißschuh, Tomas Mraz, Tommi Rantala,
1011 Topi Miettinen, VD-Lycos, ven, Vladimir Yerilov, Wieland Hoffmann,
1012 William A. Kennington III, William Wold, Xi Ruoyao, Yuri Chornoivan,
1013 Yu Watanabe, Zach Smith, Zbigniew Jędrzejewski-Szmek, Zhang Xianwei
1015 – Camerino, 2019-09-03
1019 * In .link files, MACAddressPolicy=persistent (the default) is changed
1020 to cover more devices. For devices like bridges, tun, tap, bond, and
1021 similar interfaces that do not have other identifying information,
1022 the interface name is used as the basis for persistent seed for MAC
1023 and IPv4LL addresses. The way that devices that were handled
1024 previously is not changed, and this change is about covering more
1025 devices then previously by the "persistent" policy.
1027 MACAddressPolicy=random may be used to force randomized MACs and
1028 IPv4LL addresses for a device if desired.
1030 Hint: the log output from udev (at debug level) was enhanced to
1031 clarify what policy is followed and which attributes are used.
1032 `SYSTEMD_LOG_LEVEL=debug udevadm test-builtin net_setup_link /sys/class/net/<name>`
1033 may be used to view this.
1035 Hint: if a bridge interface is created without any slaves, and gains
1036 a slave later, then now the bridge does not inherit slave's MAC.
1037 To inherit slave's MAC, for example, create the following file:
1039 # /etc/systemd/network/98-bridge-inherit-mac.link
1044 MACAddressPolicy=none
1047 * The .device units generated by systemd-fstab-generator and other
1048 generators do not automatically pull in the corresponding .mount unit
1049 as a Wants= dependency. This means that simply plugging in the device
1050 will not cause the mount unit to be started automatically. But please
1051 note that the mount unit may be started for other reasons, in
1052 particular if it is part of local-fs.target, and any unit which
1053 (transitively) depends on local-fs.target is started.
1055 * networkctl list/status/lldp now accept globbing wildcards for network
1056 interface names to match against all existing interfaces.
1058 * The $PIDFILE environment variable is set to point the absolute path
1059 configured with PIDFile= for processes of that service.
1061 * The fallback DNS server list was augmented with Cloudflare public DNS
1062 servers. Use `-Ddns-servers=` to set a different fallback.
1064 * A new special target usb-gadget.target will be started automatically
1065 when a USB Device Controller is detected (which means that the system
1066 is a USB peripheral).
1068 * A new unit setting CPUQuotaPeriodSec= assigns the time period
1069 relatively to which the CPU time quota specified by CPUQuota= is
1072 * A new unit setting ProtectHostname= may be used to prevent services
1073 from modifying hostname information (even if they otherwise would
1074 have privileges to do so).
1076 * A new unit setting NetworkNamespacePath= may be used to specify a
1077 namespace for service or socket units through a path referring to a
1078 Linux network namespace pseudo-file.
1080 * The PrivateNetwork= setting and JoinsNamespaceOf= dependencies now
1081 have an effect on .socket units: when used the listening socket is
1082 created within the configured network namespace instead of the host
1085 * ExecStart= command lines in unit files may now be prefixed with ':'
1086 in which case environment variable substitution is
1087 disabled. (Supported for the other ExecXYZ= settings, too.)
1089 * .timer units gained two new boolean settings OnClockChange= and
1090 OnTimezoneChange= which may be used to also trigger a unit when the
1091 system clock is changed or the local timezone is
1092 modified. systemd-run has been updated to make these options easily
1093 accessible from the command line for transient timers.
1095 * Two new conditions for units have been added: ConditionMemory= may be
1096 used to conditionalize a unit based on installed system
1097 RAM. ConditionCPUs= may be used to conditionalize a unit based on
1098 installed CPU cores.
1100 * The @default system call filter group understood by SystemCallFilter=
1101 has been updated to include the new rseq() system call introduced in
1104 * A new time-set.target has been added that indicates that the system
1105 time has been set from a local source (possibly imprecise). The
1106 existing time-sync.target is stronger and indicates that the time has
1107 been synchronized with a precise external source. Services where
1108 approximate time is sufficient should use the new target.
1110 * "systemctl start" (and related commands) learnt a new
1111 --show-transaction option. If specified brief information about all
1112 jobs queued because of the requested operation is shown.
1114 * systemd-networkd recognizes a new operation state 'enslaved', used
1115 (instead of 'degraded' or 'carrier') for interfaces which form a
1116 bridge, bond, or similar, and an new 'degraded-carrier' operational
1117 state used for the bond or bridge master interface when one of the
1118 enslaved devices is not operational.
1120 * .network files learnt the new IgnoreCarrierLoss= option for leaving
1121 networks configured even if the carrier is lost.
1123 * The RequiredForOnline= setting in .network files may now specify a
1124 minimum operational state required for the interface to be considered
1125 "online" by systemd-networkd-wait-online. Related to this
1126 systemd-networkd-wait-online gained a new option --operational-state=
1127 to configure the same, and its --interface= option was updated to
1128 optionally also take an operational state specific for an interface.
1130 * systemd-networkd-wait-online gained a new setting --any for waiting
1131 for only one of the requested interfaces instead of all of them.
1133 * systemd-networkd now implements L2TP tunnels.
1135 * Two new .network settings UseAutonomousPrefix= and UseOnLinkPrefix=
1136 may be used to cause autonomous and onlink prefixes received in IPv6
1137 Router Advertisements to be ignored.
1139 * New MulticastFlood=, NeighborSuppression=, and Learning= .network
1140 file settings may be used to tweak bridge behaviour.
1142 * The new TripleSampling= option in .network files may be used to
1143 configure CAN triple sampling.
1145 * A new .netdev settings PrivateKeyFile= and PresharedKeyFile= may be
1146 used to point to private or preshared key for a WireGuard interface.
1148 * /etc/crypttab now supports the same-cpu-crypt and
1149 submit-from-crypt-cpus options to tweak encryption work scheduling
1152 * systemd-tmpfiles will now take a BSD file lock before operating on a
1153 contents of directory. This may be used to temporarily exclude
1154 directories from aging by taking the same lock (useful for example
1155 when extracting a tarball into /tmp or /var/tmp as a privileged user,
1156 which might create files with really old timestamps, which
1157 nevertheless should not be deleted). For further details, see:
1159 https://systemd.io/TEMPORARY_DIRECTORIES
1161 * systemd-tmpfiles' h line type gained support for the
1162 FS_PROJINHERIT_FL ('P') file attribute (introduced in kernel 4.5),
1163 controlling project quota inheritance.
1165 * sd-boot and bootctl now implement support for an Extended Boot Loader
1166 (XBOOTLDR) partition, that is intended to be mounted to /boot, in
1167 addition to the ESP partition mounted to /efi or /boot/efi.
1168 Configuration file fragments, kernels, initrds and other EFI images
1169 to boot will be loaded from both the ESP and XBOOTLDR partitions.
1170 The XBOOTLDR partition was previously described by the Boot Loader
1171 Specification, but implementation was missing in sd-boot. Support for
1172 this concept allows using the sd-boot boot loader in more
1173 conservative scenarios where the boot loader itself is placed in the
1174 ESP but the kernels to boot (and their metadata) in a separate
1177 * A system may now be booted with systemd.volatile=overlay on the
1178 kernel command line, which causes the root file system to be set up
1179 an overlayfs mount combining the root-only root directory with a
1180 writable tmpfs. In this setup, the underlying root device is not
1181 modified, and any changes are lost at reboot.
1183 * Similar, systemd-nspawn can now boot containers with a volatile
1184 overlayfs root with the new --volatile=overlay switch.
1186 * systemd-nspawn can now consume OCI runtime bundles using a new
1187 --oci-bundle= option. This implementation is fully usable, with most
1188 features in the specification implemented, but since this a lot of
1189 new code and functionality, this feature should most likely not
1190 be used in production yet.
1192 * systemd-nspawn now supports various options described by the OCI
1193 runtime specification on the command-line and in .nspawn files:
1194 --inaccessible=/Inaccessible= may be used to mask parts of the file
1195 system tree, --console=/--pipe may be used to configure how standard
1196 input, output, and error are set up.
1198 * busctl learned the `emit` verb to generate D-Bus signals.
1200 * systemd-analyze cat-config may be used to gather and display
1201 configuration spread over multiple files, for example system and user
1202 presets, tmpfiles.d, sysusers.d, udev rules, etc.
1204 * systemd-analyze calendar now takes an optional new parameter
1205 --iterations= which may be used to show a maximum number of iterations
1206 the specified expression will elapse next.
1208 * The sd-bus C API gained support for naming method parameters in the
1211 * systemd-logind gained D-Bus APIs to specify the "reboot parameter"
1212 the reboot() system call expects.
1214 * journalctl learnt a new --cursor-file= option that points to a file
1215 from which a cursor should be loaded in the beginning and to which
1216 the updated cursor should be stored at the end.
1218 * ACRN hypervisor and Windows Subsystem for Linux (WSL) are now
1219 detected by systemd-detect-virt (and may also be used in
1220 ConditionVirtualization=).
1222 * The behaviour of systemd-logind may now be modified with environment
1223 variables $SYSTEMD_REBOOT_TO_FIRMWARE_SETUP,
1224 $SYSTEMD_REBOOT_TO_BOOT_LOADER_MENU, and
1225 $SYSTEMD_REBOOT_TO_BOOT_LOADER_ENTRY. They cause logind to either
1226 skip the relevant operation completely (when set to false), or to
1227 create a flag file in /run/systemd (when set to true), instead of
1228 actually commencing the real operation when requested. The presence
1229 of /run/systemd/reboot-to-firmware-setup,
1230 /run/systemd/reboot-to-boot-loader-menu, and
1231 /run/systemd/reboot-to-boot-loader-entry, may be used by alternative
1232 boot loader implementations to replace some steps logind performs
1233 during reboot with their own operations.
1235 * systemctl can be used to request a reboot into the boot loader menu
1236 or a specific boot loader entry with the new --boot-load-menu= and
1237 --boot-loader-entry= options to a reboot command. (This requires a
1238 boot loader that supports this, for example sd-boot.)
1240 * kernel-install will no longer unconditionally create the output
1241 directory (e.g. /efi/<machine-id>/<kernel-version>) for boot loader
1242 snippets, but will do only if the machine-specific parent directory
1243 (i.e. /efi/<machine-id>/) already exists. bootctl has been modified
1244 to create this parent directory during sd-boot installation.
1246 This makes it easier to use kernel-install with plugins which support
1247 a different layout of the bootloader partitions (for example grub2).
1249 * During package installation (with `ninja install`), we would create
1250 symlinks for getty@tty1.service, systemd-networkd.service,
1251 systemd-networkd.socket, systemd-resolved.service,
1252 remote-cryptsetup.target, remote-fs.target,
1253 systemd-networkd-wait-online.service, and systemd-timesyncd.service
1254 in /etc, as if `systemctl enable` was called for those units, to make
1255 the system usable immediately after installation. Now this is not
1256 done anymore, and instead calling `systemctl preset-all` is
1257 recommended after the first installation of systemd.
1259 * A new boolean sandboxing option RestrictSUIDSGID= has been added that
1260 is built on seccomp. When turned on creation of SUID/SGID files is
1263 * The NoNewPrivileges= and the new RestrictSUIDSGID= options are now
1264 implied if DynamicUser= is turned on for a service. This hardens
1265 these services, so that they neither can benefit from nor create
1266 SUID/SGID executables. This is a minor compatibility breakage, given
1267 that when DynamicUser= was first introduced SUID/SGID behaviour was
1268 unaffected. However, the security benefit of these two options is
1269 substantial, and the setting is still relatively new, hence we opted
1270 to make it mandatory for services with dynamic users.
1272 Contributions from: Adam Jackson, Alexander Tsoy, Andrey Yashkin,
1273 Andrzej Pietrasiewicz, Anita Zhang, Balint Reczey, Beniamino Galvani,
1274 Ben Iofel, Benjamin Berg, Benjamin Dahlhoff, Chris, Chris Morin,
1275 Christopher Wong, Claudius Ellsel, Clemens Gruber, dana, Daniel Black,
1276 Davide Cavalca, David Michael, David Rheinsberg, emersion, Evgeny
1277 Vereshchagin, Filipe Brandenburger, Franck Bui, Frantisek Sumsal,
1278 Giacinto Cifelli, Hans de Goede, Hugo Kindel, Ignat Korchagin, Insun
1279 Pyo, Jan Engelhardt, Jonas Dorel, Jonathan Lebon, Jonathon Kowalski,
1280 Jörg Sommer, Jörg Thalheim, Jussi Pakkanen, Kai-Heng Feng, Lennart
1281 Poettering, Lubomir Rintel, Luís Ferreira, Martin Pitt, Matthias
1282 Klumpp, Michael Biebl, Michael Niewöhner, Michael Olbrich, Michal
1283 Sekletar, Mike Lothian, Paul Menzel, Piotr Drąg, Riccardo Schirone,
1284 Robin Elvedi, Roman Kulikov, Ronald Tschalär, Ross Burton, Ryan
1285 Gonzalez, Sebastian Krzyszkowiak, Stephane Chazelas, StKob, Susant
1286 Sahani, Sylvain Plantefève, Szabolcs Fruhwald, Taro Yamada, Theo
1287 Ouzhinski, Thomas Haller, Tobias Jungel, Tom Yan, Tony Asleson, Topi
1288 Miettinen, unixsysadmin, Van Laser, Vesa Jääskeläinen, Yu, Li-Yu,
1289 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
1291 — Warsaw, 2019-04-11
1295 * The default locale can now be configured at compile time. Otherwise,
1296 a suitable default will be selected automatically (one of C.UTF-8,
1297 en_US.UTF-8, and C).
1299 * The version string shown by systemd and other tools now includes the
1300 git commit hash when built from git. An override may be specified
1301 during compilation, which is intended to be used by distributions to
1302 include the package release information.
1304 * systemd-cat can now filter standard input and standard error streams
1305 for different syslog priorities using the new --stderr-priority=
1308 * systemd-journald and systemd-journal-remote reject entries which
1309 contain too many fields (CVE-2018-16865) and set limits on the
1310 process' command line length (CVE-2018-16864).
1312 * $DBUS_SESSION_BUS_ADDRESS environment variable is set by pam_systemd
1315 * A new network device NamePolicy "keep" is implemented for link files,
1316 and used by default in 99-default.link (the fallback configuration
1317 provided by systemd). With this policy, if the network device name
1318 was already set by userspace, the device will not be renamed again.
1319 This matches the naming scheme that was implemented before
1320 systemd-240. If naming-scheme < 240 is specified, the "keep" policy
1321 is also enabled by default, even if not specified. Effectively, this
1322 means that if naming-scheme >= 240 is specified, network devices will
1323 be renamed according to the configuration, even if they have been
1324 renamed already, if "keep" is not specified as the naming policy in
1325 the .link file. The 99-default.link file provided by systemd includes
1326 "keep" for backwards compatibility, but it is recommended for user
1327 installed .link files to *not* include it.
1329 The "kernel" policy, which keeps kernel names declared to be
1330 "persistent", now works again as documented.
1332 * kernel-install script now optionally takes the paths to one or more
1333 initrd files, and passes them to all plugins.
1335 * The mincore() system call has been dropped from the @system-service
1336 system call filter group, as it is pretty exotic and may potentially
1337 used for side-channel attacks.
1339 * -fPIE is dropped from compiler and linker options. Please specify
1340 -Db_pie=true option to meson to build position-independent
1341 executables. Note that the meson option is supported since meson-0.49.
1343 * The fs.protected_regular and fs.protected_fifos sysctls, which were
1344 added in Linux 4.19 to make some data spoofing attacks harder, are
1345 now enabled by default. While this will hopefully improve the
1346 security of most installations, it is technically a backwards
1347 incompatible change; to disable these sysctls again, place the
1348 following lines in /etc/sysctl.d/60-protected.conf or a similar file:
1350 fs.protected_regular = 0
1351 fs.protected_fifos = 0
1353 Note that the similar hardlink and symlink protection has been
1354 enabled since v199, and may be disabled likewise.
1356 * The files read from the EnvironmentFile= setting in unit files now
1357 parse backslashes inside quotes literally, matching the behaviour of
1360 * udevadm trigger, udevadm control, udevadm settle and udevadm monitor
1361 now automatically become NOPs when run in a chroot() environment.
1363 * The tmpfiles.d/ "C" line type will now copy directory trees not only
1364 when the destination is so far missing, but also if it already exists
1365 as a directory and is empty. This is useful to cater for systems
1366 where directory trees are put together from multiple separate mount
1367 points but otherwise empty.
1369 * A new function sd_bus_close_unref() (and the associated
1370 sd_bus_close_unrefp()) has been added to libsystemd, that combines
1371 sd_bus_close() and sd_bus_unref() in one.
1373 * udevadm control learnt a new option for --ping for testing whether a
1374 systemd-udevd instance is running and reacting.
1376 * udevadm trigger learnt a new option for --wait-daemon for waiting
1377 systemd-udevd daemon to be initialized.
1379 Contributions from: Aaron Plattner, Alberts Muktupāvels, Alex Mayer,
1380 Ayman Bagabas, Beniamino Galvani, Burt P, Chris Down, Chris Lamb, Chris
1381 Morin, Christian Hesse, Claudius Ellsel, dana, Daniel Axtens, Daniele
1382 Medri, Dave Reisner, David Santamaría Rogado, Diego Canuhe, Dimitri
1383 John Ledkov, Evgeny Vereshchagin, Fabrice Fontaine, Filipe
1384 Brandenburger, Franck Bui, Frantisek Sumsal, govwin, Hans de Goede,
1385 James Hilliard, Jan Engelhardt, Jani Uusitalo, Jan Janssen, Jan
1386 Synacek, Jonathan McDowell, Jonathan Roemer, Jonathon Kowalski, Joost
1387 Heitbrink, Jörg Thalheim, Lance, Lennart Poettering, Louis Taylor,
1388 Lucas Werkmeister, Mantas Mikulėnas, Marc-Antoine Perennou,
1389 marvelousblack, Michael Biebl, Michael Sloan, Michal Sekletar, Mike
1390 Auty, Mike Gilbert, Mikhail Kasimov, Neil Brown, Niklas Hambüchen,
1391 Patrick Williams, Paul Seyfert, Peter Hutterer, Philip Withnall, Roger
1392 James, Ronnie P. Thomas, Ryan Gonzalez, Sam Morris, Stephan Edel,
1393 Stephan Gerhold, Susant Sahani, Taro Yamada, Thomas Haller, Topi
1394 Miettinen, YiFei Zhu, YmrDtnJu, YunQiang Su, Yu Watanabe, Zbigniew
1395 Jędrzejewski-Szmek, zsergeant77, Дамјан Георгиевски
1397 — Berlin, 2019-02-14
1401 * NoNewPrivileges=yes has been set for all long-running services
1402 implemented by systemd. Previously, this was problematic due to
1403 SELinux (as this would also prohibit the transition from PID1's label
1404 to the service's label). This restriction has since been lifted, but
1405 an SELinux policy update is required.
1406 (See e.g. https://github.com/fedora-selinux/selinux-policy/pull/234.)
1408 * DynamicUser=yes is dropped from systemd-networkd.service,
1409 systemd-resolved.service and systemd-timesyncd.service, which was
1410 enabled in v239 for systemd-networkd.service and systemd-resolved.service,
1411 and since v236 for systemd-timesyncd.service. The users and groups
1412 systemd-network, systemd-resolve and systemd-timesync are created
1413 by systemd-sysusers again. Distributors or system administrators
1414 may need to create these users and groups if they not exist (or need
1415 to re-enable DynamicUser= for those units) while upgrading systemd.
1416 Also, the clock file for systemd-timesyncd may need to move from
1417 /var/lib/private/systemd/timesync/clock to /var/lib/systemd/timesync/clock.
1419 * When unit files are loaded from disk, previously systemd would
1420 sometimes (depending on the unit loading order) load units from the
1421 target path of symlinks in .wants/ or .requires/ directories of other
1422 units. This meant that unit could be loaded from different paths
1423 depending on whether the unit was requested explicitly or as a
1424 dependency of another unit, not honouring the priority of directories
1425 in search path. It also meant that it was possible to successfully
1426 load and start units which are not found in the unit search path, as
1427 long as they were requested as a dependency and linked to from
1428 .wants/ or .requires/. The target paths of those symlinks are not
1429 used for loading units anymore and the unit file must be found in
1432 * A new service type has been added: Type=exec. It's very similar to
1433 Type=simple but ensures the service manager will wait for both fork()
1434 and execve() of the main service binary to complete before proceeding
1435 with follow-up units. This is primarily useful so that the manager
1436 propagates any errors in the preparation phase of service execution
1437 back to the job that requested the unit to be started. For example,
1438 consider a service that has ExecStart= set to a file system binary
1439 that doesn't exist. With Type=simple starting the unit would be
1440 considered instantly successful, as only fork() has to complete
1441 successfully and the manager does not wait for execve(), and hence
1442 its failure is seen "too late". With the new Type=exec service type
1443 starting the unit will fail, as the manager will wait for the
1444 execve() and notice its failure, which is then propagated back to the
1447 NOTE: with the next release 241 of systemd we intend to change the
1448 systemd-run tool to default to Type=exec for transient services
1449 started by it. This should be mostly safe, but in specific corner
1450 cases might result in problems, as the systemd-run tool will then
1451 block on NSS calls (such as user name look-ups due to User=) done
1452 between the fork() and execve(), which under specific circumstances
1453 might cause problems. It is recommended to specify "-p Type=simple"
1454 explicitly in the few cases where this applies. For regular,
1455 non-transient services (i.e. those defined with unit files on disk)
1456 we will continue to default to Type=simple.
1458 * The Linux kernel's current default RLIMIT_NOFILE resource limit for
1459 userspace processes is set to 1024 (soft) and 4096
1460 (hard). Previously, systemd passed this on unmodified to all
1461 processes it forked off. With this systemd release the hard limit
1462 systemd passes on is increased to 512K, overriding the kernel's
1463 defaults and substantially increasing the number of simultaneous file
1464 descriptors unprivileged userspace processes can allocate. Note that
1465 the soft limit remains at 1024 for compatibility reasons: the
1466 traditional UNIX select() call cannot deal with file descriptors >=
1467 1024 and increasing the soft limit globally might thus result in
1468 programs unexpectedly allocating a high file descriptor and thus
1469 failing abnormally when attempting to use it with select() (of
1470 course, programs shouldn't use select() anymore, and prefer
1471 poll()/epoll, but the call unfortunately remains undeservedly popular
1472 at this time). This change reflects the fact that file descriptor
1473 handling in the Linux kernel has been optimized in more recent
1474 kernels and allocating large numbers of them should be much cheaper
1475 both in memory and in performance than it used to be. Programs that
1476 want to take benefit of the increased limit have to "opt-in" into
1477 high file descriptors explicitly by raising their soft limit. Of
1478 course, when they do that they must acknowledge that they cannot use
1479 select() anymore (and neither can any shared library they use — or
1480 any shared library used by any shared library they use and so on).
1481 Which default hard limit is most appropriate is of course hard to
1482 decide. However, given reports that ~300K file descriptors are used
1483 in real-life applications we believe 512K is sufficiently high as new
1484 default for now. Note that there are also reports that using very
1485 high hard limits (e.g. 1G) is problematic: some software allocates
1486 large arrays with one element for each potential file descriptor
1487 (Java, …) — a high hard limit thus triggers excessively large memory
1488 allocations in these applications. Hopefully, the new default of 512K
1489 is a good middle ground: higher than what real-life applications
1490 currently need, and low enough for avoid triggering excessively large
1491 allocations in problematic software. (And yes, somebody should fix
1494 * The fs.nr_open and fs.file-max sysctls are now automatically bumped
1495 to the highest possible values, as separate accounting of file
1496 descriptors is no longer necessary, as memcg tracks them correctly as
1497 part of the memory accounting anyway. Thus, from the four limits on
1498 file descriptors currently enforced (fs.file-max, fs.nr_open,
1499 RLIMIT_NOFILE hard, RLIMIT_NOFILE soft) we turn off the first two,
1500 and keep only the latter two. A set of build-time options
1501 (-Dbump-proc-sys-fs-file-max=false and -Dbump-proc-sys-fs-nr-open=false)
1502 has been added to revert this change in behaviour, which might be
1503 an option for systems that turn off memcg in the kernel.
1505 * When no /etc/locale.conf file exists (and hence no locale settings
1506 are in place), systemd will now use the "C.UTF-8" locale by default,
1507 and set LANG= to it. This locale is supported by various
1508 distributions including Fedora, with clear indications that upstream
1509 glibc is going to make it available too. This locale enables UTF-8
1510 mode by default, which appears appropriate for 2018.
1512 * The "net.ipv4.conf.all.rp_filter" sysctl will now be set to 2 by
1513 default. This effectively switches the RFC3704 Reverse Path filtering
1514 from Strict mode to Loose mode. This is more appropriate for hosts
1515 that have multiple links with routes to the same networks (e.g.
1516 a client with a Wi-Fi and Ethernet both connected to the internet).
1518 Consult the kernel documentation for details on this sysctl:
1519 https://www.kernel.org/doc/Documentation/networking/ip-sysctl.txt
1521 * CPUAccounting=yes no longer enables the CPU controller when using
1522 kernel 4.15+ and the unified cgroup hierarchy, as required accounting
1523 statistics are now provided independently from the CPU controller.
1525 * Support for disabling a particular cgroup controller within a sub-tree
1526 has been added through the DisableControllers= directive.
1528 * cgroup_no_v1=all on the kernel command line now also implies
1529 using the unified cgroup hierarchy, unless one explicitly passes
1530 systemd.unified_cgroup_hierarchy=0 on the kernel command line.
1532 * The new "MemoryMin=" unit file property may now be used to set the
1533 memory usage protection limit of processes invoked by the unit. This
1534 controls the cgroup v2 memory.min attribute. Similarly, the new
1535 "IODeviceLatencyTargetSec=" property has been added, wrapping the new
1536 cgroup v2 io.latency cgroup property for configuring per-service I/O
1539 * systemd now supports the cgroup v2 devices BPF logic, as counterpart
1540 to the cgroup v1 "devices" cgroup controller.
1542 * systemd-escape now is able to combine --unescape with --template. It
1543 also learnt a new option --instance for extracting and unescaping the
1544 instance part of a unit name.
1546 * sd-bus now provides the sd_bus_message_readv() which is similar to
1547 sd_bus_message_read() but takes a va_list object. The pair
1548 sd_bus_set_method_call_timeout() and sd_bus_get_method_call_timeout()
1549 has been added for configuring the default method call timeout to
1550 use. sd_bus_error_move() may be used to efficiently move the contents
1551 from one sd_bus_error structure to another, invalidating the
1552 source. sd_bus_set_close_on_exit() and sd_bus_get_close_on_exit() may
1553 be used to control whether a bus connection object is automatically
1554 flushed when an sd-event loop is exited.
1556 * When processing classic BSD syslog log messages, journald will now
1557 save the original time-stamp string supplied in the new
1558 SYSLOG_TIMESTAMP= journal field. This permits consumers to
1559 reconstruct the original BSD syslog message more correctly.
1561 * StandardOutput=/StandardError= in service files gained support for
1562 new "append:…" parameters, for connecting STDOUT/STDERR of a service
1563 to a file, and appending to it.
1565 * The signal to use as last step of killing of unit processes is now
1566 configurable. Previously it was hard-coded to SIGKILL, which may now
1567 be overridden with the new KillSignal= setting. Note that this is the
1568 signal used when regular termination (i.e. SIGTERM) does not suffice.
1569 Similarly, the signal used when aborting a program in case of a
1570 watchdog timeout may now be configured too (WatchdogSignal=).
1572 * The XDG_SESSION_DESKTOP environment variable may now be configured in
1573 the pam_systemd argument line, using the new desktop= switch. This is
1574 useful to initialize it properly from a display manager without
1575 having to touch C code.
1577 * Most configuration options that previously accepted percentage values
1578 now also accept permille values with the '‰' suffix (instead of '%').
1580 * systemd-resolved may now optionally use OpenSSL instead of GnuTLS for
1583 * systemd-resolved's configuration file resolved.conf gained a new
1584 option ReadEtcHosts= which may be used to turn off processing and
1585 honoring /etc/hosts entries.
1587 * The "--wait" switch may now be passed to "systemctl
1588 is-system-running", in which case the tool will synchronously wait
1589 until the system finished start-up.
1591 * hostnamed gained a new bus call to determine the DMI product UUID.
1593 * On x86-64 systemd will now prefer using the RDRAND processor
1594 instruction over /dev/urandom whenever it requires randomness that
1595 neither has to be crypto-grade nor should be reproducible. This
1596 should substantially reduce the amount of entropy systemd requests
1597 from the kernel during initialization on such systems, though not
1598 reduce it to zero. (Why not zero? systemd still needs to allocate
1599 UUIDs and such uniquely, which require high-quality randomness.)
1601 * networkd gained support for Foo-Over-UDP, ERSPAN and ISATAP
1602 tunnels. It also gained a new option ForceDHCPv6PDOtherInformation=
1603 for forcing the "Other Information" bit in IPv6 RA messages. The
1604 bonding logic gained four new options AdActorSystemPriority=,
1605 AdUserPortKey=, AdActorSystem= for configuring various 802.3ad
1606 aspects, and DynamicTransmitLoadBalancing= for enabling dynamic
1607 shuffling of flows. The tunnel logic gained a new
1608 IPv6RapidDeploymentPrefix= option for configuring IPv6 Rapid
1609 Deployment. The policy rule logic gained four new options IPProtocol=,
1610 SourcePort= and DestinationPort=, InvertRule=. The bridge logic gained
1611 support for the MulticastToUnicast= option. networkd also gained
1612 support for configuring static IPv4 ARP or IPv6 neighbor entries.
1614 * .preset files (as read by 'systemctl preset') may now be used to
1615 instantiate services.
1617 * /etc/crypttab now understands the sector-size= option to configure
1618 the sector size for an encrypted partition.
1620 * Key material for encrypted disks may now be placed on a formatted
1621 medium, and referenced from /etc/crypttab by the UUID of the file
1622 system, followed by "=" suffixed by the path to the key file.
1624 * The "collect" udev component has been removed without replacement, as
1625 it is neither used nor maintained.
1627 * When the RuntimeDirectory=, StateDirectory=, CacheDirectory=,
1628 LogsDirectory=, ConfigurationDirectory= settings are used in a
1629 service the executed processes will now receive a set of environment
1630 variables containing the full paths of these directories.
1631 Specifically, RUNTIME_DIRECTORY=, STATE_DIRECTORY, CACHE_DIRECTORY,
1632 LOGS_DIRECTORY, CONFIGURATION_DIRECTORY are now set if these options
1633 are used. Note that these options may be used multiple times per
1634 service in which case the resulting paths will be concatenated and
1635 separated by colons.
1637 * Predictable interface naming has been extended to cover InfiniBand
1638 NICs. They will be exposed with an "ib" prefix.
1640 * tmpfiles.d/ line types may now be suffixed with a '-' character, in
1641 which case the respective line failing is ignored.
1643 * .link files may now be used to configure the equivalent to the
1644 "ethtool advertise" commands.
1646 * The sd-device.h and sd-hwdb.h APIs are now exported, as an
1647 alternative to libudev.h. Previously, the latter was just an internal
1648 wrapper around the former, but now these two APIs are exposed
1651 * sd-id128.h gained a new function sd_id128_get_boot_app_specific()
1652 which calculates an app-specific boot ID similar to how
1653 sd_id128_get_machine_app_specific() generates an app-specific machine
1656 * A new tool systemd-id128 has been added that can be used to determine
1657 and generate various 128bit IDs.
1659 * /etc/os-release gained two new standardized fields DOCUMENTATION_URL=
1662 * systemd-hibernate-resume-generator will now honor the "noresume"
1663 kernel command line option, in which case it will bypass resuming
1664 from any hibernated image.
1666 * The systemd-sleep.conf configuration file gained new options
1667 AllowSuspend=, AllowHibernation=, AllowSuspendThenHibernate=,
1668 AllowHybridSleep= for prohibiting specific sleep modes even if the
1669 kernel exports them.
1671 * portablectl is now officially supported and has thus moved to
1674 * bootctl learnt the two new commands "set-default" and "set-oneshot"
1675 for setting the default boot loader item to boot to (either
1676 persistently or only for the next boot). This is currently only
1677 compatible with sd-boot, but may be implemented on other boot loaders
1678 too, that follow the boot loader interface. The updated interface is
1679 now documented here:
1681 https://systemd.io/BOOT_LOADER_INTERFACE
1683 * A new kernel command line option systemd.early_core_pattern= is now
1684 understood which may be used to influence the core_pattern PID 1
1685 installs during early boot.
1687 * busctl learnt two new options -j and --json= for outputting method
1688 call replies, properties and monitoring output in JSON.
1690 * journalctl's JSON output now supports simple ANSI coloring as well as
1691 a new "json-seq" mode for generating RFC7464 output.
1693 * Unit files now support the %g/%G specifiers that resolve to the UNIX
1694 group/GID of the service manager runs as, similar to the existing
1695 %u/%U specifiers that resolve to the UNIX user/UID.
1697 * systemd-logind learnt a new global configuration option
1698 UserStopDelaySec= that may be set in logind.conf. It specifies how
1699 long the systemd --user instance shall remain started after a user
1700 logs out. This is useful to speed up repetitive re-connections of the
1701 same user, as it means the user's service manager doesn't have to be
1702 stopped/restarted on each iteration, but can be reused between
1703 subsequent options. This setting defaults to 10s. systemd-logind also
1704 exports two new properties on its Manager D-Bus objects indicating
1705 whether the system's lid is currently closed, and whether the system
1708 * systemd gained support for a generic boot counting logic, which
1709 generically permits automatic reverting to older boot loader entries
1710 if newer updated ones don't work. The boot loader side is implemented
1711 in sd-boot, but is kept open for other boot loaders too. For details
1714 https://systemd.io/AUTOMATIC_BOOT_ASSESSMENT
1716 * The SuccessAction=/FailureAction= unit file settings now learnt two
1717 new parameters: "exit" and "exit-force", which result in immediate
1718 exiting of the service manager, and are only useful in systemd --user
1719 and container environments.
1721 * Unit files gained support for a pair of options
1722 FailureActionExitStatus=/SuccessActionExitStatus= for configuring the
1723 exit status to use as service manager exit status when
1724 SuccessAction=/FailureAction= is set to exit or exit-force.
1726 * A pair of LogRateLimitIntervalSec=/LogRateLimitBurst= per-service
1727 options may now be used to configure the log rate limiting applied by
1728 journald per-service.
1730 * systemd-analyze gained a new verb "timespan" for parsing and
1731 normalizing time span values (i.e. strings like "5min 7s 8us").
1733 * systemd-analyze also gained a new verb "security" for analyzing the
1734 security and sand-boxing settings of services in order to determine an
1735 "exposure level" for them, indicating whether a service would benefit
1736 from more sand-boxing options turned on for them.
1738 * "systemd-analyze syscall-filter" will now also show system calls
1739 supported by the local kernel but not included in any of the defined
1742 * .nspawn files now understand the Ephemeral= setting, matching the
1743 --ephemeral command line switch.
1745 * sd-event gained the new APIs sd_event_source_get_floating() and
1746 sd_event_source_set_floating() for controlling whether a specific
1747 event source is "floating", i.e. destroyed along with the even loop
1750 * Unit objects on D-Bus gained a new "Refs" property that lists all
1751 clients that currently have a reference on the unit (to ensure it is
1754 * The JoinControllers= option in system.conf is no longer supported, as
1755 it didn't work correctly, is hard to support properly, is legacy (as
1756 the concept only exists on cgroup v1) and apparently wasn't used.
1758 * Journal messages that are generated whenever a unit enters the failed
1759 state are now tagged with a unique MESSAGE_ID. Similarly, messages
1760 generated whenever a service process exits are now made recognizable,
1761 too. A tagged message is also emitted whenever a unit enters the
1762 "dead" state on success.
1764 * systemd-run gained a new switch --working-directory= for configuring
1765 the working directory of the service to start. A shortcut -d is
1766 equivalent, setting the working directory of the service to the
1767 current working directory of the invoking program. The new --shell
1768 (or just -S) option has been added for invoking the $SHELL of the
1769 caller as a service, and implies --pty --same-dir --wait --collect
1770 --service-type=exec. Or in other words, "systemd-run -S" is now the
1771 quickest way to quickly get an interactive in a fully clean and
1772 well-defined system service context.
1774 * machinectl gained a new verb "import-fs" for importing an OS tree
1775 from a directory. Moreover, when a directory or tarball is imported
1776 and single top-level directory found with the OS itself below the OS
1777 tree is automatically mangled and moved one level up.
1779 * systemd-importd will no longer set up an implicit btrfs loop-back
1780 file system on /var/lib/machines. If one is already set up, it will
1781 continue to be used.
1783 * A new generator "systemd-run-generator" has been added. It will
1784 synthesize a unit from one or more program command lines included in
1785 the kernel command line. This is very useful in container managers
1788 # systemd-nspawn -i someimage.raw -b systemd.run='"some command line"'
1790 This will run "systemd-nspawn" on an image, invoke the specified
1791 command line and immediately shut down the container again, returning
1792 the command line's exit code.
1794 * The block device locking logic is now documented:
1796 https://systemd.io/BLOCK_DEVICE_LOCKING
1798 * loginctl and machinectl now optionally output the various tables in
1799 JSON using the --output= switch. It is our intention to add similar
1800 support to systemctl and all other commands.
1802 * udevadm's query and trigger verb now optionally take a .device unit
1805 * systemd-udevd's network naming logic now understands a new
1806 net.naming-scheme= kernel command line switch, which may be used to
1807 pick a specific version of the naming scheme. This helps stabilizing
1808 interface names even as systemd/udev are updated and the naming logic
1811 * sd-id128.h learnt two new auxiliary helpers: sd_id128_is_allf() and
1812 SD_ID128_ALLF to test if a 128bit ID is set to all 0xFF bytes, and to
1813 initialize one to all 0xFF.
1815 * After loading the SELinux policy systemd will now recursively relabel
1816 all files and directories listed in
1817 /run/systemd/relabel-extra.d/*.relabel (which should be simple
1818 newline separated lists of paths) in addition to the ones it already
1819 implicitly relabels in /run, /dev and /sys. After the relabelling is
1820 completed the *.relabel files (and /run/systemd/relabel-extra.d/) are
1821 removed. This is useful to permit initrds (i.e. code running before
1822 the SELinux policy is in effect) to generate files in the host
1823 filesystem safely and ensure that the correct label is applied during
1824 the transition to the host OS.
1826 * KERNEL API BREAKAGE: Linux kernel 4.18 changed behaviour regarding
1827 mknod() handling in user namespaces. Previously mknod() would always
1828 fail with EPERM in user namespaces. Since 4.18 mknod() will succeed
1829 but device nodes generated that way cannot be opened, and attempts to
1830 open them result in EPERM. This breaks the "graceful fallback" logic
1831 in systemd's PrivateDevices= sand-boxing option. This option is
1832 implemented defensively, so that when systemd detects it runs in a
1833 restricted environment (such as a user namespace, or an environment
1834 where mknod() is blocked through seccomp or absence of CAP_SYS_MKNOD)
1835 where device nodes cannot be created the effect of PrivateDevices= is
1836 bypassed (following the logic that 2nd-level sand-boxing is not
1837 essential if the system systemd runs in is itself already sand-boxed
1838 as a whole). This logic breaks with 4.18 in container managers where
1839 user namespacing is used: suddenly PrivateDevices= succeeds setting
1840 up a private /dev/ file system containing devices nodes — but when
1841 these are opened they don't work.
1843 At this point it is recommended that container managers utilizing
1844 user namespaces that intend to run systemd in the payload explicitly
1845 block mknod() with seccomp or similar, so that the graceful fallback
1848 We are very sorry for the breakage and the requirement to change
1849 container configurations for newer kernels. It's purely caused by an
1850 incompatible kernel change. The relevant kernel developers have been
1851 notified about this userspace breakage quickly, but they chose to
1854 * PermissionsStartOnly= setting is deprecated (but is still supported
1855 for backwards compatibility). The same functionality is provided by
1856 the more flexible "+", "!", and "!!" prefixes to ExecStart= and other
1859 * $DBUS_SESSION_BUS_ADDRESS environment variable is not set by
1860 pam_systemd anymore.
1862 * The naming scheme for network devices was changed to always rename
1863 devices, even if they were already renamed by userspace. The "kernel"
1864 policy was changed to only apply as a fallback, if no other naming
1867 * The requirements to build systemd is bumped to meson-0.46 and
1870 Contributions from: afg, Alan Jenkins, Aleksei Timofeyev, Alexander
1871 Filippov, Alexander Kurtz, Alexey Bogdanenko, Andreas Henriksson,
1872 Andrew Jorgensen, Anita Zhang, apnix-uk, Arkan49, Arseny Maslennikov,
1873 asavah, Asbjørn Apeland, aszlig, Bastien Nocera, Ben Boeckel, Benedikt
1874 Morbach, Benjamin Berg, Bruce Zhang, Carlo Caione, Cedric Viou, Chen
1875 Qi, Chris Chiu, Chris Down, Chris Morin, Christian Rebischke, Claudius
1876 Ellsel, Colin Guthrie, dana, Daniel, Daniele Medri, Daniel Kahn
1877 Gillmor, Daniel Rusek, Daniel van Vugt, Dariusz Gadomski, Dave Reisner,
1878 David Anderson, Davide Cavalca, David Leeds, David Malcolm, David
1879 Strauss, David Tardon, Dimitri John Ledkov, Dmitry Torokhov, dj-kaktus,
1880 Dongsu Park, Elias Probst, Emil Soleyman, Erik Kooistra, Ervin Peters,
1881 Evgeni Golov, Evgeny Vereshchagin, Fabrice Fontaine, Faheel Ahmad,
1882 Faizal Luthfi, Felix Yan, Filipe Brandenburger, Franck Bui, Frank
1883 Schaefer, Frantisek Sumsal, Gautier Husson, Gianluca Boiano, Giuseppe
1884 Scrivano, glitsj16, Hans de Goede, Harald Hoyer, Harry Mallon, Harshit
1885 Jain, Helmut Grohne, Henry Tung, Hui Yiqun, imayoda, Insun Pyo, Iwan
1886 Timmer, Jan Janssen, Jan Pokorný, Jan Synacek, Jason A. Donenfeld,
1887 javitoom, Jérémy Nouhaud, Jeremy Su, Jiuyang Liu, João Paulo Rechi
1888 Vita, Joe Hershberger, Joe Rayhawk, Joerg Behrmann, Joerg Steffens,
1889 Jonas Dorel, Jon Ringle, Josh Soref, Julian Andres Klode, Jun Bo Bi,
1890 Jürg Billeter, Keith Busch, Khem Raj, Kirill Marinushkin, Larry
1891 Bernstone, Lennart Poettering, Lion Yang, Li Song, Lorenz
1892 Hübschle-Schneider, Lubomir Rintel, Lucas Werkmeister, Ludwin Janvier,
1893 Lukáš Nykrýn, Luke Shumaker, mal, Marc-Antoine Perennou, Marcin
1894 Skarbek, Marco Trevisan (Treviño), Marian Cepok, Mario Hros, Marko
1895 Myllynen, Markus Grimm, Martin Pitt, Martin Sobotka, Martin Wilck,
1896 Mathieu Trudel-Lapierre, Matthew Leeds, Michael Biebl, Michael Olbrich,
1897 Michael 'pbone' Pobega, Michael Scherer, Michal Koutný, Michal
1898 Sekletar, Michal Soltys, Mike Gilbert, Mike Palmer, Muhammet Kara, Neal
1899 Gompa, Neil Brown, Network Silence, Niklas Tibbling, Nikolas Nyby,
1900 Nogisaka Sadata, Oliver Smith, Patrik Flykt, Pavel Hrdina, Paweł
1901 Szewczyk, Peter Hutterer, Piotr Drąg, Ray Strode, Reinhold Mueller,
1902 Renaud Métrich, Roman Gushchin, Ronny Chevalier, Rubén Suárez Alvarez,
1903 Ruixin Bao, RussianNeuroMancer, Ryutaroh Matsumoto, Saleem Rashid, Sam
1904 Morris, Samuel Morris, Sandy Carter, scootergrisen, Sébastien Bacher,
1905 Sergey Ptashnick, Shawn Landden, Shengyao Xue, Shih-Yuan Lee
1906 (FourDollars), Silvio Knizek, Sjoerd Simons, Stasiek Michalski, Stephen
1907 Gallagher, Steven Allen, Steve Ramage, Susant Sahani, Sven Joachim,
1908 Sylvain Plantefève, Tanu Kaskinen, Tejun Heo, Thiago Macieira, Thomas
1909 Blume, Thomas Haller, Thomas H. P. Andersen, Tim Ruffing, TJ, Tobias
1910 Jungel, Todd Walton, Tommi Rantala, Tomsod M, Tony Novak, Tore
1911 Anderson, Trevonn, Victor Laskurain, Victor Tapia, Violet Halo, Vojtech
1912 Trefny, welaq, William A. Kennington III, William Douglas, Wyatt Ward,
1913 Xiang Fan, Xi Ruoyao, Xuanwo, Yann E. Morin, YmrDtnJu, Yu Watanabe,
1914 Zbigniew Jędrzejewski-Szmek, Zhang Xianwei, Zsolt Dollenstein
1916 — Warsaw, 2018-12-21
1920 * NETWORK INTERFACE DEVICE NAMING CHANGES: systemd-udevd's "net_id"
1921 builtin will name network interfaces differently than in previous
1922 versions for virtual network interfaces created with SR-IOV and NPAR
1923 and for devices where the PCI network controller device does not have
1924 a slot number associated.
1926 SR-IOV virtual devices are now named based on the name of the parent
1927 interface, with a suffix of "v<N>", where <N> is the virtual device
1928 number. Previously those virtual devices were named as if completely
1931 The ninth and later NPAR virtual devices will be named following the
1932 scheme used for the first eight NPAR partitions. Previously those
1933 devices were not renamed and the kernel default (eth<n>) was used.
1935 "net_id" will also generate names for PCI devices where the PCI
1936 network controller device does not have an associated slot number
1937 itself, but one of its parents does. Previously those devices were
1938 not renamed and the kernel default (eth<n>) was used.
1940 * AF_INET and AF_INET6 are dropped from RestrictAddressFamilies= in
1941 systemd-logind.service. Since v235, IPAddressDeny=any has been set to
1942 the unit. So, it is expected that the default behavior of
1943 systemd-logind is not changed. However, if distribution packagers or
1944 administrators disabled or modified IPAddressDeny= setting by a
1945 drop-in config file, then it may be necessary to update the file to
1946 re-enable AF_INET and AF_INET6 to support network user name services,
1949 * When the RestrictNamespaces= unit property is specified multiple
1950 times, then the specified types are merged now. Previously, only the
1951 last assignment was used. So, if distribution packagers or
1952 administrators modified the setting by a drop-in config file, then it
1953 may be necessary to update the file.
1955 * When OnFailure= is used in combination with Restart= on a service
1956 unit, then the specified units will no longer be triggered on
1957 failures that result in restarting. Previously, the specified units
1958 would be activated each time the unit failed, even when the unit was
1959 going to be restarted automatically. This behaviour contradicted the
1960 documentation. With this release the code is adjusted to match the
1963 * systemd-tmpfiles will now print a notice whenever it encounters
1964 tmpfiles.d/ lines referencing the /var/run/ directory. It will
1965 recommend reworking them to use the /run/ directory instead (for
1966 which /var/run/ is simply a symlinked compatibility alias). This way
1967 systemd-tmpfiles can properly detect line conflicts and merge lines
1968 referencing the same file by two paths, without having to access
1971 * systemctl disable/unmask/preset/preset-all cannot be used with
1972 --runtime. Previously this was allowed, but resulted in unintuitive
1973 behaviour that wasn't useful. systemctl disable/unmask will now undo
1974 both runtime and persistent enablement/masking, i.e. it will remove
1975 any relevant symlinks both in /run and /etc.
1977 * Note that all long-running system services shipped with systemd will
1978 now default to a system call whitelist (rather than a blacklist, as
1979 before). In particular, systemd-udevd will now enforce one too. For
1980 most cases this should be safe, however downstream distributions
1981 which disabled sandboxing of systemd-udevd (specifically the
1982 MountFlags= setting), might want to disable this security feature
1983 too, as the default whitelisting will prohibit all mount, swap,
1984 reboot and clock changing operations from udev rules.
1986 * sd-boot acquired new loader configuration settings to optionally turn
1987 off Windows and MacOS boot partition discovery as well as
1988 reboot-into-firmware menu items. It is also able to pick a better
1989 screen resolution for HiDPI systems, and now provides loader
1990 configuration settings to change the resolution explicitly.
1992 * systemd-resolved now supports DNS-over-TLS. It's still
1993 turned off by default, use DNSOverTLS=opportunistic to turn it on in
1994 resolved.conf. We intend to make this the default as soon as couple
1995 of additional techniques for optimizing the initial latency caused by
1996 establishing a TLS/TCP connection are implemented.
1998 * systemd-resolved.service and systemd-networkd.service now set
1999 DynamicUser=yes. The users systemd-resolve and systemd-network are
2000 not created by systemd-sysusers anymore.
2002 NOTE: This has a chance of breaking nss-ldap and similar NSS modules
2003 that embed a network facing module into any process using getpwuid()
2004 or related call: the dynamic allocation of the user ID for
2005 systemd-resolved.service means the service manager has to check NSS
2006 if the user name is already taken when forking off the service. Since
2007 the user in the common case won't be defined in /etc/passwd the
2008 lookup is likely to trigger nss-ldap which in turn might use NSS to
2009 ask systemd-resolved for hostname lookups. This will hence result in
2010 a deadlock: a user name lookup in order to start
2011 systemd-resolved.service will result in a host name lookup for which
2012 systemd-resolved.service needs to be started already. There are
2013 multiple ways to work around this problem: pre-allocate the
2014 "systemd-resolve" user on such systems, so that nss-ldap won't be
2015 triggered; or use a different NSS package that doesn't do networking
2016 in-process but provides a local asynchronous name cache; or configure
2017 the NSS package to avoid lookups for UIDs in the range `pkg-config
2018 systemd --variable=dynamicuidmin` … `pkg-config systemd
2019 --variable=dynamicuidmax`, so that it does not consider itself
2020 authoritative for the same UID range systemd allocates dynamic users
2023 * The systemd-resolve tool has been renamed to resolvectl (it also
2024 remains available under the old name, for compatibility), and its
2025 interface is now verb-based, similar in style to the other <xyz>ctl
2026 tools, such as systemctl or loginctl.
2028 * The resolvectl/systemd-resolve tool also provides 'resolvconf'
2029 compatibility. It may be symlinked under the 'resolvconf' name, in
2030 which case it will take arguments and input compatible with the
2031 Debian and FreeBSD resolvconf tool.
2033 * Support for suspend-then-hibernate has been added, i.e. a sleep mode
2034 where the system initially suspends, and after a timeout resumes and
2037 * networkd's ClientIdentifier= now accepts a new option "duid-only". If
2038 set the client will only send a DUID as client identifier.
2040 * The nss-systemd glibc NSS module will now enumerate dynamic users and
2041 groups in effect. Previously, it could resolve UIDs/GIDs to user
2042 names/groups and vice versa, but did not support enumeration.
2044 * journald's Compress= configuration setting now optionally accepts a
2045 byte threshold value. All journal objects larger than this threshold
2046 will be compressed, smaller ones will not. Previously this threshold
2047 was not configurable and set to 512.
2049 * A new system.conf setting NoNewPrivileges= is now available which may
2050 be used to turn off acquisition of new privileges system-wide
2051 (i.e. set Linux' PR_SET_NO_NEW_PRIVS for PID 1 itself, and thus also
2052 for all its children). Note that turning this option on means setuid
2053 binaries and file system capabilities lose their special powers.
2054 While turning on this option is a big step towards a more secure
2055 system, doing so is likely to break numerous pre-existing UNIX tools,
2056 in particular su and sudo.
2058 * A new service systemd-time-sync-wait.service has been added. If
2059 enabled it will delay the time-sync.target unit at boot until time
2060 synchronization has been received from the network. This
2061 functionality is useful on systems lacking a local RTC or where it is
2062 acceptable that the boot process shall be delayed by external network
2065 * When hibernating, systemd will now inform the kernel of the image
2066 write offset, on kernels new enough to support this. This means swap
2067 files should work for hibernation now.
2069 * When loading unit files, systemd will now look for drop-in unit files
2070 extensions in additional places. Previously, for a unit file name
2071 "foo-bar-baz.service" it would look for dropin files in
2072 "foo-bar-baz.service.d/*.conf". Now, it will also look in
2073 "foo-bar-.service.d/*.conf" and "foo-.service.d/", i.e. at the
2074 service name truncated after all inner dashes. This scheme allows
2075 writing drop-ins easily that apply to a whole set of unit files at
2076 once. It's particularly useful for mount and slice units (as their
2077 naming is prefix based), but is also useful for service and other
2078 units, for packages that install multiple unit files at once,
2079 following a strict naming regime of beginning the unit file name with
2080 the package's name. Two new specifiers are now supported in unit
2081 files to match this: %j and %J are replaced by the part of the unit
2082 name following the last dash.
2084 * Unit files and other configuration files that support specifier
2085 expansion now understand another three new specifiers: %T and %V will
2086 resolve to /tmp and /var/tmp respectively, or whatever temporary
2087 directory has been set for the calling user. %E will expand to either
2088 /etc (for system units) or $XDG_CONFIG_HOME (for user units).
2090 * The ExecStart= lines of unit files are no longer required to
2091 reference absolute paths. If non-absolute paths are specified the
2092 specified binary name is searched within the service manager's
2093 built-in $PATH, which may be queried with 'systemd-path
2094 search-binaries-default'. It's generally recommended to continue to
2095 use absolute paths for all binaries specified in unit files.
2097 * Units gained a new load state "bad-setting", which is used when a
2098 unit file was loaded, but contained fatal errors which prevent it
2099 from being started (for example, a service unit has been defined
2100 lacking both ExecStart= and ExecStop= lines).
2102 * coredumpctl's "gdb" verb has been renamed to "debug", in order to
2103 support alternative debuggers, for example lldb. The old name
2104 continues to be available however, for compatibility reasons. Use the
2105 new --debugger= switch or the $SYSTEMD_DEBUGGER environment variable
2106 to pick an alternative debugger instead of the default gdb.
2108 * systemctl and the other tools will now output escape sequences that
2109 generate proper clickable hyperlinks in various terminal emulators
2110 where useful (for example, in the "systemctl status" output you can
2111 now click on the unit file name to quickly open it in the
2112 editor/viewer of your choice). Note that not all terminal emulators
2113 support this functionality yet, but many do. Unfortunately, the
2114 "less" pager doesn't support this yet, hence this functionality is
2115 currently automatically turned off when a pager is started (which
2116 happens quite often due to auto-paging). We hope to remove this
2117 limitation as soon as "less" learns these escape sequences. This new
2118 behaviour may also be turned off explicitly with the $SYSTEMD_URLIFY
2119 environment variable. For details on these escape sequences see:
2120 https://gist.github.com/egmontkob/eb114294efbcd5adb1944c9f3cb5feda
2122 * networkd's .network files now support a new IPv6MTUBytes= option for
2123 setting the MTU used by IPv6 explicitly as well as a new MTUBytes=
2124 option in the [Route] section to configure the MTU to use for
2125 specific routes. It also gained support for configuration of the DHCP
2126 "UserClass" option through the new UserClass= setting. It gained
2127 three new options in the new [CAN] section for configuring CAN
2128 networks. The MULTICAST and ALLMULTI interface flags may now be
2129 controlled explicitly with the new Multicast= and AllMulticast=
2132 * networkd will now automatically make use of the kernel's route
2133 expiration feature, if it is available.
2135 * udevd's .link files now support setting the number of receive and
2136 transmit channels, using the RxChannels=, TxChannels=,
2137 OtherChannels=, CombinedChannels= settings.
2139 * Support for UDPSegmentationOffload= has been removed, given its
2140 limited support in hardware, and waning software support.
2142 * networkd's .netdev files now support creating "netdevsim" interfaces.
2144 * PID 1 learnt a new bus call GetUnitByControlGroup() which may be used
2145 to query the unit belonging to a specific kernel control group.
2147 * systemd-analyze gained a new verb "cat-config", which may be used to
2148 dump the contents of any configuration file, with all its matching
2149 drop-in files added in, and honouring the usual search and masking
2150 logic applied to systemd configuration files. For example use
2151 "systemd-analyze cat-config systemd/system.conf" to get the complete
2152 system configuration file of systemd how it would be loaded by PID 1
2153 itself. Similar to this, various tools such as systemd-tmpfiles or
2154 systemd-sysusers, gained a new option "--cat-config", which does the
2155 corresponding operation for their own configuration settings. For
2156 example, "systemd-tmpfiles --cat-config" will now output the full
2157 list of tmpfiles.d/ lines in place.
2159 * timedatectl gained three new verbs: "show" shows bus properties of
2160 systemd-timedated, "timesync-status" shows the current NTP
2161 synchronization state of systemd-timesyncd, and "show-timesync"
2162 shows bus properties of systemd-timesyncd.
2164 * systemd-timesyncd gained a bus interface on which it exposes details
2167 * A new environment variable $SYSTEMD_TIMEDATED_NTP_SERVICES is now
2168 understood by systemd-timedated. It takes a colon-separated list of
2169 unit names of NTP client services. The list is used by
2170 "timedatectl set-ntp".
2172 * systemd-nspawn gained a new --rlimit= switch for setting initial
2173 resource limits for the container payload. There's a new switch
2174 --hostname= to explicitly override the container's hostname. A new
2175 --no-new-privileges= switch may be used to control the
2176 PR_SET_NO_NEW_PRIVS flag for the container payload. A new
2177 --oom-score-adjust= switch controls the OOM scoring adjustment value
2178 for the payload. The new --cpu-affinity= switch controls the CPU
2179 affinity of the container payload. The new --resolv-conf= switch
2180 allows more detailed control of /etc/resolv.conf handling of the
2181 container. Similarly, the new --timezone= switch allows more detailed
2182 control of /etc/localtime handling of the container.
2184 * systemd-detect-virt gained a new --list switch, which will print a
2185 list of all currently known VM and container environments.
2187 * Support for "Portable Services" has been added, see
2188 doc/PORTABLE_SERVICES.md for details. Currently, the support is still
2189 experimental, but this is expected to change soon. Reflecting this
2190 experimental state, the "portablectl" binary is not installed into
2191 /usr/bin yet. The binary has to be called with the full path
2192 /usr/lib/systemd/portablectl instead.
2194 * journalctl's and systemctl's -o switch now knows a new log output
2195 mode "with-unit". The output it generates is very similar to the
2196 regular "short" mode, but displays the unit name instead of the
2197 syslog tag for each log line. Also, the date is shown with timezone
2198 information. This mode is probably more useful than the classic
2199 "short" output mode for most purposes, except where pixel-perfect
2200 compatibility with classic /var/log/messages formatting is required.
2202 * A new --dump-bus-properties switch has been added to the systemd
2203 binary, which may be used to dump all supported D-Bus properties.
2204 (Options which are still supported, but are deprecated, are *not*
2207 * sd-bus gained a set of new calls:
2208 sd_bus_slot_set_floating()/sd_bus_slot_get_floating() may be used to
2209 enable/disable the "floating" state of a bus slot object,
2210 i.e. whether the slot object pins the bus it is allocated for into
2211 memory or if the bus slot object gets disconnected when the bus goes
2212 away. sd_bus_open_with_description(),
2213 sd_bus_open_user_with_description(),
2214 sd_bus_open_system_with_description() may be used to allocate bus
2215 objects and set their description string already during allocation.
2217 * sd-event gained support for watching inotify events from the event
2218 loop, in an efficient way, sharing inotify handles between multiple
2219 users. For this a new function sd_event_add_inotify() has been added.
2221 * sd-event and sd-bus gained support for calling special user-supplied
2222 destructor functions for userdata pointers associated with
2223 sd_event_source, sd_bus_slot, and sd_bus_track objects. For this new
2224 functions sd_bus_slot_set_destroy_callback,
2225 sd_bus_slot_get_destroy_callback, sd_bus_track_set_destroy_callback,
2226 sd_bus_track_get_destroy_callback,
2227 sd_event_source_set_destroy_callback,
2228 sd_event_source_get_destroy_callback have been added.
2230 * The "net.ipv4.tcp_ecn" sysctl will now be turned on by default.
2232 * PID 1 will now automatically reschedule .timer units whenever the
2233 local timezone changes. (They previously got rescheduled
2234 automatically when the system clock changed.)
2236 * New documentation has been added to document cgroups delegation,
2237 portable services and the various code quality tools we have set up:
2239 https://github.com/systemd/systemd/blob/master/docs/CGROUP_DELEGATION.md
2240 https://github.com/systemd/systemd/blob/master/docs/PORTABLE_SERVICES.md
2241 https://github.com/systemd/systemd/blob/master/docs/CODE_QUALITY.md
2243 * The Boot Loader Specification has been added to the source tree.
2245 https://github.com/systemd/systemd/blob/master/docs/BOOT_LOADER_SPECIFICATION.md
2247 While moving it into our source tree we have updated it and further
2248 changes are now accepted through the usual github PR workflow.
2250 * pam_systemd will now look for PAM userdata fields systemd.memory_max,
2251 systemd.tasks_max, systemd.cpu_weight, systemd.io_weight set by
2252 earlier PAM modules. The data in these fields is used to initialize
2253 the session scope's resource properties. Thus external PAM modules
2254 may now configure per-session limits, for example sourced from
2255 external user databases.
2257 * socket units with Accept=yes will now maintain a "refused" counter in
2258 addition to the existing "accepted" counter, counting connections
2259 refused due to the enforced limits.
2261 * The "systemd-path search-binaries-default" command may now be use to
2262 query the default, built-in $PATH PID 1 will pass to the services it
2265 * A new unit file setting PrivateMounts= has been added. It's a boolean
2266 option. If enabled the unit's processes are invoked in their own file
2267 system namespace. Note that this behaviour is also implied if any
2268 other file system namespacing options (such as PrivateTmp=,
2269 PrivateDevices=, ProtectSystem=, …) are used. This option is hence
2270 primarily useful for services that do not use any of the other file
2271 system namespacing options. One such service is systemd-udevd.service
2272 where this is now used by default.
2274 * ConditionSecurity= gained a new value "uefi-secureboot" that is true
2275 when the system is booted in UEFI "secure mode".
2277 * A new unit "system-update-pre.target" is added, which defines an
2278 optional synchronization point for offline system updates, as
2279 implemented by the pre-existing "system-update.target" unit. It
2280 allows ordering services before the service that executes the actual
2281 update process in a generic way.
2283 * Systemd now emits warnings whenever .include syntax is used.
2285 Contributions from: Adam Duskett, Alan Jenkins, Alessandro Casale,
2286 Alexander Kurtz, Alex Gartrell, Anssi Hannula, Arnaud Rebillout, Brian
2287 J. Murrell, Bruno Vernay, Chris Lamb, Chris Lesiak, Christian Brauner,
2288 Christian Hesse, Christian Rebischke, Colin Guthrie, Daniel Dao, Daniel
2289 Lin, Danylo Korostil, Davide Cavalca, David Tardon, Dimitri John
2290 Ledkov, Dmitriy Geels, Douglas Christman, Elia Geretto, emelenas, Emil
2291 Velikov, Evgeny Vereshchagin, Felipe Sateler, Feng Sun, Filipe
2292 Brandenburger, Franck Bui, futpib, Giuseppe Scrivano, Guillem Jover,
2293 guixxx, Hannes Reinecke, Hans de Goede, Harald Hoyer, Henrique Dante de
2294 Almeida, Hiram van Paassen, Ian Miell, Igor Gnatenko, Ivan Shapovalov,
2295 Iwan Timmer, James Cowgill, Jan Janssen, Jan Synacek, Jared Kazimir,
2296 Jérémy Rosen, João Paulo Rechi Vita, Joost Heitbrink, Jui-Chi Ricky
2297 Liang, Jürg Billeter, Kai-Heng Feng, Karol Augustin, Kay Sievers,
2298 Krzysztof Nowicki, Lauri Tirkkonen, Lennart Poettering, Leonard König,
2299 Long Li, Luca Boccassi, Lucas Werkmeister, Marcel Hoppe, Marc
2300 Kleine-Budde, Mario Limonciello, Martin Jansa, Martin Wilck, Mathieu
2301 Malaterre, Matteo F. Vescovi, Matthew McGinn, Matthias-Christian Ott,
2302 Michael Biebl, Michael Olbrich, Michael Prokop, Michal Koutný, Michal
2303 Sekletar, Mike Gilbert, Mikhail Kasimov, Milan Broz, Milan Pässler,
2304 Mladen Pejaković, Muhammet Kara, Nicolas Boichat, Omer Katz, Paride
2305 Legovini, Paul Menzel, Paul Milliken, Pavel Hrdina, Peter A. Bigot,
2306 Peter D'Hoye, Peter Hutterer, Peter Jones, Philip Sequeira, Philip
2307 Withnall, Piotr Drąg, Radostin Stoyanov, Ricardo Salveti de Araujo,
2308 Ronny Chevalier, Rosen Penev, Rubén Suárez Alvarez, Ryan Gonzalez,
2309 Salvo Tomaselli, Sebastian Reichel, Sergey Ptashnick, Sergio Lindo
2310 Mansilla, Stefan Schweter, Stephen Hemminger, Stuart Hayes, Susant
2311 Sahani, Sylvain Plantefève, Thomas H. P. Andersen, Tobias Jungel,
2312 Tomasz Torcz, Vito Caputo, Will Dietz, Will Thompson, Wim van Mourik,
2313 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
2315 — Berlin, 2018-06-22
2319 * The MemoryAccounting= unit property now defaults to on. After
2320 discussions with the upstream control group maintainers we learnt
2321 that the negative impact of cgroup memory accounting on current
2322 kernels is finally relatively minimal, so that it should be safe to
2323 enable this by default without affecting system performance. Besides
2324 memory accounting only task accounting is turned on by default, all
2325 other forms of resource accounting (CPU, IO, IP) remain off for now,
2326 because it's not clear yet that their impact is small enough to move
2327 from opt-in to opt-out. We recommend downstreams to leave memory
2328 accounting on by default if kernel 4.14 or higher is primarily
2329 used. On very resource constrained systems or when support for old
2330 kernels is a necessity, -Dmemory-accounting-default=false can be used
2331 to revert this change.
2333 * rpm scriptlets to update the udev hwdb and rules (%udev_hwdb_update,
2334 %udev_rules_update) and the journal catalog (%journal_catalog_update)
2335 from the upgrade scriptlets of individual packages now do nothing.
2336 Transfiletriggers have been added which will perform those updates
2337 once at the end of the transaction.
2339 Similar transfiletriggers have been added to execute any sysctl.d
2340 and binfmt.d rules. Thus, it should be unnecessary to provide any
2341 scriptlets to execute this configuration from package installation
2344 * systemd-sysusers gained a mode where the configuration to execute is
2345 specified on the command line, but this configuration is not executed
2346 directly, but instead it is merged with the configuration on disk,
2347 and the result is executed. This is useful for package installation
2348 scripts which want to create the user before installing any files on
2349 disk (in case some of those files are owned by that user), while
2350 still allowing local admin overrides.
2352 This functionality is exposed to rpm scriptlets through a new
2353 %sysusers_create_package macro. Old %sysusers_create and
2354 %sysusers_create_inline macros are deprecated.
2356 A transfiletrigger for sysusers.d configuration is now installed,
2357 which means that it should be unnecessary to call systemd-sysusers from
2358 package installation scripts, unless the package installs any files
2359 owned by those newly-created users, in which case
2360 %sysusers_create_package should be used.
2362 * Analogous change has been done for systemd-tmpfiles: it gained a mode
2363 where the command-line configuration is merged with the configuration
2364 on disk. This is exposed as the new %tmpfiles_create_package macro,
2365 and %tmpfiles_create is deprecated. A transfiletrigger is installed
2366 for tmpfiles.d, hence it should be unnecessary to call systemd-tmpfiles
2367 from package installation scripts.
2369 * sysusers.d configuration for a user may now also specify the group
2370 number, in addition to the user number ("u username 123:456"), or
2371 without the user number ("u username -:456").
2373 * Configution items for systemd-sysusers can now be specified as
2374 positional arguments when the new --inline switch is used.
2376 * The login shell of users created through sysusers.d may now be
2377 specified (previously, it was always /bin/sh for root and
2378 /sbin/nologin for other users).
2380 * systemd-analyze gained a new --global switch to look at global user
2381 configuration. It also gained a unit-paths verb to list the unit load
2382 paths that are compiled into systemd (which can be used with
2383 --systemd, --user, or --global).
2385 * udevadm trigger gained a new --settle/-w option to wait for any
2386 triggered events to finish (but just those, and not any other events
2387 which are triggered meanwhile).
2389 * The action that systemd-logind takes when the lid is closed and the
2390 machine is connected to external power can now be configured using
2391 HandleLidSwitchExternalPower= in logind.conf. Previously, this action
2392 was determined by HandleLidSwitch=, and, for backwards compatibility,
2393 is still is, if HandleLidSwitchExternalPower= is not explicitly set.
2395 * journalctl will periodically call sd_journal_process() to make it
2396 resilient against inotify queue overruns when journal files are
2397 rotated very quickly.
2399 * Two new functions in libsystemd — sd_bus_get_n_queued_read and
2400 sd_bus_get_n_queued_write — may be used to check the number of
2401 pending bus messages.
2403 * systemd gained a new
2404 org.freedesktop.systemd1.Manager.AttachProcessesToUnit dbus call
2405 which can be used to migrate foreign processes to scope and service
2406 units. The primary user for this new API is systemd itself: the
2407 systemd --user instance uses this call of the systemd --system
2408 instance to migrate processes if it itself gets the request to
2409 migrate processes and the kernel refuses this due to access
2410 restrictions. Thanks to this "systemd-run --scope --user …" works
2411 again in pure cgroup v2 environments when invoked from the user
2414 * A new TemporaryFileSystem= setting can be used to mask out part of
2415 the real file system tree with tmpfs mounts. This may be combined
2416 with BindPaths= and BindReadOnlyPaths= to hide files or directories
2417 not relevant to the unit, while still allowing some paths lower in
2418 the tree to be accessed.
2420 ProtectHome=tmpfs may now be used to hide user home and runtime
2421 directories from units, in a way that is mostly equivalent to
2422 "TemporaryFileSystem=/home /run/user /root".
2424 * Non-service units are now started with KeyringMode=shared by default.
2425 This means that mount and swapon and other mount tools have access
2426 to keys in the main keyring.
2428 * /sys/fs/bpf is now mounted automatically.
2430 * QNX virtualization is now detected by systemd-detect-virt and may
2431 be used in ConditionVirtualization=.
2433 * IPAccounting= may now be enabled also for slice units.
2435 * A new -Dsplit-bin= build configuration switch may be used to specify
2436 whether bin and sbin directories are merged, or if they should be
2437 included separately in $PATH and various listings of executable
2438 directories. The build configuration scripts will try to autodetect
2439 the proper values of -Dsplit-usr= and -Dsplit-bin= based on build
2440 system, but distributions are encouraged to configure this
2443 * A new -Dok-color= build configuration switch may be used to change
2444 the colour of "OK" status messages.
2446 * UPGRADE ISSUE: serialization of units using JoinsNamespaceOf= with
2447 PrivateNetwork=yes was buggy in previous versions of systemd. This
2448 means that after the upgrade and daemon-reexec, any such units must
2451 * INCOMPATIBILITY: as announced in the NEWS for 237, systemd-tmpfiles
2452 will not exclude read-only files owned by root from cleanup.
2454 Contributions from: Alan Jenkins, Alexander F Rødseth, Alexis Jeandet,
2455 Andika Triwidada, Andrei Gherzan, Ansgar Burchardt, antizealot1337,
2456 Batuhan Osman Taşkaya, Beniamino Galvani, Bill Yodlowsky, Caio Marcelo
2457 de Oliveira Filho, CuBiC, Daniele Medri, Daniel Mouritzen, Daniel
2458 Rusek, Davide Cavalca, Dimitri John Ledkov, Douglas Christman, Evgeny
2459 Vereshchagin, Faalagorn, Filipe Brandenburger, Franck Bui, futpib,
2460 Giacomo Longo, Gunnar Hjalmarsson, Hans de Goede, Hermann Gausterer,
2461 Iago López Galeiras, Jakub Filak, Jan Synacek, Jason A. Donenfeld,
2462 Javier Martinez Canillas, Jérémy Rosen, Lennart Poettering, Lucas
2463 Werkmeister, Mao Huang, Marco Gulino, Michael Biebl, Michael Vogt,
2464 MilhouseVH, Neal Gompa (ニール・ゴンパ), Oleander Reis, Olof Mogren,
2465 Patrick Uiterwijk, Peter Hutterer, Peter Portante, Piotr Drąg, Robert
2466 Antoni Buj Gelonch, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
2467 Fowler, SjonHortensius, snorreflorre, Susant Sahani, Sylvain
2468 Plantefève, Thomas Blume, Thomas Haller, Vito Caputo, Yu Watanabe,
2469 Zbigniew Jędrzejewski-Szmek, Марко М. Костић (Marko M. Kostić)
2471 — Warsaw, 2018-03-05
2475 * Some keyboards come with a zoom see-saw or rocker which until now got
2476 mapped to the Linux "zoomin/out" keys in hwdb. However, these
2477 keycodes are not recognized by any major desktop. They now produce
2478 Up/Down key events so that they can be used for scrolling.
2480 * INCOMPATIBILITY: systemd-tmpfiles' "f" lines changed behaviour
2481 slightly: previously, if an argument was specified for lines of this
2482 type (i.e. the right-most column was set) this string was appended to
2483 existing files each time systemd-tmpfiles was run. This behaviour was
2484 different from what the documentation said, and not particularly
2485 useful, as repeated systemd-tmpfiles invocations would not be
2486 idempotent and grow such files without bounds. With this release
2487 behaviour has been altered to match what the documentation says:
2488 lines of this type only have an effect if the indicated files don't
2489 exist yet, and only then the argument string is written to the file.
2491 * FUTURE INCOMPATIBILITY: In systemd v238 we intend to slightly change
2492 systemd-tmpfiles behaviour: previously, read-only files owned by root
2493 were always excluded from the file "aging" algorithm (i.e. the
2494 automatic clean-up of directories like /tmp based on
2495 atime/mtime/ctime). We intend to drop this restriction, and age files
2496 by default even when owned by root and read-only. This behaviour was
2497 inherited from older tools, but there have been requests to remove
2498 it, and it's not obvious why this restriction was made in the first
2499 place. Please speak up now, if you are aware of software that reqires
2500 this behaviour, otherwise we'll remove the restriction in v238.
2502 * A new environment variable $SYSTEMD_OFFLINE is now understood by
2503 systemctl. It takes a boolean argument. If on, systemctl assumes it
2504 operates on an "offline" OS tree, and will not attempt to talk to the
2505 service manager. Previously, this mode was implicitly enabled if a
2506 chroot() environment was detected, and this new environment variable
2507 now provides explicit control.
2509 * .path and .socket units may now be created transiently, too.
2510 Previously only service, mount, automount and timer units were
2511 supported as transient units. The systemd-run tool has been updated
2512 to expose this new functionality, you may hence use it now to bind
2513 arbitrary commands to path or socket activation on-the-fly from the
2514 command line. Moreover, almost all properties are now exposed for the
2515 unit types that already supported transient operation.
2517 * The systemd-mount command gained support for a new --owner= parameter
2518 which takes a user name, which is then resolved and included in uid=
2519 and gid= mount options string of the file system to mount.
2521 * A new unit condition ConditionControlGroupController= has been added
2522 that checks whether a specific cgroup controller is available.
2524 * Unit files, udev's .link files, and systemd-networkd's .netdev and
2525 .network files all gained support for a new condition
2526 ConditionKernelVersion= for checking against specific kernel
2529 * In systemd-networkd, the [IPVLAN] section in .netdev files gained
2530 support for configuring device flags in the Flags= setting. In the
2531 same files, the [Tunnel] section gained support for configuring
2532 AllowLocalRemote=. The [Route] section in .network files gained
2533 support for configuring InitialCongestionWindow=,
2534 InitialAdvertisedReceiveWindow= and QuickAck=. The [DHCP] section now
2535 understands RapidCommit=.
2537 * systemd-networkd's DHCPv6 support gained support for Prefix
2540 * sd-bus gained support for a new "watch-bind" feature. When this
2541 feature is enabled, an sd_bus connection may be set up to connect to
2542 an AF_UNIX socket in the file system as soon as it is created. This
2543 functionality is useful for writing early-boot services that
2544 automatically connect to the system bus as soon as it is started,
2545 without ugly time-based polling. systemd-networkd and
2546 systemd-resolved have been updated to make use of this
2547 functionality. busctl exposes this functionality in a new
2548 --watch-bind= command line switch.
2550 * sd-bus will now optionally synthesize a local "Connected" signal as
2551 soon as a D-Bus connection is set up fully. This message mirrors the
2552 already existing "Disconnected" signal which is synthesized when the
2553 connection is terminated. This signal is generally useful but
2554 particularly handy in combination with the "watch-bind" feature
2555 described above. Synthesizing of this message has to be requested
2556 explicitly through the new API call sd_bus_set_connected_signal(). In
2557 addition a new call sd_bus_is_ready() has been added that checks
2558 whether a connection is fully set up (i.e. between the "Connected" and
2559 "Disconnected" signals).
2561 * sd-bus gained two new calls sd_bus_request_name_async() and
2562 sd_bus_release_name_async() for asynchronously registering bus
2563 names. Similar, there is now sd_bus_add_match_async() for installing
2564 a signal match asynchronously. All of systemd's own services have
2565 been updated to make use of these calls. Doing these operations
2566 asynchronously has two benefits: it reduces the risk of deadlocks in
2567 case of cyclic dependencies between bus services, and it speeds up
2568 service initialization since synchronization points for bus
2569 round-trips are removed.
2571 * sd-bus gained two new calls sd_bus_match_signal() and
2572 sd_bus_match_signal_async(), which are similar to sd_bus_add_match()
2573 and sd_bus_add_match_async() but instead of taking a D-Bus match
2574 string take match fields as normal function parameters.
2576 * sd-bus gained two new calls sd_bus_set_sender() and
2577 sd_bus_message_set_sender() for setting the sender name of outgoing
2578 messages (either for all outgoing messages or for just one specific
2579 one). These calls are only useful in direct connections as on
2580 brokered connections the broker fills in the sender anyway,
2581 overwriting whatever the client filled in.
2583 * sd-event gained a new pseudo-handle that may be specified on all API
2584 calls where an "sd_event*" object is expected: SD_EVENT_DEFAULT. When
2585 used this refers to the default event loop object of the calling
2586 thread. Note however that this does not implicitly allocate one —
2587 which has to be done prior by using sd_event_default(). Similarly
2588 sd-bus gained three new pseudo-handles SD_BUS_DEFAULT,
2589 SD_BUS_DEFAULT_USER, SD_BUS_DEFAULT_SYSTEM that may be used to refer
2590 to the default bus of the specified type of the calling thread. Here
2591 too this does not implicitly allocate bus connection objects, this
2592 has to be done prior with sd_bus_default() and friends.
2594 * sd-event gained a new call pair
2595 sd_event_source_{get|set}_io_fd_own(). This may be used to request
2596 automatic closure of the file descriptor an IO event source watches
2597 when the event source is destroyed.
2599 * systemd-networkd gained support for natively configuring WireGuard
2602 * In previous versions systemd synthesized user records both for the
2603 "nobody" (UID 65534) and "root" (UID 0) users in nss-systemd and
2604 internally. In order to simplify distribution-wide renames of the
2605 "nobody" user (like it is planned in Fedora: nfsnobody → nobody), a
2606 new transitional flag file has been added: if
2607 /etc/systemd/dont-synthesize-nobody exists synthesizing of the 65534
2608 user and group record within the systemd codebase is disabled.
2610 * systemd-notify gained a new --uid= option for selecting the source
2611 user/UID to use for notification messages sent to the service
2614 * journalctl gained a new --grep= option to list only entries in which
2615 the message matches a certain pattern. By default matching is case
2616 insensitive if the pattern is lowercase, and case sensitive
2617 otherwise. Option --case-sensitive=yes|no can be used to override
2618 this an specify case sensitivity or case insensitivity.
2620 * There's now a "systemd-analyze service-watchdogs" command for printing
2621 the current state of the service runtime watchdog, and optionally
2622 enabling or disabling the per-service watchdogs system-wide if given a
2623 boolean argument (i.e. the concept you configure in WatchdogSec=), for
2624 debugging purposes. There's also a kernel command line option
2625 systemd.service_watchdogs= for controlling the same.
2627 * Two new "log-level" and "log-target" options for systemd-analyze were
2628 added that merge the now deprecated get-log-level, set-log-level and
2629 get-log-target, set-log-target pairs. The deprecated options are still
2630 understood for backwards compatibility. The two new options print the
2631 current value when no arguments are given, and set them when a
2632 level/target is given as an argument.
2634 * sysusers.d's "u" lines now optionally accept both a UID and a GID
2635 specification, separated by a ":" character, in order to create users
2636 where UID and GID do not match.
2638 Contributions from: Adam Duskett, Alan Jenkins, Alexander Kuleshov,
2639 Alexis Deruelle, Andrew Jeddeloh, Armin Widegreen, Batuhan Osman
2640 Taşkaya, Björn Esser, bleep_blop, Bruce A. Johnson, Chris Down, Clinton
2641 Roy, Colin Walters, Daniel Rusek, Dimitri John Ledkov, Dmitry Rozhkov,
2642 Evgeny Vereshchagin, Ewout van Mansom, Felipe Sateler, Franck Bui,
2643 Frantisek Sumsal, George Gaydarov, Gianluca Boiano, Hans-Christian
2644 Noren Egtvedt, Hans de Goede, Henrik Grindal Bakken, Jan Alexander
2645 Steffens, Jan Klötzke, Jason A. Donenfeld, jdkbx, Jérémy Rosen,
2646 Jerónimo Borque, John Lin, John Paul Herold, Jonathan Rudenberg, Jörg
2647 Thalheim, Ken (Bitsko) MacLeod, Larry Bernstone, Lennart Poettering,
2648 Lucas Werkmeister, Maciej S. Szmigiero, Marek Čermák, Martin Pitt,
2649 Mathieu Malaterre, Matthew Thode, Matthias-Christian Ott, Max Harmathy,
2650 Michael Biebl, Michael Vogt, Michal Koutný, Michal Sekletar, Michał
2651 Szczepański, Mike Gilbert, Nathaniel McCallum, Nicolas Chauvet, Olaf
2652 Hering, Olivier Schwander, Patrik Flykt, Paul Cercueil, Peter Hutterer,
2653 Piotr Drąg, Raphael Vogelgsang, Reverend Homer, Robert Kolchmeyer,
2654 Samuel Dionne-Riel, Sergey Ptashnick, Shawn Landden, Susant Sahani,
2655 Sylvain Plantefève, Thomas H. P. Andersen, Thomas Huth, Tomasz
2656 Bachorski, Vladislav Vishnyakov, Wieland Hoffmann, Yu Watanabe, Zachary
2657 Winnerman, Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски, Дилян
2664 * The modprobe.d/ drop-in for the bonding.ko kernel module introduced
2665 in v235 has been extended to also set the dummy.ko module option
2666 numdummies=0, preventing the kernel from automatically creating
2667 dummy0. All dummy interfaces must now be explicitly created.
2669 * Unknown '%' specifiers in configuration files are now rejected. This
2670 applies to units and tmpfiles.d configuration. Any percent characters
2671 that are followed by a letter or digit that are not supposed to be
2672 interpreted as the beginning of a specifier should be escaped by
2673 doubling ("%%"). (So "size=5%" is still accepted, as well as
2674 "size=5%,foo=bar", but not "LABEL=x%y%z" since %y and %z are not
2675 valid specifiers today.)
2677 * systemd-resolved now maintains a new dynamic
2678 /run/systemd/resolve/stub-resolv.conf compatibility file. It is
2679 recommended to make /etc/resolv.conf a symlink to it. This file
2680 points at the systemd-resolved stub DNS 127.0.0.53 resolver and
2681 includes dynamically acquired search domains, achieving more correct
2682 DNS resolution by software that bypasses local DNS APIs such as NSS.
2684 * The "uaccess" udev tag has been dropped from /dev/kvm and
2685 /dev/dri/renderD*. These devices now have the 0666 permissions by
2686 default (but this may be changed at build-time). /dev/dri/renderD*
2687 will now be owned by the "render" group along with /dev/kfd.
2689 * "DynamicUser=yes" has been enabled for systemd-timesyncd.service,
2690 systemd-journal-gatewayd.service and
2691 systemd-journal-upload.service. This means "nss-systemd" must be
2692 enabled in /etc/nsswitch.conf to ensure the UIDs assigned to these
2693 services are resolved properly.
2695 * In /etc/fstab two new mount options are now understood:
2696 x-systemd.makefs and x-systemd.growfs. The former has the effect that
2697 the configured file system is formatted before it is mounted, the
2698 latter that the file system is resized to the full block device size
2699 after it is mounted (i.e. if the file system is smaller than the
2700 partition it resides on, it's grown). This is similar to the fsck
2701 logic in /etc/fstab, and pulls in systemd-makefs@.service and
2702 systemd-growfs@.service as necessary, similar to
2703 systemd-fsck@.service. Resizing is currently only supported on ext4
2706 * In systemd-networkd, the IPv6 RA logic now optionally may announce
2707 DNS server and domain information.
2709 * Support for the LUKS2 on-disk format for encrypted partitions has
2710 been added. This requires libcryptsetup2 during compilation and
2713 * The systemd --user instance will now signal "readiness" when its
2714 basic.target unit has been reached, instead of when the run queue ran
2715 empty for the first time.
2717 * Tmpfiles.d with user configuration are now also supported.
2718 systemd-tmpfiles gained a new --user switch, and snippets placed in
2719 ~/.config/user-tmpfiles.d/ and corresponding directories will be
2720 executed by systemd-tmpfiles --user running in the new
2721 systemd-tmpfiles-setup.service and systemd-tmpfiles-clean.service
2722 running in the user session.
2724 * Unit files and tmpfiles.d snippets learnt three new % specifiers:
2725 %S resolves to the top-level state directory (/var/lib for the system
2726 instance, $XDG_CONFIG_HOME for the user instance), %C resolves to the
2727 top-level cache directory (/var/cache for the system instance,
2728 $XDG_CACHE_HOME for the user instance), %L resolves to the top-level
2729 logs directory (/var/log for the system instance,
2730 $XDG_CONFIG_HOME/log/ for the user instance). This matches the
2731 existing %t specifier, that resolves to the top-level runtime
2732 directory (/run for the system instance, and $XDG_RUNTIME_DIR for the
2735 * journalctl learnt a new parameter --output-fields= for limiting the
2736 set of journal fields to output in verbose and JSON output modes.
2738 * systemd-timesyncd's configuration file gained a new option
2739 RootDistanceMaxSec= for setting the maximum root distance of servers
2740 it'll use, as well as the new options PollIntervalMinSec= and
2741 PollIntervalMaxSec= to tweak the minimum and maximum poll interval.
2743 * bootctl gained a new command "list" for listing all available boot
2744 menu items on systems that follow the boot loader specification.
2746 * systemctl gained a new --dry-run switch that shows what would be done
2747 instead of doing it, and is currently supported by the shutdown and
2750 * ConditionSecurity= can now detect the TOMOYO security module.
2752 * Unit file [Install] sections are now also respected in unit drop-in
2753 files. This is intended to be used by drop-ins under /usr/lib/.
2755 * systemd-firstboot may now also set the initial keyboard mapping.
2757 * Udev "changed" events for devices which are exposed as systemd
2758 .device units are now propagated to units specified in
2759 ReloadPropagatedFrom= as reload requests.
2761 * If a udev device has a SYSTEMD_WANTS= property containing a systemd
2762 unit template name (i.e. a name in the form of 'foobar@.service',
2763 without the instance component between the '@' and - the '.'), then
2764 the escaped sysfs path of the device is automatically used as the
2767 * SystemCallFilter= in unit files has been extended so that an "errno"
2768 can be specified individually for each system call. Example:
2769 SystemCallFilter=~uname:EILSEQ.
2771 * The cgroup delegation logic has been substantially updated. Delegate=
2772 now optionally takes a list of controllers (instead of a boolean, as
2773 before), which lists the controllers to delegate at least.
2775 * The networkd DHCPv6 client now implements the FQDN option (RFC 4704).
2777 * A new LogLevelMax= setting configures the maximum log level any
2778 process of the service may log at (i.e. anything with a lesser
2779 priority than what is specified is automatically dropped). A new
2780 LogExtraFields= setting allows configuration of additional journal
2781 fields to attach to all log records generated by any of the unit's
2784 * New StandardInputData= and StandardInputText= settings along with the
2785 new option StandardInput=data may be used to configure textual or
2786 binary data that shall be passed to the executed service process via
2787 standard input, encoded in-line in the unit file.
2789 * StandardInput=, StandardOutput= and StandardError= may now be used to
2790 connect stdin/stdout/stderr of executed processes directly with a
2791 file or AF_UNIX socket in the file system, using the new "file:" option.
2793 * A new unit file option CollectMode= has been added, that allows
2794 tweaking the garbage collection logic for units. It may be used to
2795 tell systemd to garbage collect units that have failed automatically
2796 (normally it only GCs units that exited successfully). systemd-run
2797 and systemd-mount expose this new functionality with a new -G option.
2799 * "machinectl bind" may now be used to bind mount non-directories
2800 (i.e. regularfiles, devices, fifos, sockets).
2802 * systemd-analyze gained a new verb "calendar" for validating and
2803 testing calendar time specifications to use for OnCalendar= in timer
2804 units. Besides validating the expression it will calculate the next
2805 time the specified expression would elapse.
2807 * In addition to the pre-existing FailureAction= unit file setting
2808 there's now SuccessAction=, for configuring a shutdown action to
2809 execute when a unit completes successfully. This is useful in
2810 particular inside containers that shall terminate after some workload
2811 has been completed. Also, both options are now supported for all unit
2812 types, not just services.
2814 * networkds's IP rule support gained two new options
2815 IncomingInterface= and OutgoingInterface= for configuring the incoming
2816 and outgoing interfaces of configured rules. systemd-networkd also
2817 gained support for "vxcan" network devices.
2819 * networkd gained a new setting RequiredForOnline=, taking a
2820 boolean. If set, systemd-wait-online will take it into consideration
2821 when determining that the system is up, otherwise it will ignore the
2822 interface for this purpose.
2824 * The sd_notify() protocol gained support for a new operation: with
2825 FDSTOREREMOVE=1 file descriptors may be removed from the per-service
2826 store again, ahead of POLLHUP or POLLERR when they are removed
2829 * A new document doc/UIDS-GIDS.md has been added to the source tree,
2830 that documents the UID/GID range and assignment assumptions and
2831 requirements of systemd.
2833 * The watchdog device PID 1 will ping may now be configured through the
2834 WatchdogDevice= configuration file setting, or by setting the
2835 systemd.watchdog_service= kernel commandline option.
2837 * systemd-resolved's gained support for registering DNS-SD services on
2838 the local network using MulticastDNS. Services may either be
2839 registered by dropping in a .dnssd file in /etc/systemd/dnssd/ (or
2840 the same dir below /run, /usr/lib), or through its D-Bus API.
2842 * The sd_notify() protocol can now with EXTEND_TIMEOUT_USEC=microsecond
2843 extend the effective start, runtime, and stop time. The service must
2844 continue to send EXTEND_TIMEOUT_USEC within the period specified to
2845 prevent the service manager from making the service as timedout.
2847 * systemd-resolved's DNSSEC support gained support for RFC 8080
2848 (Ed25519 keys and signatures).
2850 * The systemd-resolve command line tool gained a new set of options
2851 --set-dns=, --set-domain=, --set-llmnr=, --set-mdns=, --set-dnssec=,
2852 --set-nta= and --revert to configure per-interface DNS configuration
2853 dynamically during runtime. It's useful for pushing DNS information
2854 into systemd-resolved from DNS hook scripts that various interface
2855 managing software supports (such as pppd).
2857 * systemd-nspawn gained a new --network-namespace-path= command line
2858 option, which may be used to make a container join an existing
2859 network namespace, by specifying a path to a "netns" file.
2861 Contributions from: Alan Jenkins, Alan Robertson, Alessandro Ghedini,
2862 Andrew Jeddeloh, Antonio Rojas, Ari, asavah, bleep_blop, Carsten
2863 Strotmann, Christian Brauner, Christian Hesse, Clinton Roy, Collin
2864 Eggert, Cong Wang, Daniel Black, Daniel Lockyer, Daniel Rusek, Dimitri
2865 John Ledkov, Dmitry Rozhkov, Dongsu Park, Edward A. James, Evgeny
2866 Vereshchagin, Florian Klink, Franck Bui, Gwendal Grignou, Hans de
2867 Goede, Harald Hoyer, Hristo Venev, Iago López Galeiras, Ikey Doherty,
2868 Jakub Wilk, Jérémy Rosen, Jiahui Xie, John Lin, José Bollo, Josef
2869 Andersson, juga0, Krzysztof Nowicki, Kyle Walker, Lars Karlitski, Lars
2870 Kellogg-Stedman, Lauri Tirkkonen, Lennart Poettering, Lubomir Rintel,
2871 Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn, Lukáš Říha, Lukasz
2872 Rubaszewski, Maciej S. Szmigiero, Mantas Mikulėnas, Marcus Folkesson,
2873 Martin Steuer, Mathieu Trudel-Lapierre, Matija Skala,
2874 Matthias-Christian Ott, Max Resch, Michael Biebl, Michael Vogt, Michal
2875 Koutný, Michal Sekletar, Mike Gilbert, Muhammet Kara, Neil Brown, Olaf
2876 Hering, Ondrej Kozina, Patrik Flykt, Patryk Kocielnik, Peter Hutterer,
2877 Piotr Drąg, Razvan Cojocaru, Robin McCorkell, Roland Hieber, Saran
2878 Tunyasuvunakool, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
2879 Arlott, Simon Peeters, Stanislav Angelovič, Stefan Agner, Susant
2880 Sahani, Sylvain Plantefève, Thomas Blume, Thomas Haller, Tiago Salem
2881 Herrmann, Tinu Weber, Tom Stellard, Topi Miettinen, Torsten Hilbrich,
2882 Vito Caputo, Vladislav Vishnyakov, WaLyong Cho, Yu Watanabe, Zbigniew
2883 Jędrzejewski-Szmek, Zeal Jagannatha
2885 — Berlin, 2017-12-14
2889 * INCOMPATIBILITY: systemd-logind.service and other long-running
2890 services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP
2891 communication with the outside. This generally improves security of
2892 the system, and is in almost all cases a safe and good choice, as
2893 these services do not and should not provide any network-facing
2894 functionality. However, systemd-logind uses the glibc NSS API to
2895 query the user database. This creates problems on systems where NSS
2896 is set up to directly consult network services for user database
2897 lookups. In particular, this creates incompatibilities with the
2898 "nss-nis" module, which attempts to directly contact the NIS/YP
2899 network servers it is configured for, and will now consistently
2900 fail. In such cases, it is possible to turn off IP sandboxing for
2901 systemd-logind.service (set IPAddressDeny= in its [Service] section
2902 to the empty string, via a .d/ unit file drop-in). Downstream
2903 distributions might want to update their nss-nis packaging to include
2904 such a drop-in snippet, accordingly, to hide this incompatibility
2905 from the user. Another option is to make use of glibc's nscd service
2906 to proxy such network requests through a privilege-separated, minimal
2907 local caching daemon, or to switch to more modern technologies such
2908 sssd, whose NSS hook-ups generally do not involve direct network
2909 access. In general, we think it's definitely time to question the
2910 implementation choices of nss-nis, i.e. whether it's a good idea
2911 today to embed a network-facing loadable module into all local
2912 processes that need to query the user database, including the most
2913 trivial and benign ones, such as "ls". For more details about
2914 IPAddressDeny= see below.
2916 * A new modprobe.d drop-in is now shipped by default that sets the
2917 bonding module option max_bonds=0. This overrides the kernel default,
2918 to avoid conflicts and ambiguity as to whether or not bond0 should be
2919 managed by systemd-networkd or not. This resolves multiple issues
2920 with bond0 properties not being applied, when bond0 is configured
2921 with systemd-networkd. Distributors may choose to not package this,
2922 however in that case users will be prevented from correctly managing
2923 bond0 interface using systemd-networkd.
2925 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
2926 which print the logging level and target of the system manager. They
2927 complement the existing "set-log-level" and "set-log-target" verbs
2928 used to change those values.
2930 * journald.conf gained a new boolean setting ReadKMsg= which defaults
2931 to on. If turned off kernel log messages will not be read by
2932 systemd-journald or included in the logs. It also gained a new
2933 setting LineMax= for configuring the maximum line length in
2934 STDOUT/STDERR log streams. The new default for this value is 48K, up
2935 from the previous hardcoded 2048.
2937 * A new unit setting RuntimeDirectoryPreserve= has been added, which
2938 allows more detailed control of what to do with a runtime directory
2939 configured with RuntimeDirectory= (i.e. a directory below /run or
2940 $XDG_RUNTIME_DIR) after a unit is stopped.
2942 * The RuntimeDirectory= setting for units gained support for creating
2943 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
2944 one top-level directory.
2946 * Units gained new options StateDirectory=, CacheDirectory=,
2947 LogsDirectory= and ConfigurationDirectory= which are closely related
2948 to RuntimeDirectory= but manage per-service directories below
2949 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
2950 possible to write unit files which when activated automatically gain
2951 properly owned service specific directories in these locations, thus
2952 making unit files self-contained and increasing compatibility with
2953 stateless systems and factory reset where /etc or /var are
2954 unpopulated at boot. Matching these new settings there's also
2955 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
2956 ConfigurationDirectoryMode= for configuring the access mode of these
2957 directories. These settings are particularly useful in combination
2958 with DynamicUser=yes as they provide secure, properly-owned,
2959 writable, and stateful locations for storage, excluded from the
2960 sandbox that such services live in otherwise.
2962 * Automake support has been removed from this release. systemd is now
2965 * systemd-journald will now aggressively cache client metadata during
2966 runtime, speeding up log write performance under pressure. This comes
2967 at a small price though: as much of the metadata is read
2968 asynchronously from /proc/ (and isn't implicitly attached to log
2969 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
2970 metadata stored alongside a log entry might be slightly
2971 out-of-date. Previously it could only be slightly newer than the log
2972 message. The time window is small however, and given that the kernel
2973 is unlikely to be improved anytime soon in this regard, this appears
2976 * nss-myhostname/systemd-resolved will now by default synthesize an
2977 A/AAAA resource record for the "_gateway" hostname, pointing to the
2978 current default IP gateway. Previously it did that for the "gateway"
2979 name, hampering adoption, as some distributions wanted to leave that
2980 host name open for local use. The old behaviour may still be
2981 requested at build time.
2983 * systemd-networkd's [Address] section in .network files gained a new
2984 Scope= setting for configuring the IP address scope. The [Network]
2985 section gained a new boolean setting ConfigureWithoutCarrier= that
2986 tells systemd-networkd to ignore link sensing when configuring the
2987 device. The [DHCP] section gained a new Anonymize= boolean option for
2988 turning on a number of options suggested in RFC 7844. A new
2989 [RoutingPolicyRule] section has been added for configuring the IP
2990 routing policy. The [Route] section has gained support for a new
2991 Type= setting which permits configuring
2992 blackhole/unreachable/prohibit routes.
2994 * The [VRF] section in .netdev files gained a new Table= setting for
2995 configuring the routing table to use. The [Tunnel] section gained a
2996 new Independent= boolean field for configuring tunnels independent of
2997 an underlying network interface. The [Bridge] section gained a new
2998 GroupForwardMask= option for configuration of propagation of link
2999 local frames between bridge ports.
3001 * The WakeOnLan= setting in .link files gained support for a number of
3002 new modes. A new TCP6SegmentationOffload= setting has been added for
3003 configuring TCP/IPv6 hardware segmentation offload.
3005 * The IPv6 RA sender implementation may now optionally send out RDNSS
3006 and RDNSSL records to supply DNS configuration to peers.
3008 * systemd-nspawn gained support for a new --system-call-filter= command
3009 line option for adding and removing entries in the default system
3010 call filter it applies. Moreover systemd-nspawn has been changed to
3011 implement a system call whitelist instead of a blacklist.
3013 * systemd-run gained support for a new --pipe command line option. If
3014 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
3015 are directly passed on to the activated transient service
3016 executable. This allows invoking arbitrary processes as systemd
3017 services (for example to take benefit of dependency management,
3018 accounting management, resource management or log management that is
3019 done automatically for services) — while still allowing them to be
3020 integrated in a classic UNIX shell pipeline.
3022 * When a service sends RELOAD=1 via sd_notify() and reload propagation
3023 using ReloadPropagationTo= is configured, a reload is now propagated
3024 to configured units. (Previously this was only done on explicitly
3025 requested reloads, using "systemctl reload" or an equivalent
3028 * For each service unit a restart counter is now kept: it is increased
3029 each time the service is restarted due to Restart=, and may be
3030 queried using "systemctl show -p NRestarts …".
3032 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
3033 @signal and @timer have been added, for usage with SystemCallFilter=
3034 in unit files and the new --system-call-filter= command line option
3035 of systemd-nspawn (see above).
3037 * ExecStart= lines in unit files gained two new modifiers: when a
3038 command line is prefixed with "!" the command will be executed as
3039 configured, except for the credentials applied by
3040 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
3041 "+", but does still apply namespacing options unlike "+". There's
3042 also "!!" now, which is mostly identical, but becomes a NOP on
3043 systems that support ambient capabilities. This is useful to write
3044 unit files that work with ambient capabilities where possible but
3045 automatically fall back to traditional privilege dropping mechanisms
3046 on systems where this is not supported.
3048 * ListenNetlink= settings in socket units now support RDMA netlink
3051 * A new unit file setting LockPersonality= has been added which permits
3052 locking down the chosen execution domain ("personality") of a service
3055 * A new special target "getty-pre.target" has been added, which is
3056 ordered before all text logins, and may be used to order services
3057 before textual logins acquire access to the console.
3059 * systemd will now attempt to load the virtio-rng.ko kernel module very
3060 early on if a VM environment supporting this is detected. This should
3061 improve entropy during early boot in virtualized environments.
3063 * A _netdev option is now supported in /etc/crypttab that operates in a
3064 similar way as the same option in /etc/fstab: it permits configuring
3065 encrypted devices that need to be ordered after the network is up.
3066 Following this logic, two new special targets
3067 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
3068 added that are to cryptsetup.target what remote-fs.target and
3069 remote-fs-pre.target are to local-fs.target.
3071 * Service units gained a new UnsetEnvironment= setting which permits
3072 unsetting specific environment variables for services that are
3073 normally passed to it (for example in order to mask out locale
3074 settings for specific services that can't deal with it).
3076 * Units acquired a new boolean option IPAccounting=. When turned on, IP
3077 traffic accounting (packet count as well as byte count) is done for
3078 the service, and shown as part of "systemctl status" or "systemd-run
3081 * Service units acquired two new options IPAddressAllow= and
3082 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
3083 for configuring a simple IP access control list for all sockets of
3084 the unit. These options are available also on .slice and .socket
3085 units, permitting flexible access list configuration for individual
3086 services as well as groups of services (as defined by a slice unit),
3087 including system-wide. Note that IP ACLs configured this way are
3088 enforced on every single IPv4 and IPv6 socket created by any process
3089 of the service unit, and apply to ingress as well as egress traffic.
3091 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
3092 structured log message is generated each time the unit is stopped,
3093 containing information about the consumed resources of this
3096 * A new setting KeyringMode= has been added to unit files, which may be
3097 used to control how the kernel keyring is set up for executed
3100 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
3101 "systemctl kexec" and "systemctl exit" are now always asynchronous in
3102 behaviour (that is: these commands return immediately after the
3103 operation was enqueued instead of waiting for the operation to
3104 complete). Previously, "systemctl poweroff" and "systemctl reboot"
3105 were asynchronous on systems using systemd-logind (i.e. almost
3106 always, and like they were on sysvinit), and the other three commands
3107 were unconditionally synchronous. With this release this is cleaned
3108 up, and callers will see the same asynchronous behaviour on all
3109 systems for all five operations.
3111 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
3114 * .timer units now accept calendar specifications in other timezones
3115 than UTC or the local timezone.
3117 * The tmpfiles snippet var.conf has been changed to create
3118 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
3119 the "utmp" group already, and it appears to be generally understood
3120 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
3121 databases. Previously this was implemented correctly for all these
3122 databases excepts btmp, which has been opened up like this now
3123 too. Note that while the other databases are world-readable
3124 (i.e. 0644), btmp is not and remains more restrictive.
3126 * The systemd-resolve tool gained a new --reset-server-features
3127 switch. When invoked like this systemd-resolved will forget
3128 everything it learnt about the features supported by the configured
3129 upstream DNS servers, and restarts the feature probing logic on the
3130 next resolver look-up for them at the highest feature level
3133 * The status dump systemd-resolved sends to the logs upon receiving
3134 SIGUSR1 now also includes information about all DNS servers it is
3135 configured to use, and the features levels it probed for them.
3137 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
3138 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
3139 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
3140 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
3141 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
3142 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
3143 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
3144 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
3145 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
3146 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
3147 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
3148 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
3149 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
3150 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
3151 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
3152 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
3153 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
3154 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
3155 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
3156 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
3158 — Berlin, 2017-10-06
3162 * Meson is now supported as build system in addition to Automake. It is
3163 our plan to remove Automake in one of our next releases, so that
3164 Meson becomes our exclusive build system. Hence, please start using
3165 the Meson build system in your downstream packaging. There's plenty
3166 of documentation around how to use Meson, the extremely brief
3169 ./autogen.sh && ./configure && make && sudo make install
3173 meson build && ninja -C build && sudo ninja -C build install
3175 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
3176 which permits configuring a timeout on the time a job is
3177 running. This is particularly useful for setting timeouts on jobs for
3180 * Unit files gained two new options ConditionUser= and ConditionGroup=
3181 for conditionalizing units based on the identity of the user/group
3182 running a systemd user instance.
3184 * systemd-networkd now understands a new FlowLabel= setting in the
3185 [VXLAN] section of .network files, as well as a Priority= in
3186 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
3187 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
3188 gained support for configuration of GENEVE links, and IPv6 address
3189 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
3191 * .link files now understand a new Port= setting.
3193 * systemd-networkd's DHCP support gained support for DHCP option 119
3194 (domain search list).
3196 * systemd-networkd gained support for serving IPv6 address ranges using
3197 the Router Advertisement protocol. The new .network configuration
3198 section [IPv6Prefix] may be used to configure the ranges to
3199 serve. This is implemented based on a new, minimal, native server
3200 implementation of RA.
3202 * journalctl's --output= switch gained support for a new parameter
3203 "short-iso-precise" for a mode where timestamps are shown as precise
3206 * systemd-udevd's "net_id" builtin may now generate stable network
3207 interface names from IBM PowerVM VIO devices as well as ACPI platform
3210 * MulticastDNS support in systemd-resolved may now be explicitly
3211 enabled/disabled using the new MulticastDNS= configuration file
3214 * systemd-resolved may now optionally use libidn2 instead of the libidn
3215 for processing internationalized domain names. Support for libidn2
3216 should be considered experimental and should not be enabled by
3219 * "machinectl pull-tar" and related call may now do verification of
3220 downloaded images using SUSE-style .sha256 checksum files in addition
3221 to the already existing support for validating using Ubuntu-style
3224 * sd-bus gained support for a new sd_bus_message_appendv() call which
3225 is va_list equivalent of sd_bus_message_append().
3227 * sd-boot gained support for validating images using SHIM/MOK.
3229 * The SMACK code learnt support for "onlycap".
3231 * systemd-mount --umount is now much smarter in figuring out how to
3232 properly unmount a device given its mount or device path.
3234 * The code to call libnss_dns as a fallback from libnss_resolve when
3235 the communication with systemd-resolved fails was removed. This
3236 fallback was redundant and interfered with the [!UNAVAIL=return]
3237 suffix. See nss-resolve(8) for the recommended configuration.
3239 * systemd-logind may now be restarted without losing state. It stores
3240 the file descriptors for devices it manages in the system manager
3241 using the FDSTORE= mechanism. Please note that further changes in
3242 other components may be required to make use of this (for example
3243 Xorg has code to listen for stops of systemd-logind and terminate
3244 itself when logind is stopped or restarted, in order to avoid using
3245 stale file descriptors for graphical devices, which is now
3246 counterproductive and must be reverted in order for restarts of
3247 systemd-logind to be safe. See
3248 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
3250 * All kernel-install plugins are called with the environment variable
3251 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
3252 /etc/machine-id. If the machine ID could not be determined,
3253 $KERNEL_INSTALL_MACHINE_ID will be empty. Plugins should not put
3254 anything in the entry directory (passed as the second argument) if
3255 $KERNEL_INSTALL_MACHINE_ID is empty. For backwards compatibility, a
3256 temporary directory is passed as the entry directory and removed
3257 after all the plugins exit.
3259 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
3260 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
3261 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
3262 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
3263 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
3264 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
3265 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
3266 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
3267 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
3268 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
3269 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
3270 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
3271 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
3272 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
3273 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
3274 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
3275 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
3276 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
3277 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
3278 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
3279 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
3280 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
3281 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
3282 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
3283 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
3284 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
3285 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
3286 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
3287 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
3290 — Berlin, 2017-07-12
3294 * The "hybrid" control group mode has been modified to improve
3295 compatibility with "legacy" cgroups-v1 setups. Specifically, the
3296 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
3297 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
3298 cgroups-v1 hierarchy), the only externally visible change being that
3299 the cgroups-v2 hierarchy is also mounted, to
3300 /sys/fs/cgroup/unified. This should provide a large degree of
3301 compatibility with "legacy" cgroups-v1, while taking benefit of the
3302 better management capabilities of cgroups-v2.
3304 * The default control group setup mode may be selected both a boot-time
3305 via a set of kernel command line parameters (specifically:
3306 systemd.unified_cgroup_hierarchy= and
3307 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
3308 default selected on the configure command line
3309 (--with-default-hierarchy=). The upstream default is "hybrid"
3310 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
3311 this will change in a future systemd version to be "unified" (pure
3312 cgroups-v2 mode). The third option for the compile time option is
3313 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
3314 distributions to default to "hybrid" mode for release distributions,
3315 starting with v233. We recommend "unified" for development
3316 distributions (specifically: distributions such as Fedora's rawhide)
3317 as that's where things are headed in the long run. Use "legacy" for
3318 greatest stability and compatibility only.
3320 * Note one current limitation of "unified" and "hybrid" control group
3321 setup modes: the kernel currently does not permit the systemd --user
3322 instance (i.e. unprivileged code) to migrate processes between two
3323 disconnected cgroup subtrees, even if both are managed and owned by
3324 the user. This effectively means "systemd-run --user --scope" doesn't
3325 work when invoked from outside of any "systemd --user" service or
3326 scope. Specifically, it is not supported from session scopes. We are
3327 working on fixing this in a future systemd version. (See #3388 for
3328 further details about this.)
3330 * DBus policy files are now installed into /usr rather than /etc. Make
3331 sure your system has dbus >= 1.9.18 running before upgrading to this
3332 version, or override the install path with --with-dbuspolicydir= .
3334 * All python scripts shipped with systemd (specifically: the various
3335 tests written in Python) now require Python 3.
3337 * systemd unit tests can now run standalone (without the source or
3338 build directories), and can be installed into /usr/lib/systemd/tests/
3339 with 'make install-tests'.
3341 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
3342 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
3345 * Support for the %c, %r, %R specifiers in unit files has been
3346 removed. Specifiers are not supposed to be dependent on configuration
3347 in the unit file itself (so that they resolve the same regardless
3348 where used in the unit files), but these specifiers were influenced
3349 by the Slice= option.
3351 * The shell invoked by debug-shell.service now defaults to /bin/sh in
3352 all cases. If distributions want to use a different shell for this
3353 purpose (for example Fedora's /sbin/sushell) they need to specify
3354 this explicitly at configure time using --with-debug-shell=.
3356 * The confirmation spawn prompt has been reworked to offer the
3359 (c)ontinue, proceed without asking anymore
3360 (D)ump, show the state of the unit
3361 (f)ail, don't execute the command and pretend it failed
3363 (i)nfo, show a short summary of the unit
3364 (j)obs, show jobs that are in progress
3365 (s)kip, don't execute the command and pretend it succeeded
3366 (y)es, execute the command
3368 The 'n' choice for the confirmation spawn prompt has been removed,
3369 because its meaning was confusing.
3371 The prompt may now also be redirected to an alternative console by
3372 specifying the console as parameter to systemd.confirm_spawn=.
3374 * Services of Type=notify require a READY=1 notification to be sent
3375 during startup. If no such message is sent, the service now fails,
3376 even if the main process exited with a successful exit code.
3378 * Services that fail to start up correctly now always have their
3379 ExecStopPost= commands executed. Previously, they'd enter "failed"
3380 state directly, without executing these commands.
3382 * The option MulticastDNS= of network configuration files has acquired
3383 an actual implementation. With MulticastDNS=yes a host can resolve
3384 names of remote hosts and reply to mDNS A and AAAA requests.
3386 * When units are about to be started an additional check is now done to
3387 ensure that all dependencies of type BindsTo= (when used in
3388 combination with After=) have been started.
3390 * systemd-analyze gained a new verb "syscall-filter" which shows which
3391 system call groups are defined for the SystemCallFilter= unit file
3392 setting, and which system calls they contain.
3394 * A new system call filter group "@filesystem" has been added,
3395 consisting of various file system related system calls. Group
3396 "@reboot" has been added, covering reboot, kexec and shutdown related
3397 calls. Finally, group "@swap" has been added covering swap
3398 configuration related calls.
3400 * A new unit file option RestrictNamespaces= has been added that may be
3401 used to restrict access to the various process namespace types the
3402 Linux kernel provides. Specifically, it may be used to take away the
3403 right for a service unit to create additional file system, network,
3404 user, and other namespaces. This sandboxing option is particularly
3405 relevant due to the high amount of recently discovered namespacing
3406 related vulnerabilities in the kernel.
3408 * systemd-udev's .link files gained support for a new AutoNegotiation=
3409 setting for configuring Ethernet auto-negotiation.
3411 * systemd-networkd's .network files gained support for a new
3412 ListenPort= setting in the [DHCP] section to explicitly configure the
3413 UDP client port the DHCP client shall listen on.
3415 * .network files gained a new Unmanaged= boolean setting for explicitly
3416 excluding one or more interfaces from management by systemd-networkd.
3418 * The systemd-networkd ProxyARP= option has been renamed to
3419 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
3420 renamed to ReduceARPProxy=. The old names continue to be available
3423 * systemd-networkd gained support for configuring IPv6 Proxy NDP
3424 addresses via the new IPv6ProxyNDPAddress= .network file setting.
3426 * systemd-networkd's bonding device support gained support for two new
3427 configuration options ActiveSlave= and PrimarySlave=.
3429 * The various options in the [Match] section of .network files gained
3430 support for negative matching.
3432 * New systemd-specific mount options are now understood in /etc/fstab:
3434 x-systemd.mount-timeout= may be used to configure the maximum
3435 permitted runtime of the mount command.
3437 x-systemd.device-bound may be set to bind a mount point to its
3438 backing device unit, in order to automatically remove a mount point
3439 if its backing device is unplugged. This option may also be
3440 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
3441 on the block device, which is now automatically set for all CDROM
3442 drives, so that mounted CDs are automatically unmounted when they are
3443 removed from the drive.
3445 x-systemd.after= and x-systemd.before= may be used to explicitly
3446 order a mount after or before another unit or mount point.
3448 * Enqueued start jobs for device units are now automatically garbage
3449 collected if there are no jobs waiting for them anymore.
3451 * systemctl list-jobs gained two new switches: with --after, for every
3452 queued job the jobs it's waiting for are shown; with --before the
3453 jobs which it's blocking are shown.
3455 * systemd-nspawn gained support for ephemeral boots from disk images
3456 (or in other words: --ephemeral and --image= may now be
3457 combined). Moreover, ephemeral boots are now supported for normal
3458 directories, even if the backing file system is not btrfs. Of course,
3459 if the file system does not support file system snapshots or
3460 reflinks, the initial copy operation will be relatively expensive, but
3461 this should still be suitable for many use cases.
3463 * Calendar time specifications in .timer units now support
3464 specifications relative to the end of a month by using "~" instead of
3465 "-" as separator between month and day. For example, "*-02~03" means
3466 "the third last day in February". In addition a new syntax for
3467 repeated events has been added using the "/" character. For example,
3468 "9..17/2:00" means "every two hours from 9am to 5pm".
3470 * systemd-socket-proxyd gained a new parameter --connections-max= for
3471 configuring the maximum number of concurrent connections.
3473 * sd-id128 gained a new API for generating unique IDs for the host in a
3474 way that does not leak the machine ID. Specifically,
3475 sd_id128_get_machine_app_specific() derives an ID based on the
3476 machine ID a in well-defined, non-reversible, stable way. This is
3477 useful whenever an identifier for the host is needed but where the
3478 identifier shall not be useful to identify the system beyond the
3479 scope of the application itself. (Internally this uses HMAC-SHA256 as
3480 keyed hash function using the machine ID as input.)
3482 * NotifyAccess= gained a new supported value "exec". When set
3483 notifications are accepted from all processes systemd itself invoked,
3484 including all control processes.
3486 * .nspawn files gained support for defining overlay mounts using the
3487 Overlay= and OverlayReadOnly= options. Previously this functionality
3488 was only available on the systemd-nspawn command line.
3490 * systemd-nspawn's --bind= and --overlay= options gained support for
3491 bind/overlay mounts whose source lies within the container tree by
3492 prefixing the source path with "+".
3494 * systemd-nspawn's --bind= and --overlay= options gained support for
3495 automatically allocating a temporary source directory in /var/tmp
3496 that is removed when the container dies. Specifically, if the source
3497 directory is specified as empty string this mechanism is selected. An
3498 example usage is --overlay=+/var::/var, which creates an overlay
3499 mount based on the original /var contained in the image, overlayed
3500 with a temporary directory in the host's /var/tmp. This way changes
3501 to /var are automatically flushed when the container shuts down.
3503 * systemd-nspawn --image= option does now permit raw file system block
3504 devices (in addition to images containing partition tables, as
3507 * The disk image dissection logic in systemd-nspawn gained support for
3508 automatically setting up LUKS encrypted as well as Verity protected
3509 partitions. When a container is booted from an encrypted image the
3510 passphrase is queried at start-up time. When a container with Verity
3511 data is started, the root hash is search in a ".roothash" file
3512 accompanying the disk image (alternatively, pass the root hash via
3513 the new --root-hash= command line option).
3515 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
3516 be used to dissect disk images the same way as systemd-nspawn does
3517 it, following the Bootable Partition Specification. It may even be
3518 used to mount disk images with complex partition setups (including
3519 LUKS and Verity partitions) to a local host directory, in order to
3520 inspect them. This tool is not considered public API (yet), and is
3521 thus not installed into /usr/bin. Please do not rely on its
3522 existence, since it might go away or be changed in later systemd
3525 * A new generator "systemd-verity-generator" has been added, similar in
3526 style to "systemd-cryptsetup-generator", permitting automatic setup of
3527 Verity root partitions when systemd boots up. In order to make use of
3528 this your partition setup should follow the Discoverable Partitions
3529 Specification, and the GPT partition ID of the root file system
3530 partition should be identical to the upper 128bit of the Verity root
3531 hash. The GPT partition ID of the Verity partition protecting it
3532 should be the lower 128bit of the Verity root hash. If the partition
3533 image follows this model it is sufficient to specify a single
3534 "roothash=" kernel command line argument to both configure which root
3535 image and verity partition to use as well as the root hash for
3536 it. Note that systemd-nspawn's Verity support follows the same
3537 semantics, meaning that disk images with proper Verity data in place
3538 may be booted in containers with systemd-nspawn as well as on
3539 physical systems via the verity generator. Also note that the "mkosi"
3540 tool available at https://github.com/systemd/mkosi has been updated
3541 to generate Verity protected disk images following this scheme. In
3542 fact, it has been updated to generate disk images that optionally
3543 implement a complete UEFI SecureBoot trust chain, involving a signed
3544 kernel and initrd image that incorporates such a root hash as well as
3545 a Verity-enabled root partition.
3547 * The hardware database (hwdb) udev supports has been updated to carry
3548 accelerometer quirks.
3550 * All system services are now run with a fresh kernel keyring set up
3551 for them. The invocation ID is stored by default in it, thus
3552 providing a safe, non-overridable way to determine the invocation
3555 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
3556 options for bind mounting arbitrary paths in a service-specific
3557 way. When these options are used, arbitrary host or service files and
3558 directories may be mounted to arbitrary locations in the service's
3561 * Documentation has been added that lists all of systemd's low-level
3562 environment variables:
3564 https://github.com/systemd/systemd/blob/master/docs/ENVIRONMENT.md
3566 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
3567 whether a specific socket file descriptor matches a specified socket
3570 * systemd-firstboot has been updated to check for the
3571 systemd.firstboot= kernel command line option. It accepts a boolean
3572 and when set to false the first boot questions are skipped.
3574 * systemd-fstab-generator has been updated to check for the
3575 systemd.volatile= kernel command line option, which either takes an
3576 optional boolean parameter or the special value "state". If used the
3577 system may be booted in a "volatile" boot mode. Specifically,
3578 "systemd.volatile" is used, the root directory will be mounted as
3579 tmpfs, and only /usr is mounted from the actual root file system. If
3580 "systemd.volatile=state" is used, the root directory will be mounted
3581 as usual, but /var is mounted as tmpfs. This concept provides similar
3582 functionality as systemd-nspawn's --volatile= option, but provides it
3583 on physical boots. Use this option for implementing stateless
3584 systems, or testing systems with all state and/or configuration reset
3585 to the defaults. (Note though that many distributions are not
3586 prepared to boot up without a populated /etc or /var, though.)
3588 * systemd-gpt-auto-generator gained support for LUKS encrypted root
3589 partitions. Previously it only supported LUKS encrypted partitions
3590 for all other uses, except for the root partition itself.
3592 * Socket units gained support for listening on AF_VSOCK sockets for
3593 communication in virtualized QEMU environments.
3595 * The "configure" script gained a new option --with-fallback-hostname=
3596 for specifying the fallback hostname to use if none is configured in
3597 /etc/hostname. For example, by specifying
3598 --with-fallback-hostname=fedora it is possible to default to a
3599 hostname of "fedora" on pristine installations.
3601 * systemd-cgls gained support for a new --unit= switch for listing only
3602 the control groups of a specific unit. Similar --user-unit= has been
3603 added for listing only the control groups of a specific user unit.
3605 * systemd-mount gained a new --umount switch for unmounting a mount or
3606 automount point (and all mount/automount points below it).
3608 * systemd will now refuse full configuration reloads (via systemctl
3609 daemon-reload and related calls) unless at least 16MiB of free space
3610 are available in /run. This is a safety precaution in order to ensure
3611 that generators can safely operate after the reload completed.
3613 * A new unit file option RootImage= has been added, which has a similar
3614 effect as RootDirectory= but mounts the service's root directory from
3615 a disk image instead of plain directory. This logic reuses the same
3616 image dissection and mount logic that systemd-nspawn already uses,
3617 and hence supports any disk images systemd-nspawn supports, including
3618 those following the Discoverable Partition Specification, as well as
3619 Verity enabled images. This option enables systemd to run system
3620 services directly off disk images acting as resource bundles,
3621 possibly even including full integrity data.
3623 * A new MountAPIVFS= unit file option has been added, taking a boolean
3624 argument. If enabled /proc, /sys and /dev (collectively called the
3625 "API VFS") will be mounted for the service. This is only relevant if
3626 RootDirectory= or RootImage= is used for the service, as these mounts
3627 are of course in place in the host mount namespace anyway.
3629 * systemd-nspawn gained support for a new --pivot-root= switch. If
3630 specified the root directory within the container image is pivoted to
3631 the specified mount point, while the original root disk is moved to a
3632 different place. This option enables booting of ostree images
3633 directly with systemd-nspawn.
3635 * The systemd build scripts will no longer complain if the NTP server
3636 addresses are not changed from the defaults. Google now supports
3637 these NTP servers officially. We still recommend downstreams to
3638 properly register an NTP pool with the NTP pool project though.
3640 * coredumpctl gained a new "--reverse" option for printing the list
3641 of coredumps in reverse order.
3643 * coredumpctl will now show additional information about truncated and
3644 inaccessible coredumps, as well as coredumps that are still being
3645 processed. It also gained a new --quiet switch for suppressing
3646 additional informational message in its output.
3648 * coredumpctl gained support for only showing coredumps newer and/or
3649 older than specific timestamps, using the new --since= and --until=
3650 options, reminiscent of journalctl's options by the same name.
3652 * The systemd-coredump logic has been improved so that it may be reused
3653 to collect backtraces in non-compiled languages, for example in
3654 scripting languages such as Python.
3656 * machinectl will now show the UID shift of local containers, if user
3657 namespacing is enabled for them.
3659 * systemd will now optionally run "environment generator" binaries at
3660 configuration load time. They may be used to add environment
3661 variables to the environment block passed to services invoked. One
3662 user environment generator is shipped by default that sets up
3663 environment variables based on files dropped into /etc/environment.d
3664 and ~/.config/environment.d/.
3666 * systemd-resolved now includes the new, recently published 2017 DNSSEC
3669 * hostnamed has been updated to report a new chassis type of
3670 "convertible" to cover "foldable" laptops that can both act as a
3671 tablet and as a laptop, such as various Lenovo Yoga devices.
3673 Contributions from: Adrián López, Alexander Galanin, Alexander
3674 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
3675 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
3676 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
3677 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
3678 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
3679 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
3680 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
3681 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
3682 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
3683 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
3684 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
3685 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
3686 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
3687 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
3688 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
3689 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
3690 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
3691 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
3692 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
3693 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
3694 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
3695 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
3696 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
3697 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
3698 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
3699 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
3700 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
3703 — Berlin, 2017-03-01
3707 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
3708 RestrictAddressFamilies= enabled. These sandboxing options should
3709 generally be compatible with the various external udev call-out
3710 binaries we are aware of, however there may be exceptions, in
3711 particular when exotic languages for these call-outs are used. In
3712 this case, consider turning off these settings locally.
3714 * The new RemoveIPC= option can be used to remove IPC objects owned by
3715 the user or group of a service when that service exits.
3717 * The new ProtectKernelModules= option can be used to disable explicit
3718 load and unload operations of kernel modules by a service. In
3719 addition access to /usr/lib/modules is removed if this option is set.
3721 * ProtectSystem= option gained a new value "strict", which causes the
3722 whole file system tree with the exception of /dev, /proc, and /sys,
3723 to be remounted read-only for a service.
3725 * The new ProtectKernelTunables= option can be used to disable
3726 modification of configuration files in /sys and /proc by a service.
3727 Various directories and files are remounted read-only, so access is
3728 restricted even if the file permissions would allow it.
3730 * The new ProtectControlGroups= option can be used to disable write
3731 access by a service to /sys/fs/cgroup.
3733 * Various systemd services have been hardened with
3734 ProtectKernelTunables=yes, ProtectControlGroups=yes,
3735 RestrictAddressFamilies=.
3737 * Support for dynamically creating users for the lifetime of a service
3738 has been added. If DynamicUser=yes is specified, user and group IDs
3739 will be allocated from the range 61184..65519 for the lifetime of the
3740 service. They can be resolved using the new nss-systemd.so NSS
3741 module. The module must be enabled in /etc/nsswitch.conf. Services
3742 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
3743 any resources allocated by the service will be cleaned up when the
3744 service exits. They also have ProtectHome=read-only and
3745 ProtectSystem=strict enabled, so they are not able to make any
3746 permanent modifications to the system.
3748 * The nss-systemd module also always resolves root and nobody, making
3749 it possible to have no /etc/passwd or /etc/group files in minimal
3750 container or chroot environments.
3752 * Services may be started with their own user namespace using the new
3753 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
3754 under which the service is running are mapped. All other users are
3757 * Support for the cgroup namespace has been added to systemd-nspawn. If
3758 supported by kernel, the container system started by systemd-nspawn
3759 will have its own view of the cgroup hierarchy. This new behaviour
3760 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
3762 * The new MemorySwapMax= option can be used to limit the maximum swap
3763 usage under the unified cgroup hierarchy.
3765 * Support for the CPU controller in the unified cgroup hierarchy has
3766 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
3767 options. This controller requires out-of-tree patches for the kernel
3768 and the support is provisional.
3770 * Mount and automount units may now be created transiently
3771 (i.e. dynamically at runtime via the bus API, instead of requiring
3772 unit files in the file system).
3774 * systemd-mount is a new tool which may mount file systems – much like
3775 mount(8), optionally pulling in additional dependencies through
3776 transient .mount and .automount units. For example, this tool
3777 automatically runs fsck on a backing block device before mounting,
3778 and allows the automount logic to be used dynamically from the
3779 command line for establishing mount points. This tool is particularly
3780 useful when dealing with removable media, as it will ensure fsck is
3781 run – if necessary – before the first access and that the file system
3782 is quickly unmounted after each access by utilizing the automount
3783 logic. This maximizes the chance that the file system on the
3784 removable media stays in a clean state, and if it isn't in a clean
3785 state is fixed automatically.
3787 * LazyUnmount=yes option for mount units has been added to expose the
3788 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
3791 * /efi will be used as the mount point of the EFI boot partition, if
3792 the directory is present, and the mount point was not configured
3793 through other means (e.g. fstab). If /efi directory does not exist,
3794 /boot will be used as before. This makes it easier to automatically
3795 mount the EFI partition on systems where /boot is used for something
3798 * When operating on GPT disk images for containers, systemd-nspawn will
3799 now mount the ESP to /boot or /efi according to the same rules as PID
3800 1 running on a host. This allows tools like "bootctl" to operate
3801 correctly within such containers, in order to make container images
3802 bootable on physical systems.
3804 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
3806 * Two new user session targets have been added to support running
3807 graphical sessions under the systemd --user instance:
3808 graphical-session.target and graphical-session-pre.target. See
3809 systemd.special(7) for a description of how those targets should be
3812 * The vconsole initialization code has been significantly reworked to
3813 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
3814 support unicode keymaps. Font and keymap configuration will now be
3815 copied to all allocated virtual consoles.
3817 * FreeBSD's bhyve virtualization is now detected.
3819 * Information recorded in the journal for core dumps now includes the
3820 contents of /proc/mountinfo and the command line of the process at
3821 the top of the process hierarchy (which is usually the init process
3824 * systemd-journal-gatewayd learned the --directory= option to serve
3825 files from the specified location.
3827 * journalctl --root=… can be used to peruse the journal in the
3828 /var/log/ directories inside of a container tree. This is similar to
3829 the existing --machine= option, but does not require the container to
3832 * The hardware database has been extended to support
3833 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
3836 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
3837 specify the click rate for mice which include a horizontal wheel with
3838 a click rate that is different than the one for the vertical wheel.
3840 * systemd-run gained a new --wait option that makes service execution
3841 synchronous. (Specifically, the command will not return until the
3842 specified service binary exited.)
3844 * systemctl gained a new --wait option that causes the start command to
3845 wait until the units being started have terminated again.
3847 * A new journal output mode "short-full" has been added which displays
3848 timestamps with abbreviated English day names and adds a timezone
3849 suffix. Those timestamps include more information than the default
3850 "short" output mode, and can be passed directly to journalctl's
3851 --since= and --until= options.
3853 * /etc/resolv.conf will be bind-mounted into containers started by
3854 systemd-nspawn, if possible, so any changes to resolv.conf contents
3855 are automatically propagated to the container.
3857 * The number of instances for socket-activated services originating
3858 from a single IP address can be limited with
3859 MaxConnectionsPerSource=, extending the existing setting of
3862 * systemd-networkd gained support for vcan ("Virtual CAN") interface
3865 * .netdev and .network configuration can now be extended through
3868 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
3869 Segmentation Offload, Generic Receive Offload, Large Receive Offload
3870 can be enabled and disabled using the new UDPSegmentationOffload=,
3871 TCPSegmentationOffload=, GenericSegmentationOffload=,
3872 GenericReceiveOffload=, LargeReceiveOffload= options in the
3873 [Link] section of .link files.
3875 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
3876 Port VLAN ID can be configured for bridge devices using the new STP=,
3877 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
3878 section of .netdev files.
3880 * The route table to which routes received over DHCP or RA should be
3881 added can be configured with the new RouteTable= option in the [DHCP]
3882 and [IPv6AcceptRA] sections of .network files.
3884 * The Address Resolution Protocol can be disabled on links managed by
3885 systemd-networkd using the ARP=no setting in the [Link] section of
3888 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
3889 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
3890 encode information about the result and exit codes of the current
3891 service runtime cycle.
3893 * systemd-sysctl will now configure kernel parameters in the order
3894 they occur in the configuration files. This matches what sysctl
3895 has been traditionally doing.
3897 * kernel-install "plugins" that are executed to perform various
3898 tasks after a new kernel is added and before an old one is removed
3899 can now return a special value to terminate the procedure and
3900 prevent any later plugins from running.
3902 * Journald's SplitMode=login setting has been deprecated. It has been
3903 removed from documentation, and its use is discouraged. In a future
3904 release it will be completely removed, and made equivalent to current
3905 default of SplitMode=uid.
3907 * Storage=both option setting in /etc/systemd/coredump.conf has been
3908 removed. With fast LZ4 compression storing the core dump twice is not
3911 * The --share-system systemd-nspawn option has been replaced with an
3912 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
3913 this functionality is discouraged. In addition the variables
3914 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
3915 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
3916 individual namespaces.
3918 * "machinectl list" now shows the IP address of running containers in
3919 the output, as well as OS release information.
3921 * "loginctl list" now shows the TTY of each session in the output.
3923 * sd-bus gained new API calls sd_bus_track_set_recursive(),
3924 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
3925 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
3926 tracking objects in a "recursive" mode, where a single client can be
3927 counted multiple times, if it takes multiple references.
3929 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
3930 sd_bus_get_exit_on_disconnect(). They may be used to make a
3931 process using sd-bus automatically exit if the bus connection is
3934 * Bus clients of the service manager may now "pin" loaded units into
3935 memory, by taking an explicit reference on them. This is useful to
3936 ensure the client can retrieve runtime data about the service even
3937 after the service completed execution. Taking such a reference is
3938 available only for privileged clients and should be helpful to watch
3939 running services in a race-free manner, and in particular collect
3940 information about exit statuses and results.
3942 * The nss-resolve module has been changed to strictly return UNAVAIL
3943 when communication via D-Bus with resolved failed, and NOTFOUND when
3944 a lookup completed but was negative. This means it is now possible to
3945 neatly configure fallbacks using nsswitch.conf result checking
3946 expressions. Taking benefit of this, the new recommended
3947 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
3949 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
3951 * A new setting CtrlAltDelBurstAction= has been added to
3952 /etc/systemd/system.conf which may be used to configure the precise
3953 behaviour if the user on the console presses Ctrl-Alt-Del more often
3954 than 7 times in 2s. Previously this would unconditionally result in
3955 an expedited, immediate reboot. With this new setting the precise
3956 operation may be configured in more detail, and also turned off
3959 * In .netdev files two new settings RemoteChecksumTx= and
3960 RemoteChecksumRx= are now understood that permit configuring the
3961 remote checksumming logic for VXLAN networks.
3963 * The service manager learnt a new "invocation ID" concept for invoked
3964 services. Each runtime cycle of a service will get a new invocation
3965 ID (a 128bit random UUID) assigned that identifies the current
3966 run of the service uniquely and globally. A new invocation ID
3967 is generated each time a service starts up. The journal will store
3968 the invocation ID of a service along with any logged messages, thus
3969 making the invocation ID useful for matching the online runtime of a
3970 service with the offline log data it generated in a safe way without
3971 relying on synchronized timestamps. In many ways this new service
3972 invocation ID concept is similar to the kernel's boot ID concept that
3973 uniquely and globally identifies the runtime of each boot. The
3974 invocation ID of a service is passed to the service itself via an
3975 environment variable ($INVOCATION_ID). A new bus call
3976 GetUnitByInvocationID() has been added that is similar to GetUnit()
3977 but instead of retrieving the bus path for a unit by its name
3978 retrieves it by its invocation ID. The returned path is valid only as
3979 long as the passed invocation ID is current.
3981 * systemd-resolved gained a new "DNSStubListener" setting in
3982 resolved.conf. It either takes a boolean value or the special values
3983 "udp" and "tcp", and configures whether to enable the stub DNS
3984 listener on 127.0.0.53:53.
3986 * IP addresses configured via networkd may now carry additional
3987 configuration settings supported by the kernel. New options include:
3988 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
3989 PrefixRoute=, AutoJoin=.
3991 * The PAM configuration fragment file for "user@.service" shipped with
3992 systemd (i.e. the --user instance of systemd) has been stripped to
3993 the minimum necessary to make the system boot. Previously, it
3994 contained Fedora-specific stanzas that did not apply to other
3995 distributions. It is expected that downstream distributions add
3996 additional configuration lines, matching their needs to this file,
3997 using it only as rough template of what systemd itself needs. Note
3998 that this reduced fragment does not even include an invocation of
3999 pam_limits which most distributions probably want to add, even though
4000 systemd itself does not need it. (There's also the new build time
4001 option --with-pamconfdir=no to disable installation of the PAM
4004 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
4005 capability is now also dropped from its set (in addition to
4006 CAP_SYS_MKNOD as before).
4008 * In service unit files it is now possible to connect a specific named
4009 file descriptor with stdin/stdout/stdout of an executed service. The
4010 name may be specified in matching .socket units using the
4011 FileDescriptorName= setting.
4013 * A number of journal settings may now be configured on the kernel
4014 command line. Specifically, the following options are now understood:
4015 systemd.journald.max_level_console=,
4016 systemd.journald.max_level_store=,
4017 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
4018 systemd.journald.max_level_wall=.
4020 * "systemctl is-enabled --full" will now show by which symlinks a unit
4021 file is enabled in the unit dependency tree.
4023 * Support for VeraCrypt encrypted partitions has been added to the
4024 "cryptsetup" logic and /etc/crypttab.
4026 * systemd-detect-virt gained support for a new --private-users switch
4027 that checks whether the invoking processes are running inside a user
4028 namespace. Similar, a new special value "private-users" for the
4029 existing ConditionVirtualization= setting has been added, permitting
4030 skipping of specific units in user namespace environments.
4032 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
4033 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
4034 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
4035 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
4036 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
4037 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
4038 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
4039 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
4040 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
4041 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
4042 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
4043 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
4044 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
4045 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
4046 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
4047 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
4048 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
4049 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
4050 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
4051 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
4052 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
4053 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
4054 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
4055 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
4056 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
4057 Jędrzejewski-Szmek, Zeal Jagannatha
4059 — Santa Fe, 2016-11-03
4063 * In service units the various ExecXYZ= settings have been extended
4064 with an additional special character as first argument of the
4065 assigned value: if the character '+' is used the specified command
4066 line it will be run with full privileges, regardless of User=,
4067 Group=, CapabilityBoundingSet= and similar options. The effect is
4068 similar to the existing PermissionsStartOnly= option, but allows
4069 configuration of this concept for each executed command line
4072 * Services may now alter the service watchdog timeout at runtime by
4073 sending a WATCHDOG_USEC= message via sd_notify().
4075 * MemoryLimit= and related unit settings now optionally take percentage
4076 specifications. The percentage is taken relative to the amount of
4077 physical memory in the system (or in case of containers, the assigned
4078 amount of memory). This allows scaling service resources neatly with
4079 the amount of RAM available on the system. Similarly, systemd-logind's
4080 RuntimeDirectorySize= option now also optionally takes percentage
4083 * In similar fashion TasksMax= takes percentage values now, too. The
4084 value is taken relative to the configured maximum number of processes
4085 on the system. The per-service task maximum has been changed to 15%
4086 using this functionality. (Effectively this is an increase of 512 →
4087 4915 for service units, given the kernel's default pid_max setting.)
4089 * Calendar time specifications in .timer units now understand a ".."
4090 syntax for time ranges. Example: "4..7:10" may now be used for
4091 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
4094 * The InaccessableDirectories=, ReadOnlyDirectories= and
4095 ReadWriteDirectories= unit file settings have been renamed to
4096 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
4097 applied to all kinds of file nodes, and not just directories, with
4098 the exception of symlinks. Specifically these settings may now be
4099 used on block and character device nodes, UNIX sockets and FIFOS as
4100 well as regular files. The old names of these settings remain
4101 available for compatibility.
4103 * systemd will now log about all service processes it kills forcibly
4104 (using SIGKILL) because they remained after the clean shutdown phase
4105 of the service completed. This should help identifying services that
4106 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
4107 systemd-logind's configuration a similar log message is generated for
4108 processes killed at the end of each session due to this setting.
4110 * systemd will now set the $JOURNAL_STREAM environment variable for all
4111 services whose stdout/stderr are connected to the Journal (which
4112 effectively means by default: all services). The variable contains
4113 the device and inode number of the file descriptor used for
4114 stdout/stderr. This may be used by invoked programs to detect whether
4115 their stdout/stderr is connected to the Journal, in which case they
4116 can switch over to direct Journal communication, thus being able to
4117 pass extended, structured metadata along with their log messages. As
4118 one example, this is now used by glib's logging primitives.
4120 * When using systemd's default tmp.mount unit for /tmp, the mount point
4121 will now be established with the "nosuid" and "nodev" options. This
4122 avoids privilege escalation attacks that put traps and exploits into
4123 /tmp. However, this might cause problems if you e. g. put container
4124 images or overlays into /tmp; if you need this, override tmp.mount's
4125 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
4128 * systemd now supports the "memory" cgroup controller also on
4131 * The systemd-cgtop tool now optionally takes a control group path as
4132 command line argument. If specified, the control group list shown is
4133 limited to subgroups of that group.
4135 * The SystemCallFilter= unit file setting gained support for
4136 pre-defined, named system call filter sets. For example
4137 SystemCallFilter=@clock is now an effective way to make all clock
4138 changing-related system calls unavailable to a service. A number of
4139 similar pre-defined groups are defined. Writing system call filters
4140 for system services is simplified substantially with this new
4141 concept. Accordingly, all of systemd's own, long-running services now
4142 enable system call filtering based on this, by default.
4144 * A new service setting MemoryDenyWriteExecute= has been added, taking
4145 a boolean value. If turned on, a service may no longer create memory
4146 mappings that are writable and executable at the same time. This
4147 enhances security for services where this is enabled as it becomes
4148 harder to dynamically write and then execute memory in exploited
4149 service processes. This option has been enabled for all of systemd's
4150 own long-running services.
4152 * A new RestrictRealtime= service setting has been added, taking a
4153 boolean argument. If set the service's processes may no longer
4154 acquire realtime scheduling. This improves security as realtime
4155 scheduling may otherwise be used to easily freeze the system.
4157 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
4158 value. This may be used for requesting that the system manager inside
4159 of the container reports start-up completion to nspawn which then
4160 propagates this notification further to the service manager
4161 supervising nspawn itself. A related option NotifyReady= in .nspawn
4162 files has been added too. This functionality allows ordering of the
4163 start-up of multiple containers using the usual systemd ordering
4166 * machinectl gained a new command "stop" that is an alias for
4169 * systemd-resolved gained support for contacting DNS servers on
4170 link-local IPv6 addresses.
4172 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
4173 its caches. A method call for requesting the same operation has been
4174 added to the bus API too, and is made available via "systemd-resolve
4177 * systemd-resolve gained a new --status switch. If passed a brief
4178 summary of the used DNS configuration with per-interface information
4181 * resolved.conf gained a new Cache= boolean option, defaulting to
4182 on. If turned off local DNS caching is disabled. This comes with a
4183 performance penalty in particular when DNSSEC is enabled. Note that
4184 resolved disables its internal caching implicitly anyway, when the
4185 configured DNS server is on a host-local IP address such as ::1 or
4186 127.0.0.1, thus automatically avoiding double local caching.
4188 * systemd-resolved now listens on the local IP address 127.0.0.53:53
4189 for DNS requests. This improves compatibility with local programs
4190 that do not use the libc NSS or systemd-resolved's bus APIs for name
4191 resolution. This minimal DNS service is only available to local
4192 programs and does not implement the full DNS protocol, but enough to
4193 cover local DNS clients. A new, static resolv.conf file, listing just
4194 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
4195 now recommended to make /etc/resolv.conf a symlink to this file in
4196 order to route all DNS lookups to systemd-resolved, regardless if
4197 done via NSS, the bus API or raw DNS packets. Note that this local
4198 DNS service is not as fully featured as the libc NSS or
4199 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
4200 used to deliver link-local address information (as this implies
4201 sending a local interface index along), LLMNR/mDNS support via this
4202 interface is severely restricted. It is thus strongly recommended for
4203 all applications to use the libc NSS API or native systemd-resolved
4206 * systemd-networkd's bridge support learned a new setting
4207 VLANFiltering= for controlling VLAN filtering. Moreover a new section
4208 in .network files has been added for configuring VLAN bridging in
4209 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
4211 * systemd-networkd's IPv6 Router Advertisement code now makes use of
4212 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
4213 now be acquired without relying on DHCPv6. Two new options
4214 UseDomains= and UseDNS= have been added to configure this behaviour.
4216 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
4217 renamed IPv6AcceptRA=, without altering its behaviour. The old
4218 setting name remains available for compatibility reasons.
4220 * The systemd-networkd VTI/VTI6 tunneling support gained new options
4221 Key=, InputKey= and OutputKey=.
4223 * systemd-networkd gained support for VRF ("Virtual Routing Function")
4224 interface configuration.
4226 * "systemctl edit" may now be used to create new unit files by
4227 specifying the --force switch.
4229 * sd-event gained a new function sd_event_get_iteration() for
4230 requesting the current iteration counter of the event loop. It starts
4231 at zero and is increased by one with each event loop iteration.
4233 * A new rpm macro %systemd_ordering is provided by the macros.systemd
4234 file. It can be used in lieu of %systemd_requires in packages which
4235 don't use any systemd functionality and are intended to be installed
4236 in minimal containers without systemd present. This macro provides
4237 ordering dependencies to ensure that if the package is installed in
4238 the same rpm transaction as systemd, systemd will be installed before
4239 the scriptlets for the package are executed, allowing unit presets
4242 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
4243 been added to simplify packaging of generators.
4245 * The os-release file gained VERSION_CODENAME field for the
4246 distribution nickname (e.g. VERSION_CODENAME=woody).
4248 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
4249 can be set to disable parsing of metadata and the creation
4250 of persistent symlinks for that device.
4252 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
4253 to make them available to logged-in users has been reverted.
4255 * Much of the common code of the various systemd components is now
4256 built into an internal shared library libsystemd-shared-231.so
4257 (incorporating the systemd version number in the name, to be updated
4258 with future releases) that the components link to. This should
4259 decrease systemd footprint both in memory during runtime and on
4260 disk. Note that the shared library is not for public use, and is
4261 neither API nor ABI stable, but is likely to change with every new
4262 released update. Packagers need to make sure that binaries
4263 linking to libsystemd-shared.so are updated in step with the
4266 * Configuration for "mkosi" is now part of the systemd
4267 repository. mkosi is a tool to easily build legacy-free OS images,
4268 and is available on github: https://github.com/systemd/mkosi. If
4269 "mkosi" is invoked in the build tree a new raw OS image is generated
4270 incorporating the systemd sources currently being worked on and a
4271 clean, fresh distribution installation. The generated OS image may be
4272 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
4273 UEFI PC. This functionality is particularly useful to easily test
4274 local changes made to systemd in a pristine, defined environment. See
4275 doc/HACKING for details.
4277 * configure learned the --with-support-url= option to specify the
4278 distribution's bugtracker.
4280 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
4281 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
4282 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
4283 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
4284 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
4285 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
4286 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
4287 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
4288 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
4289 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
4290 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
4291 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
4292 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
4293 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
4294 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
4295 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
4296 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
4297 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
4298 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
4300 — Berlin, 2016-07-25
4304 * DNSSEC is now turned on by default in systemd-resolved (in
4305 "allow-downgrade" mode), but may be turned off during compile time by
4306 passing "--with-default-dnssec=no" to "configure" (and of course,
4307 during runtime with DNSSEC= in resolved.conf). We recommend
4308 downstreams to leave this on at least during development cycles and
4309 report any issues with the DNSSEC logic upstream. We are very
4310 interested in collecting feedback about the DNSSEC validator and its
4311 limitations in the wild. Note however, that DNSSEC support is
4312 probably nothing downstreams should turn on in stable distros just
4313 yet, as it might create incompatibilities with a few DNS servers and
4314 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
4315 automatically whenever we detect such incompatible setups, but there
4316 might be systems we do not cover yet. Hence: please help us testing
4317 the DNSSEC code, leave this on where you can, report back, but then
4318 again don't consider turning this on in your stable, LTS or
4319 production release just yet. (Note that you have to enable
4320 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
4321 and its DNSSEC mode for host name resolution from local
4324 * systemd-resolve conveniently resolves DANE records with the --tlsa
4325 option and OPENPGPKEY records with the --openpgp option. It also
4326 supports dumping raw DNS record data via the new --raw= switch.
4328 * systemd-logind will now by default terminate user processes that are
4329 part of the user session scope unit (session-XX.scope) when the user
4330 logs out. This behavior is controlled by the KillUserProcesses=
4331 setting in logind.conf, and the previous default of "no" is now
4332 changed to "yes". This means that user sessions will be properly
4333 cleaned up after, but additional steps are necessary to allow
4334 intentionally long-running processes to survive logout.
4336 While the user is logged in at least once, user@.service is running,
4337 and any service that should survive the end of any individual login
4338 session can be started at a user service or scope using systemd-run.
4339 systemd-run(1) man page has been extended with an example which shows
4340 how to run screen in a scope unit underneath user@.service. The same
4341 command works for tmux.
4343 After the user logs out of all sessions, user@.service will be
4344 terminated too, by default, unless the user has "lingering" enabled.
4345 To effectively allow users to run long-term tasks even if they are
4346 logged out, lingering must be enabled for them. See loginctl(1) for
4347 details. The default polkit policy was modified to allow users to
4348 set lingering for themselves without authentication.
4350 Previous defaults can be restored at compile time by the
4351 --without-kill-user-processes option to "configure".
4353 * systemd-logind gained new configuration settings SessionsMax= and
4354 InhibitorsMax=, both with a default of 8192. It will not register new
4355 user sessions or inhibitors above this limit.
4357 * systemd-logind will now reload configuration on SIGHUP.
4359 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
4360 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
4361 enable. Also, support for the "io" cgroup controller in the unified
4362 hierarchy has been added, so that the "memory", "pids" and "io" are
4363 now the controllers that are supported on the unified hierarchy.
4365 WARNING: it is not possible to use previous systemd versions with
4366 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
4367 is necessary to also update systemd in the initramfs if using the
4368 unified hierarchy. An updated SELinux policy is also required.
4370 * LLDP support has been extended, and both passive (receive-only) and
4371 active (sender) modes are supported. Passive mode ("routers-only") is
4372 enabled by default in systemd-networkd. Active LLDP mode is enabled
4373 by default for containers on the internal network. The "networkctl
4374 lldp" command may be used to list information gathered. "networkctl
4375 status" will also show basic LLDP information on connected peers now.
4377 * The IAID and DUID unique identifier sent in DHCP requests may now be
4378 configured for the system and each .network file managed by
4379 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
4381 * systemd-networkd gained support for configuring proxy ARP support for
4382 each interface, via the ProxyArp= setting in .network files. It also
4383 gained support for configuring the multicast querier feature of
4384 bridge devices, via the new MulticastQuerier= setting in .netdev
4385 files. Similarly, snooping on the IGMP traffic can be controlled
4386 via the new setting MulticastSnooping=.
4388 A new setting PreferredLifetime= has been added for addresses
4389 configured in .network file to configure the lifetime intended for an
4392 The systemd-networkd DHCP server gained the option EmitRouter=, which
4393 defaults to yes, to configure whether the DHCP Option 3 (Router)
4396 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
4397 systemd-socket-activate and installed into /usr/bin. It is now fully
4400 * systemd-journald now uses separate threads to flush changes to disk
4401 when closing journal files, thus reducing impact of slow disk I/O on
4402 logging performance.
4404 * The sd-journal API gained two new calls
4405 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
4406 can be used to open journal files using file descriptors instead of
4407 file or directory paths. sd_journal_open_container() has been
4408 deprecated, sd_journal_open_directory_fd() should be used instead
4409 with the flag SD_JOURNAL_OS_ROOT.
4411 * journalctl learned a new output mode "-o short-unix" that outputs log
4412 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
4413 UTC). It also gained support for a new --no-hostname setting to
4414 suppress the hostname column in the family of "short" output modes.
4416 * systemd-ask-password now optionally skips printing of the password to
4417 stdout with --no-output which can be useful in scripts.
4419 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
4420 (devices tagged with ID_MAKER_TOOL) are now tagged with
4421 "uaccess" and are available to logged in users.
4423 * The DeviceAllow= unit setting now supports specifiers (with "%").
4425 * "systemctl show" gained a new --value switch, which allows print a
4426 only the contents of a specific unit property, without also printing
4427 the property's name. Similar support was added to "show*" verbs
4428 of loginctl and machinectl that output "key=value" lists.
4430 * A new unit type "generated" was added for files dynamically generated
4431 by generator tools. Similarly, a new unit type "transient" is used
4432 for unit files created using the runtime API. "systemctl enable" will
4433 refuse to operate on such files.
4435 * A new command "systemctl revert" has been added that may be used to
4436 revert to the vendor version of a unit file, in case local changes
4437 have been made by adding drop-ins or overriding the unit file.
4439 * "machinectl clean" gained a new verb to automatically remove all or
4440 just hidden container images.
4442 * systemd-tmpfiles gained support for a new line type "e" for emptying
4443 directories, if they exist, without creating them if they don't.
4445 * systemd-nspawn gained support for automatically patching the UID/GIDs
4446 of the owners and the ACLs of all files and directories in a
4447 container tree to match the UID/GID user namespacing range selected
4448 for the container invocation. This mode is enabled via the new
4449 --private-users-chown switch. It also gained support for
4450 automatically choosing a free, previously unused UID/GID range when
4451 starting a container, via the new --private-users=pick setting (which
4452 implies --private-users-chown). Together, these options for the first
4453 time make user namespacing for nspawn containers fully automatic and
4454 thus deployable. The systemd-nspawn@.service template unit file has
4455 been changed to use this functionality by default.
4457 * systemd-nspawn gained a new --network-zone= switch, that allows
4458 creating ad-hoc virtual Ethernet links between multiple containers,
4459 that only exist as long as at least one container referencing them is
4460 running. This allows easy connecting of multiple containers with a
4461 common link that implements an Ethernet broadcast domain. Each of
4462 these network "zones" may be named relatively freely by the user, and
4463 may be referenced by any number of containers, but each container may
4464 only reference one of these "zones". On the lower level, this is
4465 implemented by an automatically managed bridge network interface for
4466 each zone, that is created when the first container referencing its
4467 zone is created and removed when the last one referencing its zone
4470 * The default start timeout may now be configured on the kernel command
4471 line via systemd.default_timeout_start_sec=. It was already
4472 configurable via the DefaultTimeoutStartSec= option in
4473 /etc/systemd/system.conf.
4475 * Socket units gained a new TriggerLimitIntervalSec= and
4476 TriggerLimitBurst= setting to configure a limit on the activation
4477 rate of the socket unit.
4479 * The LimitNICE= setting now optionally takes normal UNIX nice values
4480 in addition to the raw integer limit value. If the specified
4481 parameter is prefixed with "+" or "-" and is in the range -20..19 the
4482 value is understood as UNIX nice value. If not prefixed like this it
4483 is understood as raw RLIMIT_NICE limit.
4485 * Note that the effect of the PrivateDevices= unit file setting changed
4486 slightly with this release: the per-device /dev file system will be
4487 mounted read-only from this version on, and will have "noexec"
4488 set. This (minor) change of behavior might cause some (exceptional)
4489 legacy software to break, when PrivateDevices=yes is set for its
4490 service. Please leave PrivateDevices= off if you run into problems
4493 * systemd-bootchart has been split out to a separate repository:
4494 https://github.com/systemd/systemd-bootchart
4496 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
4497 merged into the kernel in its current form.
4499 * The compatibility libraries libsystemd-daemon.so,
4500 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
4501 which have been deprecated since systemd-209 have been removed along
4502 with the corresponding pkg-config files. All symbols provided by
4503 those libraries are provided by libsystemd.so.
4505 * The Capabilities= unit file setting has been removed (it is ignored
4506 for backwards compatibility). AmbientCapabilities= and
4507 CapabilityBoundingSet= should be used instead.
4509 * A new special target has been added, initrd-root-device.target,
4510 which creates a synchronization point for dependencies of the root
4511 device in early userspace. Initramfs builders must ensure that this
4512 target is now included in early userspace.
4514 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
4515 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
4516 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
4517 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
4518 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
4519 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
4520 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
4521 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
4522 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
4523 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
4524 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
4525 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
4526 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
4527 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
4528 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
4529 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
4530 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
4531 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
4532 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
4533 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
4534 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
4535 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
4536 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
4537 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
4538 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
4541 — Fairfax, 2016-05-21
4545 * The systemd-resolved DNS resolver service has gained a substantial
4546 set of new features, most prominently it may now act as a DNSSEC
4547 validating stub resolver. DNSSEC mode is currently turned off by
4548 default, but is expected to be turned on by default in one of the
4549 next releases. For now, we invite everybody to test the DNSSEC logic
4550 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
4551 service also gained a full set of D-Bus interfaces, including calls
4552 to configure DNS and DNSSEC settings per link (for use by external
4553 network management software). systemd-resolved and systemd-networkd
4554 now distinguish between "search" and "routing" domains. The former
4555 are used to qualify single-label names, the latter are used purely
4556 for routing lookups within certain domains to specific links.
4557 resolved now also synthesizes RRs for all entries from /etc/hosts.
4559 * The systemd-resolve tool (which is a client utility for
4560 systemd-resolved) has been improved considerably and is now fully
4561 supported and documented. Hence it has moved from /usr/lib/systemd to
4564 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
4567 * The coredump collection logic has been reworked: when a coredump is
4568 collected it is now written to disk, compressed and processed
4569 (including stacktrace extraction) from a new instantiated service
4570 systemd-coredump@.service, instead of directly from the
4571 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
4572 processing large coredumps can take up a substantial amount of
4573 resources and time, and this previously happened entirely outside of
4574 systemd's service supervision. With the new logic the core_pattern
4575 hook only does minimal metadata collection before passing off control
4576 to the new instantiated service, which is configured with a time
4577 limit, a nice level and other settings to minimize negative impact on
4578 the rest of the system. Also note that the new logic will honour the
4579 RLIMIT_CORE setting of the crashed process, which now allows users
4580 and processes to turn off coredumping for their processes by setting
4583 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
4584 and all forked processes by default. Previously, PID 1 would leave
4585 the setting at "0" for all processes, as set by the kernel. Note that
4586 the resource limit traditionally has no effect on the generated
4587 coredumps on the system if the /proc/sys/kernel/core_pattern hook
4588 logic is used. Since the limit is now honoured (see above) its
4589 default has been changed so that the coredumping logic is enabled by
4590 default for all processes, while allowing specific opt-out.
4592 * When the stacktrace is extracted from processes of system users, this
4593 is now done as "systemd-coredump" user, in order to sandbox this
4594 potentially security sensitive parsing operation. (Note that when
4595 processing coredumps of normal users this is done under the user ID
4596 of process that crashed, as before.) Packagers should take notice
4597 that it is now necessary to create the "systemd-coredump" system user
4598 and group at package installation time.
4600 * The systemd-activate socket activation testing tool gained support
4601 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
4602 and --seqpacket switches. It also has been extended to support both
4603 new-style and inetd-style file descriptor passing. Use the new
4604 --inetd switch to request inetd-style file descriptor passing.
4606 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
4607 variable, which takes a boolean value. If set to false, ANSI color
4608 output is disabled in the tools even when run on a terminal that
4611 * The VXLAN support in networkd now supports two new settings
4612 DestinationPort= and PortRange=.
4614 * A new systemd.machine_id= kernel command line switch has been added,
4615 that may be used to set the machine ID in /etc/machine-id if it is
4616 not initialized yet. This command line option has no effect if the
4617 file is already initialized.
4619 * systemd-nspawn gained a new --as-pid2 switch that invokes any
4620 specified command line as PID 2 rather than PID 1 in the
4621 container. In this mode PID 1 is a minimal stub init process that
4622 implements the special POSIX and Linux semantics of PID 1 regarding
4623 signal and child process management. Note that this stub init process
4624 is implemented in nspawn itself and requires no support from the
4625 container image. This new logic is useful to support running
4626 arbitrary commands in the container, as normal processes are
4627 generally not prepared to run as PID 1.
4629 * systemd-nspawn gained a new --chdir= switch for setting the current
4630 working directory for the process started in the container.
4632 * "journalctl /dev/sda" will now output all kernel log messages for
4633 specified device from the current boot, in addition to all devices
4634 that are parents of it. This should make log output about devices
4635 pretty useful, as long as kernel drivers attach enough metadata to
4636 the log messages. (The usual SATA drivers do.)
4638 * The sd-journal API gained two new calls
4639 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
4640 that report whether log data from /run or /var has been found.
4642 * journalctl gained a new switch "--fields" that prints all journal
4643 record field names currently in use in the journal. This is backed
4644 by two new sd-journal API calls sd_journal_enumerate_fields() and
4645 sd_journal_restart_fields().
4647 * Most configurable timeouts in systemd now expect an argument of
4648 "infinity" to turn them off, instead of "0" as before. The semantics
4649 from now on is that a timeout of "0" means "now", and "infinity"
4650 means "never". To maintain backwards compatibility, "0" continues to
4651 turn off previously existing timeout settings.
4653 * "systemctl reload-or-try-restart" has been renamed to "systemctl
4654 try-reload-or-restart" to clarify what it actually does: the "try"
4655 logic applies to both reloading and restarting, not just restarting.
4656 The old name continues to be accepted for compatibility.
4658 * On boot-up, when PID 1 detects that the system clock is behind the
4659 release date of the systemd version in use, the clock is now set
4660 to the latter. Previously, this was already done in timesyncd, in order
4661 to avoid running with clocks set to the various clock epochs such as
4662 1902, 1938 or 1970. With this change the logic is now done in PID 1
4663 in addition to timesyncd during early boot-up, so that it is enforced
4664 before the first process is spawned by systemd. Note that the logic
4665 in timesyncd remains, as it is more comprehensive and ensures
4666 clock monotonicity by maintaining a persistent timestamp file in
4667 /var. Since /var is generally not available in earliest boot or the
4668 initrd, this part of the logic remains in timesyncd, and is not done
4671 * Support for tweaking details in net_cls.class_id through the
4672 NetClass= configuration directive has been removed, as the kernel
4673 people have decided to deprecate that controller in cgroup v2.
4674 Userspace tools such as nftables are moving over to setting rules
4675 that are specific to the full cgroup path of a task, which obsoletes
4676 these controllers anyway. The NetClass= directive is kept around for
4677 legacy compatibility reasons. For a more in-depth description of the
4678 kernel change, please refer to the respective upstream commit:
4680 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
4682 * A new service setting RuntimeMaxSec= has been added that may be used
4683 to specify a maximum runtime for a service. If the timeout is hit, the
4684 service is terminated and put into a failure state.
4686 * A new service setting AmbientCapabilities= has been added. It allows
4687 configuration of additional Linux process capabilities that are
4688 passed to the activated processes. This is only available on very
4691 * The process resource limit settings in service units may now be used
4692 to configure hard and soft limits individually.
4694 * The various libsystemd APIs such as sd-bus or sd-event now publicly
4695 expose support for gcc's __attribute__((cleanup())) C extension.
4696 Specifically, for many object destructor functions alternative
4697 versions have been added that have names suffixed with "p" and take a
4698 pointer to a pointer to the object to destroy, instead of just a
4699 pointer to the object itself. This is useful because these destructor
4700 functions may be used directly as parameters to the cleanup
4701 construct. Internally, systemd has been a heavy user of this GCC
4702 extension for a long time, and with this change similar support is
4703 now available to consumers of the library outside of systemd. Note
4704 that by using this extension in your sources compatibility with old
4705 and strictly ANSI compatible C compilers is lost. However, all gcc or
4706 LLVM versions of recent years support this extension.
4708 * Timer units gained support for a new setting RandomizedDelaySec= that
4709 allows configuring some additional randomized delay to the configured
4710 time. This is useful to spread out timer events to avoid load peaks in
4711 clusters or larger setups.
4713 * Calendar time specifications now support sub-second accuracy.
4715 * Socket units now support listening on SCTP and UDP-lite protocol
4718 * The sd-event API now comes with a full set of man pages.
4720 * Older versions of systemd contained experimental support for
4721 compressing journal files and coredumps with the LZ4 compressor that
4722 was not compatible with the lz4 binary (due to API limitations of the
4723 lz4 library). This support has been removed; only support for files
4724 compatible with the lz4 binary remains. This LZ4 logic is now
4725 officially supported and no longer considered experimental.
4727 * The dkr image import logic has been removed again from importd. dkr's
4728 micro-services focus doesn't fit into the machine image focus of
4729 importd, and quickly got out of date with the upstream dkr API.
4731 * Creation of the /run/lock/lockdev/ directory was dropped from
4732 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
4733 been available for many years. If you still need this, you need to
4734 create your own tmpfiles.d config file with:
4736 d /run/lock/lockdev 0775 root lock -
4738 * The settings StartLimitBurst=, StartLimitInterval=, StartLimitAction=
4739 and RebootArgument= have been moved from the [Service] section of
4740 unit files to [Unit], and they are now supported on all unit types,
4741 not just service units. Of course, systemd will continue to
4742 understand these settings also at the old location, in order to
4743 maintain compatibility.
4745 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
4746 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
4747 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
4748 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
4749 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
4750 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
4751 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
4752 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
4753 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
4754 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
4755 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
4756 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
4757 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
4758 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
4759 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
4760 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
4761 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
4762 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
4763 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
4765 — Berlin, 2016-02-11
4769 * A number of properties previously only settable in unit
4770 files are now also available as properties to set when
4771 creating transient units programmatically via the bus, as it
4772 is exposed with systemd-run's --property=
4773 setting. Specifically, these are: SyslogIdentifier=,
4774 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
4775 EnvironmentFile=, ReadWriteDirectories=,
4776 ReadOnlyDirectories=, InaccessibleDirectories=,
4777 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
4779 * When creating transient services via the bus API it is now
4780 possible to pass in a set of file descriptors to use as
4781 STDIN/STDOUT/STDERR for the invoked process.
4783 * Slice units may now be created transiently via the bus APIs,
4784 similar to the way service and scope units may already be
4785 created transiently.
4787 * Wherever systemd expects a calendar timestamp specification
4788 (like in journalctl's --since= and --until= switches) UTC
4789 timestamps are now supported. Timestamps suffixed with "UTC"
4790 are now considered to be in Universal Time Coordinated
4791 instead of the local timezone. Also, timestamps may now
4792 optionally be specified with sub-second accuracy. Both of
4793 these additions also apply to recurring calendar event
4794 specification, such as OnCalendar= in timer units.
4796 * journalctl gained a new "--sync" switch that asks the
4797 journal daemon to write all so far unwritten log messages to
4798 disk and sync the files, before returning.
4800 * systemd-tmpfiles learned two new line types "q" and "Q" that
4801 operate like "v", but also set up a basic btrfs quota
4802 hierarchy when used on a btrfs file system with quota
4805 * tmpfiles' "v", "q" and "Q" will now create a plain directory
4806 instead of a subvolume (even on a btrfs file system) if the
4807 root directory is a plain directory, and not a
4808 subvolume. This should simplify things with certain chroot()
4809 environments which are not aware of the concept of btrfs
4812 * systemd-detect-virt gained a new --chroot switch to detect
4813 whether execution takes place in a chroot() environment.
4815 * CPUAffinity= now takes CPU index ranges in addition to
4818 * The various memory-related resource limit settings (such as
4819 LimitAS=) now understand the usual K, M, G, ... suffixes to
4820 the base of 1024 (IEC). Similar, the time-related resource
4821 limit settings understand the usual min, h, day, ...
4824 * There's a new system.conf setting DefaultTasksMax= to
4825 control the default TasksMax= setting for services and
4826 scopes running on the system. (TasksMax= is the primary
4827 setting that exposes the "pids" cgroup controller on systemd
4828 and was introduced in the previous systemd release.) The
4829 setting now defaults to 512, which means services that are
4830 not explicitly configured otherwise will only be able to
4831 create 512 processes or threads at maximum, from this
4832 version on. Note that this means that thread- or
4833 process-heavy services might need to be reconfigured to set
4834 TasksMax= to a higher value. It is sufficient to set
4835 TasksMax= in these specific unit files to a higher value, or
4836 even "infinity". Similar, there's now a logind.conf setting
4837 UserTasksMax= that defaults to 4096 and limits the total
4838 number of processes or tasks each user may own
4839 concurrently. nspawn containers also have the TasksMax=
4840 value set by default now, to 8192. Note that all of this
4841 only has an effect if the "pids" cgroup controller is
4842 enabled in the kernel. The general benefit of these changes
4843 should be a more robust and safer system, that provides a
4844 certain amount of per-service fork() bomb protection.
4846 * systemd-nspawn gained the new --network-veth-extra= switch
4847 to define additional and arbitrarily-named virtual Ethernet
4848 links between the host and the container.
4850 * A new service execution setting PassEnvironment= has been
4851 added that allows importing select environment variables
4852 from PID1's environment block into the environment block of
4855 * Timer units gained support for a new RemainAfterElapse=
4856 setting which takes a boolean argument. It defaults to on,
4857 exposing behaviour unchanged to previous releases. If set to
4858 off, timer units are unloaded after they elapsed if they
4859 cannot elapse again. This is particularly useful for
4860 transient timer units, which shall not stay around longer
4861 than until they first elapse.
4863 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
4864 default now (the kernel default is 16). This is beneficial
4865 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
4866 allows substantially larger numbers of queued
4867 datagrams. This should increase the capability of systemd to
4868 parallelize boot-up, as logging and sd_notify() are unlikely
4869 to stall execution anymore. If you need to change the value
4870 from the new defaults, use the usual sysctl.d/ snippets.
4872 * The compression framing format used by the journal or
4873 coredump processing has changed to be in line with what the
4874 official LZ4 tools generate. LZ4 compression support in
4875 systemd was considered unsupported previously, as the format
4876 was not compatible with the normal tools. With this release
4877 this has changed now, and it is hence safe for downstream
4878 distributions to turn it on. While not compressing as well
4879 as the XZ, LZ4 is substantially faster, which makes
4880 it a good default choice for the compression logic in the
4881 journal and in coredump handling.
4883 * Any reference to /etc/mtab has been dropped from
4884 systemd. The file has been obsolete since a while, but
4885 systemd refused to work on systems where it was incorrectly
4886 set up (it should be a symlink or non-existent). Please make
4887 sure to update to util-linux 2.27.1 or newer in conjunction
4888 with this systemd release, which also drops any reference to
4889 /etc/mtab. If you maintain a distribution make sure that no
4890 software you package still references it, as this is a
4891 likely source of bugs. There's also a glibc bug pending,
4892 asking for removal of any reference to this obsolete file:
4894 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
4896 Note that only util-linux versions built with
4897 --enable-libmount-force-mountinfo are supported.
4899 * Support for the ".snapshot" unit type has been removed. This
4900 feature turned out to be little useful and little used, and
4901 has now been removed from the core and from systemctl.
4903 * The dependency types RequiresOverridable= and
4904 RequisiteOverridable= have been removed from systemd. They
4905 have been used only very sparingly to our knowledge and
4906 other options that provide a similar effect (such as
4907 systemctl --mode=ignore-dependencies) are much more useful
4908 and commonly used. Moreover, they were only half-way
4909 implemented as the option to control behaviour regarding
4910 these dependencies was never added to systemctl. By removing
4911 these dependency types the execution engine becomes a bit
4912 simpler. Unit files that use these dependencies should be
4913 changed to use the non-Overridable dependency types
4914 instead. In fact, when parsing unit files with these
4915 options, that's what systemd will automatically convert them
4916 too, but it will also warn, asking users to fix the unit
4917 files accordingly. Removal of these dependency types should
4918 only affect a negligible number of unit files in the wild.
4920 * Behaviour of networkd's IPForward= option changed
4921 (again). It will no longer maintain a per-interface setting,
4922 but propagate one way from interfaces where this is enabled
4923 to the global kernel setting. The global setting will be
4924 enabled when requested by a network that is set up, but
4925 never be disabled again. This change was made to make sure
4926 IPv4 and IPv6 behaviour regarding packet forwarding is
4927 similar (as the Linux IPv6 stack does not support
4928 per-interface control of this setting) and to minimize
4931 * In unit files the behaviour of %u, %U, %h, %s has
4932 changed. These specifiers will now unconditionally resolve
4933 to the various user database fields of the user that the
4934 systemd instance is running as, instead of the user
4935 configured in the specific unit via User=. Note that this
4936 effectively doesn't change much, as resolving of these
4937 specifiers was already turned off in the --system instance
4938 of systemd, as we cannot do NSS lookups from PID 1. In the
4939 --user instance of systemd these specifiers where correctly
4940 resolved, but hardly made any sense, since the user instance
4941 lacks privileges to do user switches anyway, and User= is
4942 hence useless. Moreover, even in the --user instance of
4943 systemd behaviour was awkward as it would only take settings
4944 from User= assignment placed before the specifier into
4945 account. In order to unify and simplify the logic around
4946 this the specifiers will now always resolve to the
4947 credentials of the user invoking the manager (which in case
4948 of PID 1 is the root user).
4950 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
4951 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
4952 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
4953 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
4954 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
4955 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
4956 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
4957 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
4958 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
4959 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
4960 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
4961 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
4962 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
4963 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
4966 — Berlin, 2015-11-18
4970 * systemd now depends on util-linux v2.27. More specifically,
4971 the newly added mount monitor feature in libmount now
4972 replaces systemd's former own implementation.
4974 * libmount mandates /etc/mtab not to be regular file, and
4975 systemd now enforces this condition at early boot.
4976 /etc/mtab has been deprecated and warned about for a very
4977 long time, so systems running systemd should already have
4978 stopped having this file around as anything else than a
4979 symlink to /proc/self/mounts.
4981 * Support for the "pids" cgroup controller has been added. It
4982 allows accounting the number of tasks in a cgroup and
4983 enforcing limits on it. This adds two new setting
4984 TasksAccounting= and TasksMax= to each unit, as well as a
4985 global option DefaultTasksAccounting=.
4987 * Support for the "net_cls" cgroup controller has been added.
4988 It allows assigning a net class ID to each task in the
4989 cgroup, which can then be used in firewall rules and traffic
4990 shaping configurations. Note that the kernel netfilter net
4991 class code does not currently work reliably for ingress
4992 packets on unestablished sockets.
4994 This adds a new config directive called NetClass= to CGroup
4995 enabled units. Allowed values are positive numbers for fixed
4996 assignments and "auto" for picking a free value
4999 * 'systemctl is-system-running' now returns 'offline' if the
5000 system is not booted with systemd. This command can now be
5001 used as a substitute for 'systemd-notify --booted'.
5003 * Watchdog timeouts have been increased to 3 minutes for all
5004 in-tree service files. Apparently, disk IO issues are more
5005 frequent than we hoped, and user reported >1 minute waiting
5008 * 'machine-id-commit' functionality has been merged into
5009 'machine-id-setup --commit'. The separate binary has been
5012 * The WorkingDirectory= directive in unit files may now be set
5013 to the special value '~'. In this case, the working
5014 directory is set to the home directory of the user
5015 configured in User=.
5017 * "machinectl shell" will now open the shell in the home
5018 directory of the selected user by default.
5020 * The CrashChVT= configuration file setting is renamed to
5021 CrashChangeVT=, following our usual logic of not
5022 abbreviating unnecessarily. The old directive is still
5023 supported for compat reasons. Also, this directive now takes
5024 an integer value between 1 and 63, or a boolean value. The
5025 formerly supported '-1' value for disabling stays around for
5028 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
5029 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
5030 RootDirectory= properties can now be set for transient
5033 * The systemd-analyze tool gained a new "set-log-target" verb
5034 to change the logging target the system manager logs to
5035 dynamically during runtime. This is similar to how
5036 "systemd-analyze set-log-level" already changes the log
5039 * In nspawn /sys is now mounted as tmpfs, with only a selected
5040 set of subdirectories mounted in from the real sysfs. This
5041 enhances security slightly, and is useful for ensuring user
5042 namespaces work correctly.
5044 * Support for USB FunctionFS activation has been added. This
5045 allows implementation of USB gadget services that are
5046 activated as soon as they are requested, so that they don't
5047 have to run continuously, similar to classic socket
5050 * The "systemctl exit" command now optionally takes an
5051 additional parameter that sets the exit code to return from
5052 the systemd manager when exiting. This is only relevant when
5053 running the systemd user instance, or when running the
5054 system instance in a container.
5056 * sd-bus gained the new API calls sd_bus_path_encode_many()
5057 and sd_bus_path_decode_many() that allow easy encoding and
5058 decoding of multiple identifier strings inside a D-Bus
5059 object path. Another new call sd_bus_default_flush_close()
5060 has been added to flush and close per-thread default
5063 * systemd-cgtop gained support for a -M/--machine= switch to
5064 show the control groups within a certain container only.
5066 * "systemctl kill" gained support for an optional --fail
5067 switch. If specified the requested operation will fail of no
5068 processes have been killed, because the unit had no
5069 processes attached, or similar.
5071 * A new systemd.crash_reboot=1 kernel command line option has
5072 been added that triggers a reboot after crashing. This can
5073 also be set through CrashReboot= in systemd.conf.
5075 * The RuntimeDirectory= setting now understands unit
5076 specifiers like %i or %f.
5078 * A new (still internal) library API sd-ipv4acd has been added,
5079 that implements address conflict detection for IPv4. It's
5080 based on code from sd-ipv4ll, and will be useful for
5081 detecting DHCP address conflicts.
5083 * File descriptors passed during socket activation may now be
5084 named. A new API sd_listen_fds_with_names() is added to
5085 access the names. The default names may be overridden,
5086 either in the .socket file using the FileDescriptorName=
5087 parameter, or by passing FDNAME= when storing the file
5088 descriptors using sd_notify().
5090 * systemd-networkd gained support for:
5092 - Setting the IPv6 Router Advertisement settings via
5093 IPv6AcceptRouterAdvertisements= in .network files.
5095 - Configuring the HelloTimeSec=, MaxAgeSec= and
5096 ForwardDelaySec= bridge parameters in .netdev files.
5098 - Configuring PreferredSource= for static routes in
5101 * The "ask-password" framework used to query for LUKS harddisk
5102 passwords or SSL passwords during boot gained support for
5103 caching passwords in the kernel keyring, if it is
5104 available. This makes sure that the user only has to type in
5105 a passphrase once if there are multiple objects to unlock
5106 with the same one. Previously, such password caching was
5107 available only when Plymouth was used; this moves the
5108 caching logic into the systemd codebase itself. The
5109 "systemd-ask-password" utility gained a new --keyname=
5110 switch to control which kernel keyring key to use for
5111 caching a password in. This functionality is also useful for
5112 enabling display managers such as gdm to automatically
5113 unlock the user's GNOME keyring if its passphrase, the
5114 user's password and the harddisk password are the same, if
5115 gdm-autologin is used.
5117 * When downloading tar or raw images using "machinectl
5118 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
5119 file is now also downloaded, if it is available and stored
5120 next to the image file.
5122 * Units of type ".socket" gained a new boolean setting
5123 Writable= which is only useful in conjunction with
5124 ListenSpecial=. If true, enables opening the specified
5125 special file in O_RDWR mode rather than O_RDONLY mode.
5127 * systemd-rfkill has been reworked to become a singleton
5128 service that is activated through /dev/rfkill on each rfkill
5129 state change and saves the settings to disk. This way,
5130 systemd-rfkill is now compatible with devices that exist
5131 only intermittendly, and even restores state if the previous
5132 system shutdown was abrupt rather than clean.
5134 * The journal daemon gained support for vacuuming old journal
5135 files controlled by the number of files that shall remain,
5136 in addition to the already existing control by size and by
5137 date. This is useful as journal interleaving performance
5138 degrades with too many separate journal files, and allows
5139 putting an effective limit on them. The new setting defaults
5140 to 100, but this may be changed by setting SystemMaxFiles=
5141 and RuntimeMaxFiles= in journald.conf. Also, the
5142 "journalctl" tool gained the new --vacuum-files= switch to
5143 manually vacuum journal files to leave only the specified
5144 number of files in place.
5146 * udev will now create /dev/disk/by-path links for ATA devices
5147 on kernels where that is supported.
5149 * Galician, Serbian, Turkish and Korean translations were added.
5151 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
5152 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
5153 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
5154 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
5155 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
5156 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
5157 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
5158 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
5159 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
5160 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
5161 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
5162 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
5163 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
5164 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
5165 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
5166 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
5167 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
5168 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
5170 — Berlin, 2015-10-07
5174 * The DHCP implementation of systemd-networkd gained a set of
5177 - The DHCP server now supports emitting DNS and NTP
5178 information. It may be enabled and configured via
5179 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
5180 and NTP information is enabled, but no servers are
5181 configured, the corresponding uplink information (if there
5182 is any) is propagated.
5184 - Server and client now support transmission and reception
5185 of timezone information. It can be configured via the
5186 newly introduced network options UseTimezone=,
5187 EmitTimezone=, and Timezone=. Transmission of timezone
5188 information is enabled between host and containers by
5189 default now: the container will change its local timezone
5190 to what the host has set.
5192 - Lease timeouts can now be configured via
5193 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
5195 - The DHCP server improved on the stability of
5196 leases. Clients are more likely to get the same lease
5197 information back, even if the server loses state.
5199 - The DHCP server supports two new configuration options to
5200 control the lease address pool metrics, PoolOffset= and
5203 * The encapsulation limit of tunnels in systemd-networkd may
5204 now be configured via 'EncapsulationLimit='. It allows
5205 modifying the maximum additional levels of encapsulation
5206 that are permitted to be prepended to a packet.
5208 * systemd now supports the concept of user buses replacing
5209 session buses, if used with dbus-1.10 (and enabled via dbus
5210 --enable-user-session). It previously only supported this on
5211 kdbus-enabled systems, and this release expands this to
5212 'dbus-daemon' systems.
5214 * systemd-networkd now supports predictable interface names
5217 * systemd now optionally supports the new Linux kernel
5218 "unified" control group hierarchy. If enabled via the kernel
5219 command-line option 'systemd.unified_cgroup_hierarchy=1',
5220 systemd will try to mount the unified cgroup hierarchy
5221 directly on /sys/fs/cgroup. If not enabled, or not
5222 available, systemd will fall back to the legacy cgroup
5223 hierarchy setup, as before. Host system and containers can
5224 mix and match legacy and unified hierarchies as they
5225 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
5226 environment variable to individually select the hierarchy to
5227 use for executed containers. By default, nspawn will use the
5228 unified hierarchy for the containers if the host uses the
5229 unified hierarchy, and the legacy hierarchy otherwise.
5230 Please note that at this point the unified hierarchy is an
5231 experimental kernel feature and is likely to change in one
5232 of the next kernel releases. Therefore, it should not be
5233 enabled by default in downstream distributions yet. The
5234 minimum required kernel version for the unified hierarchy to
5235 work is 4.2. Note that when the unified hierarchy is used
5236 for the first time delegated access to controllers is
5237 safe. Because of this systemd-nspawn containers will get
5238 access to controllers now, as will systemd user
5239 sessions. This means containers and user sessions may now
5240 manage their own resources, partitioning up what the system
5243 * A new special scope unit "init.scope" has been introduced
5244 that encapsulates PID 1 of the system. It may be used to
5245 determine resource usage and enforce resource limits on PID
5246 1 itself. PID 1 hence moved out of the root of the control
5249 * The cgtop tool gained support for filtering out kernel
5250 threads when counting tasks in a control group. Also, the
5251 count of processes is now recursively summed up by
5252 default. Two options -k and --recursive= have been added to
5253 revert to old behaviour. The tool has also been updated to
5254 work correctly in containers now.
5256 * systemd-nspawn's --bind= and --bind-ro= options have been
5257 extended to allow creation of non-recursive bind mounts.
5259 * libsystemd gained two new calls sd_pid_get_cgroup() and
5260 sd_peer_get_cgroup() which return the control group path of
5261 a process or peer of a connected AF_UNIX socket. This
5262 function call is particularly useful when implementing
5263 delegated subtrees support in the control group hierarchy.
5265 * The "sd-event" event loop API of libsystemd now supports
5266 correct dequeuing of real-time signals, without losing
5269 * When systemd requests a polkit decision when managing units it
5270 will now add additional fields to the request, including unit
5271 name and desired operation. This enables more powerful polkit
5272 policies, that make decisions depending on these parameters.
5274 * nspawn learnt support for .nspawn settings files, that may
5275 accompany the image files or directories of containers, and
5276 may contain additional settings for the container. This is
5277 an alternative to configuring container parameters via the
5278 nspawn command line.
5280 Contributions from: Cristian Rodríguez, Daniel Mack, David
5281 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
5282 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
5283 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
5284 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
5285 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
5286 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
5287 Andersen, Tom Gundersen, Torstein Husebø
5289 — Berlin, 2015-09-08
5293 * machinectl gained a new verb 'shell' which opens a fresh
5294 shell on the target container or the host. It is similar to
5295 the existing 'login' command of machinectl, but spawns the
5296 shell directly without prompting for username or
5297 password. The pseudo machine '.host' now refers to the local
5298 host and is used by default. Hence, 'machinectl shell' can
5299 be used as replacement for 'su -' which spawns a session as
5300 a fresh systemd unit in a way that is fully isolated from
5301 the originating session.
5303 * systemd-networkd learned to cope with private-zone DHCP
5304 options and allows other programs to query the values.
5306 * SELinux access control when enabling/disabling units is no
5307 longer enforced with this release. The previous implementation
5308 was incorrect, and a new corrected implementation is not yet
5309 available. As unit file operations are still protected via
5310 polkit and D-Bus policy this is not a security problem. Yet,
5311 distributions which care about optimal SELinux support should
5312 probably not stabilize on this release.
5314 * sd-bus gained support for matches of type "arg0has=", that
5315 test for membership of strings in string arrays sent in bus
5318 * systemd-resolved now dumps the contents of its DNS and LLMNR
5319 caches to the logs on reception of the SIGUSR1 signal. This
5320 is useful to debug DNS behaviour.
5322 * The coredumpctl tool gained a new --directory= option to
5323 operate on journal files in a specific directory.
5325 * "systemctl reboot" and related commands gained a new
5326 "--message=" option which may be used to set a free-text
5327 wall message when shutting down or rebooting the
5328 system. This message is also logged, which is useful for
5329 figuring out the reason for a reboot or shutdown a
5332 * The "systemd-resolve-host" tool's -i switch now takes
5333 network interface numbers as alternative to interface names.
5335 * A new unit file setting for services has been introduced:
5336 UtmpMode= allows configuration of how precisely systemd
5337 handles utmp and wtmp entries for the service if this is
5338 enabled. This allows writing services that appear similar to
5339 user sessions in the output of the "w", "who", "last" and
5342 * systemd-resolved will now locally synthesize DNS resource
5343 records for the "localhost" and "gateway" domains as well as
5344 the local hostname. This should ensure that clients querying
5345 RRs via resolved will get similar results as those going via
5346 NSS, if nss-myhostname is enabled.
5348 Contributions from: Alastair Hughes, Alex Crawford, Daniel
5349 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
5350 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
5351 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
5352 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
5353 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
5354 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
5355 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
5356 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
5357 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
5358 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
5359 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
5361 — Berlin, 2015-08-27
5365 * The systemd-efi-boot-generator functionality was merged into
5366 systemd-gpt-auto-generator.
5368 * systemd-networkd now supports Group Policy for vxlan
5369 devices. It can be enabled via the new boolean configuration
5370 option called 'GroupPolicyExtension='.
5372 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
5373 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
5374 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
5376 — Berlin, 2015-07-31
5380 * The python-systemd code has been removed from the systemd repository.
5381 A new repository has been created which accommodates the code from
5382 now on, and we kindly ask distributions to create a separate package
5383 for this: https://github.com/systemd/python-systemd
5385 * The systemd daemon will now reload its main configuration
5386 (/etc/systemd/system.conf) on daemon-reload.
5388 * sd-dhcp now exposes vendor specific extensions via
5389 sd_dhcp_lease_get_vendor_specific().
5391 * systemd-networkd gained a number of new configuration options.
5393 - A new boolean configuration option for TAP devices called
5394 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
5395 device, thus allowing to send and receive GSO packets.
5397 - A new tunnel configuration option called 'CopyDSCP='.
5398 If enabled, the DSCP field of ip6 tunnels is copied into the
5399 decapsulated packet.
5401 - A set of boolean bridge configuration options were added.
5402 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
5403 and 'UnicastFlood=' are now parsed by networkd and applied to the
5404 respective bridge link device via the respective IFLA_BRPORT_*
5407 - A new string configuration option to override the hostname sent
5408 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
5409 is true, networkd will use the configured hostname instead of the
5410 system hostname when sending DHCP requests.
5412 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
5413 networkd will configure the IPv6 flow-label of the tunnel device
5414 according to RFC2460.
5416 - The 'macvtap' virtual network devices are now supported, similar to
5417 the already supported 'macvlan' devices.
5419 * systemd-resolved now implements RFC5452 to improve resilience against
5420 cache poisoning. Additionally, source port randomization is enabled
5421 by default to further protect against DNS spoofing attacks.
5423 * nss-mymachines now supports translating UIDs and GIDs of running
5424 containers with user-namespaces enabled. If a container 'foo'
5425 translates a host uid 'UID' to the container uid 'TUID', then
5426 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
5427 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
5428 mapped as 'vg-foo-TGID'.
5430 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
5431 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
5432 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
5433 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
5434 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
5435 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
5436 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
5437 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
5438 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
5439 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
5441 — Berlin, 2015-07-29
5445 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
5446 There are no known issues with current sysfs, and udev does not need
5447 or should be used to work around such bugs.
5449 * udev does no longer enable USB HID power management. Several reports
5450 indicate, that some devices cannot handle that setting.
5452 * The udev accelerometer helper was removed. The functionality
5453 is now fully included in iio-sensor-proxy. But this means,
5454 older iio-sensor-proxy versions will no longer provide
5455 accelerometer/orientation data with this systemd version.
5456 Please upgrade iio-sensor-proxy to version 1.0.
5458 * networkd gained a new configuration option IPv6PrivacyExtensions=
5459 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
5460 for Stateless Address") on selected networks.
5462 * For the sake of fewer build-time dependencies and less code in the
5463 main repository, the python bindings are about to be removed in the
5464 next release. A new repository has been created which accommodates
5465 the code from now on, and we kindly ask distributions to create a
5466 separate package for this. The removal will take place in v223.
5468 https://github.com/systemd/python-systemd
5470 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
5471 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
5472 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
5473 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
5474 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
5475 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
5476 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
5477 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
5478 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
5479 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
5481 — Berlin, 2015-07-07
5485 * The sd-bus.h and sd-event.h APIs have now been declared
5486 stable and have been added to the official interface of
5487 libsystemd.so. sd-bus implements an alternative D-Bus client
5488 library, that is relatively easy to use, very efficient and
5489 supports both classic D-Bus as well as kdbus as transport
5490 backend. sd-event is a generic event loop abstraction that
5491 is built around Linux epoll, but adds features such as event
5492 prioritization or efficient timer handling. Both APIs are good
5493 choices for C programs looking for a bus and/or event loop
5494 implementation that is minimal and does not have to be
5495 portable to other kernels.
5497 * kdbus support is no longer compile-time optional. It is now
5498 always built-in. However, it can still be disabled at
5499 runtime using the kdbus=0 kernel command line setting, and
5500 that setting may be changed to default to off, by specifying
5501 --disable-kdbus at build-time. Note though that the kernel
5502 command line setting has no effect if the kdbus.ko kernel
5503 module is not installed, in which case kdbus is (obviously)
5504 also disabled. We encourage all downstream distributions to
5505 begin testing kdbus by adding it to the kernel images in the
5506 development distributions, and leaving kdbus support in
5509 * The minimal required util-linux version has been bumped to
5512 * Support for chkconfig (--enable-chkconfig) was removed in
5513 favor of calling an abstraction tool
5514 /lib/systemd/systemd-sysv-install. This needs to be
5515 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
5516 in README for details.
5518 * If there's a systemd unit and a SysV init script for the
5519 same service name, and the user executes "systemctl enable"
5520 for it (or a related call), then this will now enable both
5521 (or execute the related operation on both), not just the
5524 * The libudev API documentation has been converted from gtkdoc
5527 * gudev has been removed from the systemd tree, it is now an
5530 * The systemd-cgtop tool learnt a new --raw switch to generate
5531 "raw" (machine parsable) output.
5533 * networkd's IPForwarding= .network file setting learnt the
5534 new setting "kernel", which ensures that networkd does not
5535 change the IP forwarding sysctl from the default kernel
5538 * The systemd-logind bus API now exposes a new boolean
5539 property "Docked" that reports whether logind considers the
5540 system "docked", i.e. connected to a docking station or not.
5542 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
5543 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
5544 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
5545 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
5546 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
5547 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
5548 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
5549 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
5550 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
5551 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
5552 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
5553 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
5554 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
5555 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
5556 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
5557 Fink, Zbigniew Jędrzejewski-Szmek
5559 — Berlin, 2015-06-19
5563 * The gudev library has been extracted into a separate repository
5564 available at: https://git.gnome.org/browse/libgudev/
5565 It is now managed as part of the Gnome project. Distributions
5566 are recommended to pass --disable-gudev to systemd and use
5567 gudev from the Gnome project instead. gudev is still included
5568 in systemd, for now. It will be removed soon, though. Please
5569 also see the announcement-thread on systemd-devel:
5570 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
5572 * systemd now exposes a CPUUsageNSec= property for each
5573 service unit on the bus, that contains the overall consumed
5574 CPU time of a service (the sum of what each process of the
5575 service consumed). This value is only available if
5576 CPUAccounting= is turned on for a service, and is then shown
5577 in the "systemctl status" output.
5579 * Support for configuring alternative mappings of the old SysV
5580 runlevels to systemd targets has been removed. They are now
5581 hardcoded in a way that runlevels 2, 3, 4 all map to
5582 multi-user.target and 5 to graphical.target (which
5583 previously was already the default behaviour).
5585 * The auto-mounter logic gained support for mount point
5586 expiry, using a new TimeoutIdleSec= setting in .automount
5587 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
5589 * The EFI System Partition (ESP) as mounted to /boot by
5590 systemd-efi-boot-generator will now be unmounted
5591 automatically after 2 minutes of not being used. This should
5592 minimize the risk of ESP corruptions.
5594 * New /etc/fstab options x-systemd.requires= and
5595 x-systemd.requires-mounts-for= are now supported to express
5596 additional dependencies for mounts. This is useful for
5597 journalling file systems that support external journal
5598 devices or overlay file systems that require underlying file
5599 systems to be mounted.
5601 * systemd does not support direct live-upgrades (via systemctl
5602 daemon-reexec) from versions older than v44 anymore. As no
5603 distribution we are aware of shipped such old versions in a
5604 stable release this should not be problematic.
5606 * When systemd forks off a new per-connection service instance
5607 it will now set the $REMOTE_ADDR environment variable to the
5608 remote IP address, and $REMOTE_PORT environment variable to
5609 the remote IP port. This behaviour is similar to the
5610 corresponding environment variables defined by CGI.
5612 * systemd-networkd gained support for uplink failure
5613 detection. The BindCarrier= option allows binding interface
5614 configuration dynamically to the link sense of other
5615 interfaces. This is useful to achieve behaviour like in
5618 * systemd-networkd gained support for configuring the DHCP
5619 client identifier to use when requesting leases.
5621 * systemd-networkd now has a per-network UseNTP= option to
5622 configure whether NTP server information acquired via DHCP
5623 is passed on to services like systemd-timesyncd.
5625 * systemd-networkd gained support for vti6 tunnels.
5627 * Note that systemd-networkd manages the sysctl variable
5628 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
5629 it is configured for since v219. The variable controls IP
5630 forwarding, and is a per-interface alternative to the global
5631 /proc/sys/net/ipv[46]/ip_forward. This setting is
5632 configurable in the IPForward= option, which defaults to
5633 "no". This means if networkd is used for an interface it is
5634 no longer sufficient to set the global sysctl option to turn
5635 on IP forwarding! Instead, the .network file option
5636 IPForward= needs to be turned on! Note that the
5637 implementation of this behaviour was broken in v219 and has
5640 * Many bonding and vxlan options are now configurable in
5643 * systemd-nspawn gained a new --property= setting to set unit
5644 properties for the container scope. This is useful for
5645 setting resource parameters (e.g. "CPUShares=500") on
5646 containers started from the command line.
5648 * systemd-nspawn gained a new --private-users= switch to make
5649 use of user namespacing available on recent Linux kernels.
5651 * systemd-nspawn may now be called as part of a shell pipeline
5652 in which case the pipes used for stdin and stdout are passed
5653 directly to the process invoked in the container, without
5654 indirection via a pseudo tty.
5656 * systemd-nspawn gained a new switch to control the UNIX
5657 signal to use when killing the init process of the container
5660 * systemd-nspawn gained a new --overlay= switch for mounting
5661 overlay file systems into the container using the new kernel
5664 * When a container image is imported via systemd-importd and
5665 the host file system is not btrfs, a loopback block device
5666 file is created in /var/lib/machines.raw with a btrfs file
5667 system inside. It is then mounted to /var/lib/machines to
5668 enable btrfs features for container management. The loopback
5669 file and btrfs file system is grown as needed when container
5670 images are imported via systemd-importd.
5672 * systemd-machined/systemd-importd gained support for btrfs
5673 quota, to enforce container disk space limits on disk. This
5674 is exposed in "machinectl set-limit".
5676 * systemd-importd now can import containers from local .tar,
5677 .raw and .qcow2 images, and export them to .tar and .raw. It
5678 can also import dkr v2 images now from the network (on top
5681 * systemd-importd gained support for verifying downloaded
5682 images with gpg2 (previously only gpg1 was supported).
5684 * systemd-machined, systemd-logind, systemd: most bus calls are
5685 now accessible to unprivileged processes via polkit. Also,
5686 systemd-logind will now allow users to kill their own sessions
5687 without further privileges or authorization.
5689 * systemd-shutdownd has been removed. This service was
5690 previously responsible for implementing scheduled shutdowns
5691 as exposed in /usr/bin/shutdown's time parameter. This
5692 functionality has now been moved into systemd-logind and is
5693 accessible via a bus interface.
5695 * "systemctl reboot" gained a new switch --firmware-setup that
5696 can be used to reboot into the EFI firmware setup, if that
5697 is available. systemd-logind now exposes an API on the bus
5698 to trigger such reboots, in case graphical desktop UIs want
5699 to cover this functionality.
5701 * "systemctl enable", "systemctl disable" and "systemctl mask"
5702 now support a new "--now" switch. If specified the units
5703 that are enabled will also be started, and the ones
5704 disabled/masked also stopped.
5706 * The Gummiboot EFI boot loader tool has been merged into
5707 systemd, and renamed to "systemd-boot". The bootctl tool has been
5708 updated to support systemd-boot.
5710 * An EFI kernel stub has been added that may be used to create
5711 kernel EFI binaries that contain not only the actual kernel,
5712 but also an initrd, boot splash, command line and OS release
5713 information. This combined binary can then be signed as a
5714 single image, so that the firmware can verify it all in one
5715 step. systemd-boot has special support for EFI binaries created
5716 like this and can extract OS release information from them
5717 and show them in the boot menu. This functionality is useful
5718 to implement cryptographically verified boot schemes.
5720 * Optional support has been added to systemd-fsck to pass
5721 fsck's progress report to an AF_UNIX socket in the file
5724 * udev will no longer create device symlinks for all block
5725 devices by default. A blacklist for excluding special block
5726 devices from this logic has been turned into a whitelist
5727 that requires picking block devices explicitly that require
5730 * A new (currently still internal) API sd-device.h has been
5731 added to libsystemd. This modernized API is supposed to
5732 replace libudev eventually. In fact, already much of libudev
5733 is now just a wrapper around sd-device.h.
5735 * A new hwdb database for storing metadata about pointing
5736 stick devices has been added.
5738 * systemd-tmpfiles gained support for setting file attributes
5739 similar to the "chattr" tool with new 'h' and 'H' lines.
5741 * systemd-journald will no longer unconditionally set the
5742 btrfs NOCOW flag on new journal files. This is instead done
5743 with tmpfiles snippet using the new 'h' line type. This
5744 allows easy disabling of this logic, by masking the
5745 journal-nocow.conf tmpfiles file.
5747 * systemd-journald will now translate audit message types to
5748 human readable identifiers when writing them to the
5749 journal. This should improve readability of audit messages.
5751 * The LUKS logic gained support for the offset= and skip=
5752 options in /etc/crypttab, as previously implemented by
5755 * /usr/lib/os-release gained a new optional field VARIANT= for
5756 distributions that support multiple variants (such as a
5757 desktop edition, a server edition, ...)
5759 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
5760 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
5761 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
5762 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
5763 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
5764 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
5765 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
5766 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
5767 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
5768 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
5769 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
5770 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
5771 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
5772 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
5773 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
5774 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
5775 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
5776 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
5777 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
5778 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
5779 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
5780 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
5781 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
5782 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
5783 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
5784 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
5785 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
5787 — Berlin, 2015-05-22
5791 * Introduce a new API "sd-hwdb.h" for querying the hardware
5792 metadata database. With this minimal interface one can query
5793 and enumerate the udev hwdb, decoupled from the old libudev
5794 library. libudev's interface for this is now only a wrapper
5795 around sd-hwdb. A new tool systemd-hwdb has been added to
5796 interface with and update the database.
5798 * When any of systemd's tools copies files (for example due to
5799 tmpfiles' C lines) a btrfs reflink will attempted first,
5800 before bytewise copying is done.
5802 * systemd-nspawn gained a new --ephemeral switch. When
5803 specified a btrfs snapshot is taken of the container's root
5804 directory, and immediately removed when the container
5805 terminates again. Thus, a container can be started whose
5806 changes never alter the container's root directory, and are
5807 lost on container termination. This switch can also be used
5808 for starting a container off the root file system of the
5809 host without affecting the host OS. This switch is only
5810 available on btrfs file systems.
5812 * systemd-nspawn gained a new --template= switch. It takes the
5813 path to a container tree to use as template for the tree
5814 specified via --directory=, should that directory be
5815 missing. This allows instantiating containers dynamically,
5816 on first run. This switch is only available on btrfs file
5819 * When a .mount unit refers to a mount point on which multiple
5820 mounts are stacked, and the .mount unit is stopped all of
5821 the stacked mount points will now be unmounted until no
5822 mount point remains.
5824 * systemd now has an explicit notion of supported and
5825 unsupported unit types. Jobs enqueued for unsupported unit
5826 types will now fail with an "unsupported" error code. More
5827 specifically .swap, .automount and .device units are not
5828 supported in containers, .busname units are not supported on
5829 non-kdbus systems. .swap and .automount are also not
5830 supported if their respective kernel compile time options
5833 * machinectl gained support for two new "copy-from" and
5834 "copy-to" commands for copying files from a running
5835 container to the host or vice versa.
5837 * machinectl gained support for a new "bind" command to bind
5838 mount host directories into local containers. This is
5839 currently only supported for nspawn containers.
5841 * networkd gained support for configuring bridge forwarding
5842 database entries (fdb) from .network files.
5844 * A new tiny daemon "systemd-importd" has been added that can
5845 download container images in tar, raw, qcow2 or dkr formats,
5846 and make them available locally in /var/lib/machines, so
5847 that they can run as nspawn containers. The daemon can GPG
5848 verify the downloads (not supported for dkr, since it has no
5849 provisions for verifying downloads). It will transparently
5850 decompress bz2, xz, gzip compressed downloads if necessary,
5851 and restore sparse files on disk. The daemon uses privilege
5852 separation to ensure the actual download logic runs with
5853 fewer privileges than the daemon itself. machinectl has
5854 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
5855 make the functionality of importd available to the
5856 user. With this in place the Fedora and Ubuntu "Cloud"
5857 images can be downloaded and booted as containers unmodified
5858 (the Fedora images lack the appropriate GPG signature files
5859 currently, so they cannot be verified, but this will change
5860 soon, hopefully). Note that downloading images is currently
5861 only fully supported on btrfs.
5863 * machinectl is now able to list container images found in
5864 /var/lib/machines, along with some metadata about sizes of
5865 disk and similar. If the directory is located on btrfs and
5866 quota is enabled, this includes quota display. A new command
5867 "image-status" has been added that shows additional
5868 information about images.
5870 * machinectl is now able to clone container images
5871 efficiently, if the underlying file system (btrfs) supports
5872 it, with the new "machinectl clone" command. It also
5873 gained commands for renaming and removing images, as well as
5874 marking them read-only or read-write (supported also on
5875 legacy file systems).
5877 * networkd gained support for collecting LLDP network
5878 announcements, from hardware that supports this. This is
5879 shown in networkctl output.
5881 * systemd-run gained support for a new -t (--pty) switch for
5882 invoking a binary on a pty whose input and output is
5883 connected to the invoking terminal. This allows executing
5884 processes as system services while interactively
5885 communicating with them via the terminal. Most interestingly
5886 this is supported across container boundaries. Invoking
5887 "systemd-run -t /bin/bash" is an alternative to running a
5888 full login session, the difference being that the former
5889 will not register a session, nor go through the PAM session
5892 * tmpfiles gained support for a new "v" line type for creating
5893 btrfs subvolumes. If the underlying file system is a legacy
5894 file system, this automatically degrades to creating a
5895 normal directory. Among others /var/lib/machines is now
5896 created like this at boot, should it be missing.
5898 * The directory /var/lib/containers/ has been deprecated and
5899 been replaced by /var/lib/machines. The term "machines" has
5900 been used in the systemd context as generic term for both
5901 VMs and containers, and hence appears more appropriate for
5902 this, as the directory can also contain raw images bootable
5905 * systemd-nspawn when invoked with -M but without --directory=
5906 or --image= is now capable of searching for the container
5907 root directory, subvolume or disk image automatically, in
5908 /var/lib/machines. systemd-nspawn@.service has been updated
5909 to make use of this, thus allowing it to be used for raw
5912 * A new machines.target unit has been introduced that is
5913 supposed to group all containers/VMs invoked as services on
5914 the system. systemd-nspawn@.service has been updated to
5915 integrate with that.
5917 * machinectl gained a new "start" command, for invoking a
5918 container as a service. "machinectl start foo" is mostly
5919 equivalent to "systemctl start systemd-nspawn@foo.service",
5920 but handles escaping in a nicer way.
5922 * systemd-nspawn will now mount most of the cgroupfs tree
5923 read-only into each container, with the exception of the
5924 container's own subtree in the name=systemd hierarchy.
5926 * journald now sets the special FS_NOCOW file flag for its
5927 journal files. This should improve performance on btrfs, by
5928 avoiding heavy fragmentation when journald's write-pattern
5929 is used on COW file systems. It degrades btrfs' data
5930 integrity guarantees for the files to the same levels as for
5931 ext3/ext4 however. This should be OK though as journald does
5932 its own data integrity checks and all its objects are
5933 checksummed on disk. Also, journald should handle btrfs disk
5934 full events a lot more gracefully now, by processing SIGBUS
5935 errors, and not relying on fallocate() anymore.
5937 * When journald detects that journal files it is writing to
5938 have been deleted it will immediately start new journal
5941 * systemd now provides a way to store file descriptors
5942 per-service in PID 1. This is useful for daemons to ensure
5943 that fds they require are not lost during a daemon
5944 restart. The fds are passed to the daemon on the next
5945 invocation in the same way socket activation fds are
5946 passed. This is now used by journald to ensure that the
5947 various sockets connected to all the system's stdout/stderr
5948 are not lost when journald is restarted. File descriptors
5949 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
5950 an extension to sd_notify(). Note that a limit is enforced
5951 on the number of fds a service can store in PID 1, and it
5952 defaults to 0, so that no fds may be stored, unless this is
5953 explicitly turned on.
5955 * The default TERM variable to use for units connected to a
5956 terminal, when no other value is explicitly is set is now
5957 vt220 rather than vt102. This should be fairly safe still,
5958 but allows PgUp/PgDn work.
5960 * The /etc/crypttab option header= as known from Debian is now
5963 * "loginctl user-status" and "loginctl session-status" will
5964 now show the last 10 lines of log messages of the
5965 user/session following the status output. Similar,
5966 "machinectl status" will show the last 10 log lines
5967 associated with a virtual machine or container
5968 service. (Note that this is usually not the log messages
5969 done in the VM/container itself, but simply what the
5970 container manager logs. For nspawn this includes all console
5973 * "loginctl session-status" without further argument will now
5974 show the status of the session of the caller. Similar,
5975 "lock-session", "unlock-session", "activate",
5976 "enable-linger", "disable-linger" may now be called without
5977 session/user parameter in which case they apply to the
5978 caller's session/user.
5980 * An X11 session scriptlet is now shipped that uploads
5981 $DISPLAY and $XAUTHORITY into the environment of the systemd
5982 --user daemon if a session begins. This should improve
5983 compatibility with X11 enabled applications run as systemd
5986 * Generators are now subject to masking via /etc and /run, the
5987 same way as unit files.
5989 * networkd .network files gained support for configuring
5990 per-link IPv4/IPv6 packet forwarding as well as IPv4
5991 masquerading. This is by default turned on for veth links to
5992 containers, as registered by systemd-nspawn. This means that
5993 nspawn containers run with --network-veth will now get
5994 automatic routed access to the host's networks without any
5995 further configuration or setup, as long as networkd runs on
5998 * systemd-nspawn gained the --port= (-p) switch to expose TCP
5999 or UDP posts of a container on the host. With this in place
6000 it is possible to run containers with private veth links
6001 (--network-veth), and have their functionality exposed on
6002 the host as if their services were running directly on the
6005 * systemd-nspawn's --network-veth switch now gained a short
6006 version "-n", since with the changes above it is now truly
6007 useful out-of-the-box. The systemd-nspawn@.service has been
6008 updated to make use of it too by default.
6010 * systemd-nspawn will now maintain a per-image R/W lock, to
6011 ensure that the same image is not started more than once
6012 writable. (It's OK to run an image multiple times
6013 simultaneously in read-only mode.)
6015 * systemd-nspawn's --image= option is now capable of
6016 dissecting and booting MBR and GPT disk images that contain
6017 only a single active Linux partition. Previously it
6018 supported only GPT disk images with proper GPT type
6019 IDs. This allows running cloud images from major
6020 distributions directly with systemd-nspawn, without
6023 * In addition to collecting mouse dpi data in the udev
6024 hardware database, there's now support for collecting angle
6025 information for mouse scroll wheels. The database is
6026 supposed to guarantee similar scrolling behavior on mice
6027 that it knows about. There's also support for collecting
6028 information about Touchpad types.
6030 * udev's input_id built-in will now also collect touch screen
6031 dimension data and attach it to probed devices.
6033 * /etc/os-release gained support for a Distribution Privacy
6036 * networkd gained support for creating "ipvlan", "gretap",
6037 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
6039 * systemd-tmpfiles gained support for "a" lines for setting
6042 * systemd-nspawn will now mount /tmp in the container to
6043 tmpfs, automatically.
6045 * systemd now exposes the memory.usage_in_bytes cgroup
6046 attribute and shows it for each service in the "systemctl
6047 status" output, if available.
6049 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
6050 immediate reboot is triggered. This useful if shutdown is
6051 hung and is unable to complete, to expedite the
6052 operation. Note that this kind of reboot will still unmount
6053 all file systems, and hence should not result in fsck being
6056 * A .device unit for an optical block device will now be
6057 considered active only when a medium is in the drive. Also,
6058 mount units are now bound to their backing devices thus
6059 triggering automatic unmounting when devices become
6060 unavailable. With this in place systemd will now
6061 automatically unmount left-over mounts when a CD-ROM is
6062 ejected or an USB stick is yanked from the system.
6064 * networkd-wait-online now has support for waiting for
6065 specific interfaces only (with globbing), and for giving up
6066 after a configurable timeout.
6068 * networkd now exits when idle. It will be automatically
6069 restarted as soon as interfaces show up, are removed or
6070 change state. networkd will stay around as long as there is
6071 at least one DHCP state machine or similar around, that keep
6074 * networkd may now configure IPv6 link-local addressing in
6075 addition to IPv4 link-local addressing.
6077 * The IPv6 "token" for use in SLAAC may now be configured for
6078 each .network interface in networkd.
6080 * Routes configured with networkd may now be assigned a scope
6083 * networkd's [Match] sections now support globbing and lists
6084 of multiple space-separated matches per item.
6086 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
6087 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
6088 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
6089 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
6090 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
6091 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
6092 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
6093 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
6094 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
6095 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
6096 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
6097 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
6098 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
6099 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
6100 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
6101 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
6102 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
6103 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
6104 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
6105 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
6106 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
6107 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
6108 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
6109 Hoffmann, Zbigniew Jędrzejewski-Szmek
6111 — Berlin, 2015-02-16
6115 * When querying unit file enablement status (for example via
6116 "systemctl is-enabled"), a new state "indirect" is now known
6117 which indicates that a unit might not be enabled itself, but
6118 another unit listed in its Also= setting might be.
6120 * Similar to the various existing ConditionXYZ= settings for
6121 units, there are now matching AssertXYZ= settings. While
6122 failing conditions cause a unit to be skipped, but its job
6123 to succeed, failing assertions declared like this will cause
6124 a unit start operation and its job to fail.
6126 * hostnamed now knows a new chassis type "embedded".
6128 * systemctl gained a new "edit" command. When used on a unit
6129 file, this allows extending unit files with .d/ drop-in
6130 configuration snippets or editing the full file (after
6131 copying it from /usr/lib to /etc). This will invoke the
6132 user's editor (as configured with $EDITOR), and reload the
6133 modified configuration after editing.
6135 * "systemctl status" now shows the suggested enablement state
6136 for a unit, as declared in the (usually vendor-supplied)
6137 system preset files.
6139 * nss-myhostname will now resolve the single-label host name
6140 "gateway" to the locally configured default IP routing
6141 gateways, ordered by their metrics. This assigns a stable
6142 name to the used gateways, regardless which ones are
6143 currently configured. Note that the name will only be
6144 resolved after all other name sources (if nss-myhostname is
6145 configured properly) and should hence not negatively impact
6146 systems that use the single-label host name "gateway" in
6149 * systemd-inhibit now allows filtering by mode when listing
6152 * Scope and service units gained a new "Delegate" boolean
6153 property, which, when set, allows processes running inside the
6154 unit to further partition resources. This is primarily
6155 useful for systemd user instances as well as container
6158 * journald will now pick up audit messages directly from
6159 the kernel, and log them like any other log message. The
6160 audit fields are split up and fully indexed. This means that
6161 journalctl in many ways is now a (nicer!) alternative to
6162 ausearch, the traditional audit client. Note that this
6163 implements only a minimal audit client. If you want the
6164 special audit modes like reboot-on-log-overflow, please use
6165 the traditional auditd instead, which can be used in
6166 parallel to journald.
6168 * The ConditionSecurity= unit file option now understands the
6169 special string "audit" to check whether auditing is
6172 * journalctl gained two new commands --vacuum-size= and
6173 --vacuum-time= to delete old journal files until the
6174 remaining ones take up no more than the specified size on disk,
6175 or are not older than the specified time.
6177 * A new, native PPPoE library has been added to sd-network,
6178 systemd's library of light-weight networking protocols. This
6179 library will be used in a future version of networkd to
6180 enable PPPoE communication without an external pppd daemon.
6182 * The busctl tool now understands a new "capture" verb that
6183 works similar to "monitor", but writes a packet capture
6184 trace to STDOUT that can be redirected to a file which is
6185 compatible with libcap's capture file format. This can then
6186 be loaded in Wireshark and similar tools to inspect bus
6189 * The busctl tool now understands a new "tree" verb that shows
6190 the object trees of a specific service on the bus, or of all
6193 * The busctl tool now understands a new "introspect" verb that
6194 shows all interfaces and members of objects on the bus,
6195 including their signature and values. This is particularly
6196 useful to get more information about bus objects shown by
6197 the new "busctl tree" command.
6199 * The busctl tool now understands new verbs "call",
6200 "set-property" and "get-property" for invoking bus method
6201 calls, setting and getting bus object properties in a
6204 * busctl gained a new --augment-creds= argument that controls
6205 whether the tool shall augment credential information it
6206 gets from the bus with data from /proc, in a possibly
6209 * nspawn's --link-journal= switch gained two new values
6210 "try-guest" and "try-host" that work like "guest" and
6211 "host", but do not fail if the host has no persistent
6212 journalling enabled. -j is now equivalent to
6213 --link-journal=try-guest.
6215 * macvlan network devices created by nspawn will now have
6216 stable MAC addresses.
6218 * A new SmackProcessLabel= unit setting has been added, which
6219 controls the SMACK security label processes forked off by
6220 the respective unit shall use.
6222 * If compiled with --enable-xkbcommon, systemd-localed will
6223 verify x11 keymap settings by compiling the given keymap. It
6224 will spew out warnings if the compilation fails. This
6225 requires libxkbcommon to be installed.
6227 * When a coredump is collected, a larger number of metadata
6228 fields is now collected and included in the journal records
6229 created for it. More specifically, control group membership,
6230 environment variables, memory maps, working directory,
6231 chroot directory, /proc/$PID/status, and a list of open file
6232 descriptors is now stored in the log entry.
6234 * The udev hwdb now contains DPI information for mice. For
6237 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
6239 * All systemd programs that read standalone configuration
6240 files in /etc now also support a corresponding series of
6241 .conf.d configuration directories in /etc/, /run/,
6242 /usr/local/lib/, /usr/lib/, and (if configured with
6243 --enable-split-usr) /lib/. In particular, the following
6244 configuration files now have corresponding configuration
6245 directories: system.conf user.conf, logind.conf,
6246 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
6247 resolved.conf, timesyncd.conf, journal-remote.conf, and
6248 journal-upload.conf. Note that distributions should use the
6249 configuration directories in /usr/lib/; the directories in
6250 /etc/ are reserved for the system administrator.
6252 * systemd-rfkill will no longer take the rfkill device name
6253 into account when storing rfkill state on disk, as the name
6254 might be dynamically assigned and not stable. Instead, the
6255 ID_PATH udev variable combined with the rfkill type (wlan,
6256 bluetooth, ...) is used.
6258 * A new service systemd-machine-id-commit.service has been
6259 added. When used on systems where /etc is read-only during
6260 boot, and /etc/machine-id is not initialized (but an empty
6261 file), this service will copy the temporary machine ID
6262 created as replacement into /etc after the system is fully
6263 booted up. This is useful for systems that are freshly
6264 installed with a non-initialized machine ID, but should get
6265 a fixed machine ID for subsequent boots.
6267 * networkd's .netdev files now provide a large set of
6268 configuration parameters for VXLAN devices. Similarly, the
6269 bridge port cost parameter is now configurable in .network
6270 files. There's also new support for configuring IP source
6271 routing. networkd .link files gained support for a new
6272 OriginalName= match that is useful to match against the
6273 original interface name the kernel assigned. .network files
6274 may include MTU= and MACAddress= fields for altering the MTU
6275 and MAC address while being connected to a specific network
6278 * The LUKS logic gained supported for configuring
6279 UUID-specific key files. There's also new support for naming
6280 LUKS device from the kernel command line, using the new
6281 luks.name= argument.
6283 * Timer units may now be transiently created via the bus API
6284 (this was previously already available for scope and service
6285 units). In addition it is now possible to create multiple
6286 transient units at the same time with a single bus call. The
6287 "systemd-run" tool has been updated to make use of this for
6288 running commands on a specified time, in at(1)-style.
6290 * tmpfiles gained support for "t" lines, for assigning
6291 extended attributes to files. Among other uses this may be
6292 used to assign SMACK labels to files.
6294 Contributions from: Alin Rauta, Alison Chaiken, Andrej
6295 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
6296 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
6297 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
6298 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
6299 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
6300 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
6301 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
6302 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
6303 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
6304 Michael Biebl, Michael Chapman, Michael Marineau, Michal
6305 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
6306 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
6307 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
6308 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
6309 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
6310 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
6311 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
6313 — Berlin, 2014-12-10
6317 * journalctl gained the new options -t/--identifier= to match
6318 on the syslog identifier (aka "tag"), as well as --utc to
6319 show log timestamps in the UTC timezone. journalctl now also
6320 accepts -n/--lines=all to disable line capping in a pager.
6322 * journalctl gained a new switch, --flush, that synchronously
6323 flushes logs from /run/log/journal to /var/log/journal if
6324 persistent storage is enabled. systemd-journal-flush.service
6325 now waits until the operation is complete.
6327 * Services can notify the manager before they start a reload
6328 (by sending RELOADING=1) or shutdown (by sending
6329 STOPPING=1). This allows the manager to track and show the
6330 internal state of daemons and closes a race condition when
6331 the process is still running but has closed its D-Bus
6334 * Services with Type=oneshot do not have to have any ExecStart
6337 * User units are now loaded also from
6338 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
6339 /run/systemd/user directory that was already previously
6340 supported, but is under the control of the user.
6342 * Job timeouts (i.e. timeouts on the time a job that is
6343 queued stays in the run queue) can now optionally result in
6344 immediate reboot or power-off actions (JobTimeoutAction= and
6345 JobTimeoutRebootArgument=). This is useful on ".target"
6346 units, to limit the maximum time a target remains
6347 undispatched in the run queue, and to trigger an emergency
6348 operation in such a case. This is now used by default to
6349 turn off the system if boot-up (as defined by everything in
6350 basic.target) hangs and does not complete for at least
6351 15min. Also, if power-off or reboot hang for at least 30min
6352 an immediate power-off/reboot operation is triggered. This
6353 functionality is particularly useful to increase reliability
6354 on embedded devices, but also on laptops which might
6355 accidentally get powered on when carried in a backpack and
6356 whose boot stays stuck in a hard disk encryption passphrase
6359 * systemd-logind can be configured to also handle lid switch
6360 events even when the machine is docked or multiple displays
6361 are attached (HandleLidSwitchDocked= option).
6363 * A helper binary and a service have been added which can be
6364 used to resume from hibernation in the initramfs. A
6365 generator will parse the resume= option on the kernel
6366 command line to trigger resume.
6368 * A user console daemon systemd-consoled has been
6369 added. Currently, it is a preview, and will so far open a
6370 single terminal on each session of the user marked as
6371 Desktop=systemd-console.
6373 * Route metrics can be specified for DHCP routes added by
6376 * The SELinux context of socket-activated services can be set
6377 from the information provided by the networking stack
6378 (SELinuxContextFromNet= option).
6380 * Userspace firmware loading support has been removed and
6381 the minimum supported kernel version is thus bumped to 3.7.
6383 * Timeout for udev workers has been increased from 1 to 3
6384 minutes, but a warning will be printed after 1 minute to
6385 help diagnose kernel modules that take a long time to load.
6387 * Udev rules can now remove tags on devices with TAG-="foobar".
6389 * systemd's readahead implementation has been removed. In many
6390 circumstances it didn't give expected benefits even for
6391 rotational disk drives and was becoming less relevant in the
6392 age of SSDs. As none of the developers has been using
6393 rotating media anymore, and nobody stepped up to actively
6394 maintain this component of systemd it has now been removed.
6396 * Swap units can use Options= to specify discard options.
6397 Discard options specified for swaps in /etc/fstab are now
6400 * Docker containers are now detected as a separate type of
6403 * The Password Agent protocol gained support for queries where
6404 the user input is shown, useful e.g. for user names.
6405 systemd-ask-password gained a new --echo option to turn that
6408 * The default sysctl.d/ snippets will now set:
6410 net.core.default_qdisc = fq_codel
6412 This selects Fair Queuing Controlled Delay as the default
6413 queuing discipline for network interfaces. fq_codel helps
6414 fight the network bufferbloat problem. It is believed to be
6415 a good default with no tuning required for most workloads.
6416 Downstream distributions may override this choice. On 10Gbit
6417 servers that do not do forwarding, "fq" may perform better.
6418 Systems without a good clocksource should use "pfifo_fast".
6420 * If kdbus is enabled during build a new option BusPolicy= is
6421 available for service units, that allows locking all service
6422 processes into a stricter bus policy, in order to limit
6423 access to various bus services, or even hide most of them
6424 from the service's view entirely.
6426 * networkctl will now show the .network and .link file
6427 networkd has applied to a specific interface.
6429 * sd-login gained a new API call sd_session_get_desktop() to
6430 query which desktop environment has been selected for a
6433 * UNIX utmp support is now compile-time optional to support
6434 legacy-free systems.
6436 * systemctl gained two new commands "add-wants" and
6437 "add-requires" for pulling in units from specific targets
6440 * If the word "rescue" is specified on the kernel command line
6441 the system will now boot into rescue mode (aka
6442 rescue.target), which was previously available only by
6443 specifying "1" or "systemd.unit=rescue.target" on the kernel
6444 command line. This new kernel command line option nicely
6445 mirrors the already existing "emergency" kernel command line
6448 * New kernel command line options mount.usr=, mount.usrflags=,
6449 mount.usrfstype= have been added that match root=, rootflags=,
6450 rootfstype= but allow mounting a specific file system to
6453 * The $NOTIFY_SOCKET is now also passed to control processes of
6454 services, not only the main process.
6456 * This version reenables support for fsck's -l switch. This
6457 means at least version v2.25 of util-linux is required for
6458 operation, otherwise dead-locks on device nodes may
6459 occur. Again: you need to update util-linux to at least
6460 v2.25 when updating systemd to v217.
6462 * The "multi-seat-x" tool has been removed from systemd, as
6463 its functionality has been integrated into X servers 1.16,
6464 and the tool is hence redundant. It is recommended to update
6465 display managers invoking this tool to simply invoke X
6466 directly from now on, again.
6468 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
6469 message flag has been added for all of systemd's polkit
6470 authenticated method calls has been added. In particular this
6471 now allows optional interactive authorization via polkit for
6472 many of PID1's privileged operations such as unit file
6473 enabling and disabling.
6475 * "udevadm hwdb --update" learnt a new switch "--usr" for
6476 placing the rebuilt hardware database in /usr instead of
6477 /etc. When used only hardware database entries stored in
6478 /usr will be used, and any user database entries in /etc are
6479 ignored. This functionality is useful for vendors to ship a
6480 pre-built database on systems where local configuration is
6481 unnecessary or unlikely.
6483 * Calendar time specifications in .timer units now also
6484 understand the strings "semi-annually", "quarterly" and
6485 "minutely" as shortcuts (in addition to the preexisting
6486 "anually", "hourly", ...).
6488 * systemd-tmpfiles will now correctly create files in /dev
6489 at boot which are marked for creation only at boot. It is
6490 recommended to always create static device nodes with 'c!'
6491 and 'b!', so that they are created only at boot and not
6492 overwritten at runtime.
6494 * When the watchdog logic is used for a service (WatchdogSec=)
6495 and the watchdog timeout is hit the service will now be
6496 terminated with SIGABRT (instead of just SIGTERM), in order
6497 to make sure a proper coredump and backtrace is
6498 generated. This ensures that hanging services will result in
6499 similar coredump/backtrace behaviour as services that hit a
6502 Contributions from: Andreas Henriksson, Andrei Borzenkov,
6503 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
6504 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
6505 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
6506 Herrmann, David Sommerseth, David Strauss, Emil Renner
6507 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
6508 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
6509 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
6510 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
6511 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
6512 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
6513 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
6514 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
6515 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
6516 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
6517 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
6518 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
6519 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
6520 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
6521 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
6522 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
6525 — Berlin, 2014-10-28
6529 * timedated no longer reads NTP implementation unit names from
6530 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
6531 implementations should add a
6533 Conflicts=systemd-timesyncd.service
6535 to their unit files to take over and replace systemd's NTP
6536 default functionality.
6538 * systemd-sysusers gained a new line type "r" for configuring
6539 which UID/GID ranges to allocate system users/groups
6540 from. Lines of type "u" may now add an additional column
6541 that specifies the home directory for the system user to be
6542 created. Also, systemd-sysusers may now optionally read user
6543 information from STDIN instead of a file. This is useful for
6544 invoking it from RPM preinst scriptlets that need to create
6545 users before the first RPM file is installed since these
6546 files might need to be owned by them. A new
6547 %sysusers_create_inline RPM macro has been introduced to do
6548 just that. systemd-sysusers now updates the shadow files as
6549 well as the user/group databases, which should enhance
6550 compatibility with certain tools like grpck.
6552 * A number of bus APIs of PID 1 now optionally consult polkit to
6553 permit access for otherwise unprivileged clients under certain
6554 conditions. Note that this currently doesn't support
6555 interactive authentication yet, but this is expected to be
6556 added eventually, too.
6558 * /etc/machine-info now has new fields for configuring the
6559 deployment environment of the machine, as well as the
6560 location of the machine. hostnamectl has been updated with
6561 new command to update these fields.
6563 * systemd-timesyncd has been updated to automatically acquire
6564 NTP server information from systemd-networkd, which might
6565 have been discovered via DHCP.
6567 * systemd-resolved now includes a caching DNS stub resolver
6568 and a complete LLMNR name resolution implementation. A new
6569 NSS module "nss-resolve" has been added which can be used
6570 instead of glibc's own "nss-dns" to resolve hostnames via
6571 systemd-resolved. Hostnames, addresses and arbitrary RRs may
6572 be resolved via systemd-resolved D-Bus APIs. In contrast to
6573 the glibc internal resolver systemd-resolved is aware of
6574 multi-homed system, and keeps DNS server and caches separate
6575 and per-interface. Queries are sent simultaneously on all
6576 interfaces that have DNS servers configured, in order to
6577 properly handle VPNs and local LANs which might resolve
6578 separate sets of domain names. systemd-resolved may acquire
6579 DNS server information from systemd-networkd automatically,
6580 which in turn might have discovered them via DHCP. A tool
6581 "systemd-resolve-host" has been added that may be used to
6582 query the DNS logic in resolved. systemd-resolved implements
6583 IDNA and automatically uses IDNA or UTF-8 encoding depending
6584 on whether classic DNS or LLMNR is used as transport. In the
6585 next releases we intend to add a DNSSEC and mDNS/DNS-SD
6586 implementation to systemd-resolved.
6588 * A new NSS module nss-mymachines has been added, that
6589 automatically resolves the names of all local registered
6590 containers to their respective IP addresses.
6592 * A new client tool "networkctl" for systemd-networkd has been
6593 added. It currently is entirely passive and will query
6594 networking configuration from udev, rtnetlink and networkd,
6595 and present it to the user in a very friendly
6596 way. Eventually, we hope to extend it to become a full
6597 control utility for networkd.
6599 * .socket units gained a new DeferAcceptSec= setting that
6600 controls the kernels' TCP_DEFER_ACCEPT sockopt for
6601 TCP. Similarly, support for controlling TCP keep-alive
6602 settings has been added (KeepAliveTimeSec=,
6603 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
6604 turning off Nagle's algorithm on TCP has been added
6607 * logind learned a new session type "web", for use in projects
6608 like Cockpit which register web clients as PAM sessions.
6610 * timer units with at least one OnCalendar= setting will now
6611 be started only after time-sync.target has been
6612 reached. This way they will not elapse before the system
6613 clock has been corrected by a local NTP client or
6614 similar. This is particular useful on RTC-less embedded
6615 machines, that come up with an invalid system clock.
6617 * systemd-nspawn's --network-veth= switch should now result in
6618 stable MAC addresses for both the outer and the inner side
6621 * systemd-nspawn gained a new --volatile= switch for running
6622 container instances with /etc or /var unpopulated.
6624 * The kdbus client code has been updated to use the new Linux
6625 3.17 memfd subsystem instead of the old kdbus-specific one.
6627 * systemd-networkd's DHCP client and server now support
6628 FORCERENEW. There are also new configuration options to
6629 configure the vendor client identifier and broadcast mode
6632 * systemd will no longer inform the kernel about the current
6633 timezone, as this is necessarily incorrect and racy as the
6634 kernel has no understanding of DST and similar
6635 concepts. This hence means FAT timestamps will be always
6636 considered UTC, similar to what Android is already
6637 doing. Also, when the RTC is configured to the local time
6638 (rather than UTC) systemd will never synchronize back to it,
6639 as this might confuse Windows at a later boot.
6641 * systemd-analyze gained a new command "verify" for offline
6642 validation of unit files.
6644 * systemd-networkd gained support for a couple of additional
6645 settings for bonding networking setups. Also, the metric for
6646 statically configured routes may now be configured. For
6647 network interfaces where this is appropriate the peer IP
6648 address may now be configured.
6650 * systemd-networkd's DHCP client will no longer request
6651 broadcasting by default, as this tripped up some networks.
6652 For hardware where broadcast is required the feature should
6653 be switched back on using RequestBroadcast=yes.
6655 * systemd-networkd will now set up IPv4LL addresses (when
6656 enabled) even if DHCP is configured successfully.
6658 * udev will now default to respect network device names given
6659 by the kernel when the kernel indicates that these are
6660 predictable. This behavior can be tweaked by changing
6661 NamePolicy= in the relevant .link file.
6663 * A new library systemd-terminal has been added that
6664 implements full TTY stream parsing and rendering. This
6665 library is supposed to be used later on for implementing a
6666 full userspace VT subsystem, replacing the current kernel
6669 * A new tool systemd-journal-upload has been added to push
6670 journal data to a remote system running
6671 systemd-journal-remote.
6673 * journald will no longer forward all local data to another
6674 running syslog daemon. This change has been made because
6675 rsyslog (which appears to be the most commonly used syslog
6676 implementation these days) no longer makes use of this, and
6677 instead pulls the data out of the journal on its own. Since
6678 forwarding the messages to a non-existent syslog server is
6679 more expensive than we assumed we have now turned this
6680 off. If you run a syslog server that is not a recent rsyslog
6681 version, you have to turn this option on again
6682 (ForwardToSyslog= in journald.conf).
6684 * journald now optionally supports the LZ4 compressor for
6685 larger journal fields. This compressor should perform much
6686 better than XZ which was the previous default.
6688 * machinectl now shows the IP addresses of local containers,
6689 if it knows them, plus the interface name of the container.
6691 * A new tool "systemd-escape" has been added that makes it
6692 easy to escape strings to build unit names and similar.
6694 * sd_notify() messages may now include a new ERRNO= field
6695 which is parsed and collected by systemd and shown among the
6696 "systemctl status" output for a service.
6698 * A new component "systemd-firstboot" has been added that
6699 queries the most basic systemd information (timezone,
6700 hostname, root password) interactively on first
6701 boot. Alternatively it may also be used to provision these
6702 things offline on OS images installed into directories.
6704 * The default sysctl.d/ snippets will now set
6706 net.ipv4.conf.default.promote_secondaries=1
6708 This has the benefit of no flushing secondary IP addresses
6709 when primary addresses are removed.
6711 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
6712 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
6713 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
6714 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
6715 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
6716 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
6717 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
6718 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
6719 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
6720 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
6721 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
6722 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
6723 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
6724 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
6725 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
6727 — Berlin, 2014-08-19
6731 * A new tool systemd-sysusers has been added. This tool
6732 creates system users and groups in /etc/passwd and
6733 /etc/group, based on static declarative system user/group
6734 definitions in /usr/lib/sysusers.d/. This is useful to
6735 enable factory resets and volatile systems that boot up with
6736 an empty /etc directory, and thus need system users and
6737 groups created during early boot. systemd now also ships
6738 with two default sysusers.d/ files for the most basic
6739 users and groups systemd and the core operating system
6742 * A new tmpfiles snippet has been added that rebuilds the
6743 essential files in /etc on boot, should they be missing.
6745 * A directive for ensuring automatic clean-up of
6746 /var/cache/man/ has been removed from the default
6747 configuration. This line should now be shipped by the man
6748 implementation. The necessary change has been made to the
6749 man-db implementation. Note that you need to update your man
6750 implementation to one that ships this line, otherwise no
6751 automatic clean-up of /var/cache/man will take place.
6753 * A new condition ConditionNeedsUpdate= has been added that
6754 may conditionalize services to only run when /etc or /var
6755 are "older" than the vendor operating system resources in
6756 /usr. This is useful for reconstructing or updating /etc
6757 after an offline update of /usr or a factory reset, on the
6758 next reboot. Services that want to run once after such an
6759 update or reset should use this condition and order
6760 themselves before the new systemd-update-done.service, which
6761 will mark the two directories as fully updated. A number of
6762 service files have been added making use of this, to rebuild
6763 the udev hardware database, the journald message catalog and
6764 dynamic loader cache (ldconfig). The systemd-sysusers tool
6765 described above also makes use of this now. With this in
6766 place it is now possible to start up a minimal operating
6767 system with /etc empty cleanly. For more information on the
6768 concepts involved see this recent blog story:
6770 http://0pointer.de/blog/projects/stateless.html
6772 * A new system group "input" has been introduced, and all
6773 input device nodes get this group assigned. This is useful
6774 for system-level software to get access to input devices. It
6775 complements what is already done for "audio" and "video".
6777 * systemd-networkd learnt minimal DHCPv4 server support in
6778 addition to the existing DHCPv4 client support. It also
6779 learnt DHCPv6 client and IPv6 Router Solicitation client
6780 support. The DHCPv4 client gained support for static routes
6781 passed in from the server. Note that the [DHCPv4] section
6782 known in older systemd-networkd versions has been renamed to
6783 [DHCP] and is now also used by the DHCPv6 client. Existing
6784 .network files using settings of this section should be
6785 updated, though compatibility is maintained. Optionally, the
6786 client hostname may now be sent to the DHCP server.
6788 * networkd gained support for vxlan virtual networks as well
6789 as tun/tap and dummy devices.
6791 * networkd gained support for automatic allocation of address
6792 ranges for interfaces from a system-wide pool of
6793 addresses. This is useful for dynamically managing a large
6794 number of interfaces with a single network configuration
6795 file. In particular this is useful to easily assign
6796 appropriate IP addresses to the veth links of a large number
6797 of nspawn instances.
6799 * RPM macros for processing sysusers, sysctl and binfmt
6800 drop-in snippets at package installation time have been
6803 * The /etc/os-release file should now be placed in
6804 /usr/lib/os-release. The old location is automatically
6805 created as symlink. /usr/lib is the more appropriate
6806 location of this file, since it shall actually describe the
6807 vendor operating system shipped in /usr, and not the
6808 configuration stored in /etc.
6810 * .mount units gained a new boolean SloppyOptions= setting
6811 that maps to mount(8)'s -s option which enables permissive
6812 parsing of unknown mount options.
6814 * tmpfiles learnt a new "L+" directive which creates a symlink
6815 but (unlike "L") deletes a pre-existing file first, should
6816 it already exist and not already be the correct
6817 symlink. Similarly, "b+", "c+" and "p+" directives have been
6818 added as well, which create block and character devices, as
6819 well as fifos in the filesystem, possibly removing any
6820 pre-existing files of different types.
6822 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
6823 'argument' field (which so far specified the source to
6824 symlink/copy the files from) is now optional. If omitted the
6825 same file os copied from /usr/share/factory/ suffixed by the
6826 full destination path. This is useful for populating /etc
6827 with essential files, by copying them from vendor defaults
6828 shipped in /usr/share/factory/etc.
6830 * A new command "systemctl preset-all" has been added that
6831 applies the service preset settings to all installed unit
6832 files. A new switch --preset-mode= has been added that
6833 controls whether only enable or only disable operations
6836 * A new command "systemctl is-system-running" has been added
6837 that allows checking the overall state of the system, for
6838 example whether it is fully up and running.
6840 * When the system boots up with an empty /etc, the equivalent
6841 to "systemctl preset-all" is executed during early boot, to
6842 make sure all default services are enabled after a factory
6845 * systemd now contains a minimal preset file that enables the
6846 most basic services systemd ships by default.
6848 * Unit files' [Install] section gained a new DefaultInstance=
6849 field for defining the default instance to create if a
6850 template unit is enabled with no instance specified.
6852 * A new passive target cryptsetup-pre.target has been added
6853 that may be used by services that need to make they run and
6854 finish before the first LUKS cryptographic device is set up.
6856 * The /dev/loop-control and /dev/btrfs-control device nodes
6857 are now owned by the "disk" group by default, opening up
6858 access to this group.
6860 * systemd-coredump will now automatically generate a
6861 stack trace of all core dumps taking place on the system,
6862 based on elfutils' libdw library. This stack trace is logged
6865 * systemd-coredump may now optionally store coredumps directly
6866 on disk (in /var/lib/systemd/coredump, possibly compressed),
6867 instead of storing them unconditionally in the journal. This
6868 mode is the new default. A new configuration file
6869 /etc/systemd/coredump.conf has been added to configure this
6870 and other parameters of systemd-coredump.
6872 * coredumpctl gained a new "info" verb to show details about a
6873 specific coredump. A new switch "-1" has also been added
6874 that makes sure to only show information about the most
6875 recent entry instead of all entries. Also, as the tool is
6876 generally useful now the "systemd-" prefix of the binary
6877 name has been removed. Distributions that want to maintain
6878 compatibility with the old name should add a symlink from
6879 the old name to the new name.
6881 * journald's SplitMode= now defaults to "uid". This makes sure
6882 that unprivileged users can access their own coredumps with
6883 coredumpctl without restrictions.
6885 * New kernel command line options "systemd.wants=" (for
6886 pulling an additional unit during boot), "systemd.mask="
6887 (for masking a specific unit for the boot), and
6888 "systemd.debug-shell" (for enabling the debug shell on tty9)
6889 have been added. This is implemented in the new generator
6890 "systemd-debug-generator".
6892 * systemd-nspawn will now by default filter a couple of
6893 syscalls for containers, among them those required for
6894 kernel module loading, direct x86 IO port access, swap
6895 management, and kexec. Most importantly though
6896 open_by_handle_at() is now prohibited for containers,
6897 closing a hole similar to a recently discussed vulnerability
6898 in docker regarding access to files on file hierarchies the
6899 container should normally not have access to. Note that, for
6900 nspawn, we generally make no security claims anyway (and
6901 this is explicitly documented in the man page), so this is
6902 just a fix for one of the most obvious problems.
6904 * A new man page file-hierarchy(7) has been added that
6905 contains a minimized, modernized version of the file system
6906 layout systemd expects, similar in style to the FHS
6907 specification or hier(5). A new tool systemd-path(1) has
6908 been added to query many of these paths for the local
6911 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
6912 longer done. Since the directory now has a per-user size
6913 limit, and is cleaned on logout this appears unnecessary,
6914 in particular since this now brings the lifecycle of this
6915 directory closer in line with how IPC objects are handled.
6917 * systemd.pc now exports a number of additional directories,
6918 including $libdir (which is useful to identify the library
6919 path for the primary architecture of the system), and a
6920 couple of drop-in directories.
6922 * udev's predictable network interface names now use the dev_port
6923 sysfs attribute, introduced in linux 3.15 instead of dev_id to
6924 distinguish between ports of the same PCI function. dev_id should
6925 only be used for ports using the same HW address, hence the need
6928 * machined has been updated to export the OS version of a
6929 container (read from /etc/os-release and
6930 /usr/lib/os-release) on the bus. This is now shown in
6931 "machinectl status" for a machine.
6933 * A new service setting RestartForceExitStatus= has been
6934 added. If configured to a set of exit signals or process
6935 return values, the service will be restarted when the main
6936 daemon process exits with any of them, regardless of the
6939 * systemctl's -H switch for connecting to remote systemd
6940 machines has been extended so that it may be used to
6941 directly connect to a specific container on the
6942 host. "systemctl -H root@foobar:waldi" will now connect as
6943 user "root" to host "foobar", and then proceed directly to
6944 the container named "waldi". Note that currently you have to
6945 authenticate as user "root" for this to work, as entering
6946 containers is a privileged operation.
6948 Contributions from: Andreas Henriksson, Benjamin Steinwender,
6949 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
6950 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
6951 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
6952 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
6953 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
6954 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
6955 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
6956 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
6957 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
6958 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
6959 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
6961 — Berlin, 2014-07-03
6965 * As an experimental feature, udev now tries to lock the
6966 disk device node (flock(LOCK_SH|LOCK_NB)) while it
6967 executes events for the disk or any of its partitions.
6968 Applications like partitioning programs can lock the
6969 disk device node (flock(LOCK_EX)) and claim temporary
6970 device ownership that way; udev will entirely skip all event
6971 handling for this disk and its partitions. If the disk
6972 was opened for writing, the close will trigger a partition
6973 table rescan in udev's "watch" facility, and if needed
6974 synthesize "change" events for the disk and all its partitions.
6975 This is now unconditionally enabled, and if it turns out to
6976 cause major problems, we might turn it on only for specific
6977 devices, or might need to disable it entirely. Device Mapper
6978 devices are excluded from this logic.
6980 * We temporarily dropped the "-l" switch for fsck invocations,
6981 since they collide with the flock() logic above. util-linux
6982 upstream has been changed already to avoid this conflict,
6983 and we will re-add "-l" as soon as util-linux with this
6984 change has been released.
6986 * The dependency on libattr has been removed. Since a long
6987 time, the extended attribute calls have moved to glibc, and
6988 libattr is thus unnecessary.
6990 * Virtualization detection works without privileges now. This
6991 means the systemd-detect-virt binary no longer requires
6992 CAP_SYS_PTRACE file capabilities, and our daemons can run
6993 with fewer privileges.
6995 * systemd-networkd now runs under its own "systemd-network"
6996 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
6997 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
6998 loses the ability to write to files owned by root this way.
7000 * Similarly, systemd-resolved now runs under its own
7001 "systemd-resolve" user with no capabilities remaining.
7003 * Similarly, systemd-bus-proxyd now runs under its own
7004 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
7006 * systemd-networkd gained support for setting up "veth"
7007 virtual Ethernet devices for container connectivity, as well
7008 as GRE and VTI tunnels.
7010 * systemd-networkd will no longer automatically attempt to
7011 manually load kernel modules necessary for certain tunnel
7012 transports. Instead, it is assumed the kernel loads them
7013 automatically when required. This only works correctly on
7014 very new kernels. On older kernels, please consider adding
7015 the kernel modules to /etc/modules-load.d/ as a work-around.
7017 * The resolv.conf file systemd-resolved generates has been
7018 moved to /run/systemd/resolve/. If you have a symlink from
7019 /etc/resolv.conf, it might be necessary to correct it.
7021 * Two new service settings, ProtectHome= and ProtectSystem=,
7022 have been added. When enabled, they will make the user data
7023 (such as /home) inaccessible or read-only and the system
7024 (such as /usr) read-only, for specific services. This allows
7025 very light-weight per-service sandboxing to avoid
7026 modifications of user data or system files from
7027 services. These two new switches have been enabled for all
7028 of systemd's long-running services, where appropriate.
7030 * Socket units gained new SocketUser= and SocketGroup=
7031 settings to set the owner user and group of AF_UNIX sockets
7032 and FIFOs in the file system.
7034 * Socket units gained a new RemoveOnStop= setting. If enabled,
7035 all FIFOS and sockets in the file system will be removed
7036 when the specific socket unit is stopped.
7038 * Socket units gained a new Symlinks= setting. It takes a list
7039 of symlinks to create to file system sockets or FIFOs
7040 created by the specific Unix sockets. This is useful to
7041 manage symlinks to socket nodes with the same lifecycle as
7044 * The /dev/log socket and /dev/initctl FIFO have been moved to
7045 /run, and have been replaced by symlinks. This allows
7046 connecting to these facilities even if PrivateDevices=yes is
7047 used for a service (which makes /dev/log itself unavailable,
7048 but /run is left). This also has the benefit of ensuring
7049 that /dev only contains device nodes, directories and
7050 symlinks, and nothing else.
7052 * sd-daemon gained two new calls sd_pid_notify() and
7053 sd_pid_notifyf(). They are similar to sd_notify() and
7054 sd_notifyf(), but allow overriding of the source PID of
7055 notification messages if permissions permit this. This is
7056 useful to send notify messages on behalf of a different
7057 process (for example, the parent process). The
7058 systemd-notify tool has been updated to make use of this
7059 when sending messages (so that notification messages now
7060 originate from the shell script invoking systemd-notify and
7061 not the systemd-notify process itself. This should minimize
7062 a race where systemd fails to associate notification
7063 messages to services when the originating process already
7066 * A new "on-abnormal" setting for Restart= has been added. If
7067 set, it will result in automatic restarts on all "abnormal"
7068 reasons for a process to exit, which includes unclean
7069 signals, core dumps, timeouts and watchdog timeouts, but
7070 does not include clean and unclean exit codes or clean
7071 signals. Restart=on-abnormal is an alternative for
7072 Restart=on-failure for services that shall be able to
7073 terminate and avoid restarts on certain errors, by
7074 indicating so with an unclean exit code. Restart=on-failure
7075 or Restart=on-abnormal is now the recommended setting for
7076 all long-running services.
7078 * If the InaccessibleDirectories= service setting points to a
7079 mount point (or if there are any submounts contained within
7080 it), it is now attempted to completely unmount it, to make
7081 the file systems truly unavailable for the respective
7084 * The ReadOnlyDirectories= service setting and
7085 systemd-nspawn's --read-only parameter are now recursively
7086 applied to all submounts, too.
7088 * Mount units may now be created transiently via the bus APIs.
7090 * The support for SysV and LSB init scripts has been removed
7091 from the systemd daemon itself. Instead, it is now
7092 implemented as a generator that creates native systemd units
7093 from these scripts when needed. This enables us to remove a
7094 substantial amount of legacy code from PID 1, following the
7095 fact that many distributions only ship a very small number
7096 of LSB/SysV init scripts nowadays.
7098 * Privileged Xen (dom0) domains are not considered
7099 virtualization anymore by the virtualization detection
7100 logic. After all, they generally have unrestricted access to
7101 the hardware and usually are used to manage the unprivileged
7104 * systemd-tmpfiles gained a new "C" line type, for copying
7105 files or entire directories.
7107 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
7108 lines. So far, they have been non-globbing versions of the
7109 latter, and have thus been redundant. In future, it is
7110 recommended to only use "z". "m" has hence been removed
7111 from the documentation, even though it stays supported.
7113 * A tmpfiles snippet to recreate the most basic structure in
7114 /var has been added. This is enough to create the /var/run →
7115 /run symlink and create a couple of structural
7116 directories. This allows systems to boot up with an empty or
7117 volatile /var. Of course, while with this change, the core OS
7118 now is capable with dealing with a volatile /var, not all
7119 user services are ready for it. However, we hope that sooner
7120 or later, many service daemons will be changed upstream so
7121 that they are able to automatically create their necessary
7122 directories in /var at boot, should they be missing. This is
7123 the first step to allow state-less systems that only require
7124 the vendor image for /usr to boot.
7126 * systemd-nspawn has gained a new --tmpfs= switch to mount an
7127 empty tmpfs instance to a specific directory. This is
7128 particularly useful for making use of the automatic
7129 reconstruction of /var (see above), by passing --tmpfs=/var.
7131 * Access modes specified in tmpfiles snippets may now be
7132 prefixed with "~", which indicates that they shall be masked
7133 by whether the existing file or directory is currently
7134 writable, readable or executable at all. Also, if specified,
7135 the sgid/suid/sticky bits will be masked for all
7138 * A new passive target unit "network-pre.target" has been
7139 added which is useful for services that shall run before any
7140 network is configured, for example firewall scripts.
7142 * The "floppy" group that previously owned the /dev/fd*
7143 devices is no longer used. The "disk" group is now used
7144 instead. Distributions should probably deprecate usage of
7147 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
7148 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
7149 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
7150 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
7151 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
7152 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
7155 — Berlin, 2014-06-11
7159 * A new "systemd-timesyncd" daemon has been added for
7160 synchronizing the system clock across the network. It
7161 implements an SNTP client. In contrast to NTP
7162 implementations such as chrony or the NTP reference server,
7163 this only implements a client side, and does not bother with
7164 the full NTP complexity, focusing only on querying time from
7165 one remote server and synchronizing the local clock to
7166 it. Unless you intend to serve NTP to networked clients or
7167 want to connect to local hardware clocks, this simple NTP
7168 client should be more than appropriate for most
7169 installations. The daemon runs with minimal privileges, and
7170 has been hooked up with networkd to only operate when
7171 network connectivity is available. The daemon saves the
7172 current clock to disk every time a new NTP sync has been
7173 acquired, and uses this to possibly correct the system clock
7174 early at bootup, in order to accommodate for systems that
7175 lack an RTC such as the Raspberry Pi and embedded devices,
7176 and to make sure that time monotonically progresses on these
7177 systems, even if it is not always correct. To make use of
7178 this daemon, a new system user and group "systemd-timesync"
7179 needs to be created on installation of systemd.
7181 * The queue "seqnum" interface of libudev has been disabled, as
7182 it was generally incompatible with device namespacing as
7183 sequence numbers of devices go "missing" if the devices are
7184 part of a different namespace.
7186 * "systemctl list-timers" and "systemctl list-sockets" gained
7187 a --recursive switch for showing units of these types also
7188 for all local containers, similar in style to the already
7189 supported --recursive switch for "systemctl list-units".
7191 * A new RebootArgument= setting has been added for service
7192 units, which may be used to specify a kernel reboot argument
7193 to use when triggering reboots with StartLimitAction=.
7195 * A new FailureAction= setting has been added for service
7196 units which may be used to specify an operation to trigger
7197 when a service fails. This works similarly to
7198 StartLimitAction=, but unlike it, controls what is done
7199 immediately rather than only after several attempts to
7200 restart the service in question.
7202 * hostnamed got updated to also expose the kernel name,
7203 release, and version on the bus. This is useful for
7204 executing commands like hostnamectl with the -H switch.
7205 systemd-analyze makes use of this to properly display
7206 details when running non-locally.
7208 * The bootchart tool can now show cgroup information in the
7209 graphs it generates.
7211 * The CFS CPU quota cgroup attribute is now exposed for
7212 services. The new CPUQuota= switch has been added for this
7213 which takes a percentage value. Setting this will have the
7214 result that a service may never get more CPU time than the
7215 specified percentage, even if the machine is otherwise idle.
7217 * systemd-networkd learned IPIP and SIT tunnel support.
7219 * LSB init scripts exposing a dependency on $network will now
7220 get a dependency on network-online.target rather than simply
7221 network.target. This should bring LSB handling closer to
7222 what it was on SysV systems.
7224 * A new fsck.repair= kernel option has been added to control
7225 how fsck shall deal with unclean file systems at boot.
7227 * The (.ini) configuration file parser will now silently ignore
7228 sections whose names begin with "X-". This may be used to maintain
7229 application-specific extension sections in unit files.
7231 * machined gained a new API to query the IP addresses of
7232 registered containers. "machinectl status" has been updated
7233 to show these addresses in its output.
7235 * A new call sd_uid_get_display() has been added to the
7236 sd-login APIs for querying the "primary" session of a
7237 user. The "primary" session of the user is elected from the
7238 user's sessions and generally a graphical session is
7239 preferred over a text one.
7241 * A minimal systemd-resolved daemon has been added. It
7242 currently simply acts as a companion to systemd-networkd and
7243 manages resolv.conf based on per-interface DNS
7244 configuration, possibly supplied via DHCP. In the long run
7245 we hope to extend this into a local DNSSEC enabled DNS and
7248 * The systemd-networkd-wait-online tool is now enabled by
7249 default. It will delay network-online.target until a network
7250 connection has been configured. The tool primarily integrates
7251 with networkd, but will also make a best effort to make sense
7252 of network configuration performed in some other way.
7254 * Two new service options StartupCPUShares= and
7255 StartupBlockIOWeight= have been added that work similarly to
7256 CPUShares= and BlockIOWeight= however only apply during
7257 system startup. This is useful to prioritize certain services
7258 differently during bootup than during normal runtime.
7260 * hostnamed has been changed to prefer the statically
7261 configured hostname in /etc/hostname (unless set to
7262 'localhost' or empty) over any dynamic one supplied by
7263 dhcp. With this change, the rules for picking the hostname
7264 match more closely the rules of other configuration settings
7265 where the local administrator's configuration in /etc always
7266 overrides any other settings.
7268 Contributions from: Ali H. Caliskan, Alison Chaiken, Bas van
7269 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
7270 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
7271 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
7272 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
7273 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
7274 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
7275 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
7276 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
7277 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
7278 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
7279 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
7280 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
7281 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
7282 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
7283 Lindskog, WaLyong Cho, Will Woods, Zbigniew
7286 — Beijing, 2014-05-28
7290 * When restoring the screen brightness at boot, stay away from
7291 the darkest setting or from the lowest 5% of the available
7292 range, depending on which is the larger value of both. This
7293 should effectively protect the user from rebooting into a
7294 black screen, should the brightness have been set to minimum
7297 * sd-login gained a new sd_machine_get_class() call to
7298 determine the class ("vm" or "container") of a machine
7299 registered with machined.
7301 * sd-login gained new calls
7302 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
7303 to query the identity of the peer of a local AF_UNIX
7304 connection. They operate similarly to their sd_pid_get_xyz()
7307 * PID 1 will now maintain a system-wide system state engine
7308 with the states "starting", "running", "degraded",
7309 "maintenance", "stopping". These states are bound to system
7310 startup, normal runtime, runtime with at least one failed
7311 service, rescue/emergency mode and system shutdown. This
7312 state is shown in the "systemctl status" output when no unit
7313 name is passed. It is useful to determine system state, in
7314 particularly when doing so for many systems or containers at
7317 * A new command "list-machines" has been added to "systemctl"
7318 that lists all local OS containers and shows their system
7319 state (see above), if systemd runs inside of them.
7321 * systemctl gained a new "-r" switch to recursively enumerate
7322 units on all local containers, when used with the
7323 "list-unit" command (which is the default one that is
7324 executed when no parameters are specified).
7326 * The GPT automatic partition discovery logic will now honour
7327 two GPT partition flags: one may be set on a partition to
7328 cause it to be mounted read-only, and the other may be set
7329 on a partition to ignore it during automatic discovery.
7331 * Two new GPT type UUIDs have been added for automatic root
7332 partition discovery, for 32-bit and 64-bit ARM. This is not
7333 particularly useful for discovering the root directory on
7334 these architectures during bare-metal boots (since UEFI is
7335 not common there), but still very useful to allow booting of
7336 ARM disk images in nspawn with the -i option.
7338 * MAC addresses of interfaces created with nspawn's
7339 --network-interface= switch will now be generated from the
7340 machine name, and thus be stable between multiple invocations
7343 * logind will now automatically remove all IPC objects owned
7344 by a user if she or he fully logs out. This makes sure that
7345 users who are logged out cannot continue to consume IPC
7346 resources. This covers SysV memory, semaphores and message
7347 queues as well as POSIX shared memory and message
7348 queues. Traditionally, SysV and POSIX IPC had no lifecycle
7349 limits. With this functionality, that is corrected. This may
7350 be turned off by using the RemoveIPC= switch of logind.conf.
7352 * The systemd-machine-id-setup and tmpfiles tools gained a
7353 --root= switch to operate on a specific root directory,
7356 * journald can now forward logged messages to the TTYs of all
7357 logged in users ("wall"). This is the default for all
7358 emergency messages now.
7360 * A new tool systemd-journal-remote has been added to stream
7361 journal log messages across the network.
7363 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
7364 controller trees are mounted into it. Note that the
7365 directories mounted beneath it are not read-only. This is a
7366 security measure and is particularly useful because glibc
7367 actually includes a search logic to pick any tmpfs it can
7368 find to implement shm_open() if /dev/shm is not available
7369 (which it might very well be in namespaced setups).
7371 * machinectl gained a new "poweroff" command to cleanly power
7372 down a local OS container.
7374 * The PrivateDevices= unit file setting will now also drop the
7375 CAP_MKNOD capability from the capability bound set, and
7376 imply DevicePolicy=closed.
7378 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
7379 comprehensively on all long-running systemd services where
7380 this is appropriate.
7382 * systemd-udevd will now run in a disassociated mount
7383 namespace. To mount directories from udev rules, make sure to
7384 pull in mount units via SYSTEMD_WANTS properties.
7386 * The kdbus support gained support for uploading policy into
7387 the kernel. sd-bus gained support for creating "monitoring"
7388 connections that can eavesdrop into all bus communication
7389 for debugging purposes.
7391 * Timestamps may now be specified in seconds since the UNIX
7392 epoch Jan 1st, 1970 by specifying "@" followed by the value
7395 * Native tcpwrap support in systemd has been removed. tcpwrap
7396 is old code, not really maintained anymore and has serious
7397 shortcomings, and better options such as firewalls
7398 exist. For setups that require tcpwrap usage, please
7399 consider invoking your socket-activated service via tcpd,
7400 like on traditional inetd.
7402 * A new system.conf configuration option
7403 DefaultTimerAccuracySec= has been added that controls the
7404 default AccuracySec= setting of .timer units.
7406 * Timer units gained a new WakeSystem= switch. If enabled,
7407 timers configured this way will cause the system to resume
7408 from system suspend (if the system supports that, which most
7411 * Timer units gained a new Persistent= switch. If enabled,
7412 timers configured this way will save to disk when they have
7413 been last triggered. This information is then used on next
7414 reboot to possible execute overdue timer events, that
7415 could not take place because the system was powered off.
7416 This enables simple anacron-like behaviour for timer units.
7418 * systemctl's "list-timers" will now also list the time a
7419 timer unit was last triggered in addition to the next time
7420 it will be triggered.
7422 * systemd-networkd will now assign predictable IPv4LL
7423 addresses to its local interfaces.
7425 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
7426 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
7427 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
7428 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
7429 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
7430 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
7431 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
7432 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
7435 — Berlin, 2014-03-25
7439 * A new unit file setting RestrictAddressFamilies= has been
7440 added to restrict which socket address families unit
7441 processes gain access to. This takes address family names
7442 like "AF_INET" or "AF_UNIX", and is useful to minimize the
7443 attack surface of services via exotic protocol stacks. This
7444 is built on seccomp system call filters.
7446 * Two new unit file settings RuntimeDirectory= and
7447 RuntimeDirectoryMode= have been added that may be used to
7448 manage a per-daemon runtime directories below /run. This is
7449 an alternative for setting up directory permissions with
7450 tmpfiles snippets, and has the advantage that the runtime
7451 directory's lifetime is bound to the daemon runtime and that
7452 the daemon starts up with an empty directory each time. This
7453 is particularly useful when writing services that drop
7454 privileges using the User= or Group= setting.
7456 * The DeviceAllow= unit setting now supports globbing for
7457 matching against device group names.
7459 * The systemd configuration file system.conf gained new
7460 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
7461 DefaultMemoryAccounting= to globally turn on/off accounting
7462 for specific resources (cgroups) for all units. These
7463 settings may still be overridden individually in each unit
7466 * systemd-gpt-auto-generator is now able to discover /srv and
7467 root partitions in addition to /home and swap partitions. It
7468 also supports LUKS-encrypted partitions now. With this in
7469 place, automatic discovery of partitions to mount following
7470 the Discoverable Partitions Specification
7471 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
7472 is now a lot more complete. This allows booting without
7473 /etc/fstab and without root= on the kernel command line on
7474 systems prepared appropriately.
7476 * systemd-nspawn gained a new --image= switch which allows
7477 booting up disk images and Linux installations on any block
7478 device that follow the Discoverable Partitions Specification
7479 (see above). This means that installations made with
7480 appropriately updated installers may now be started and
7481 deployed using container managers, completely
7482 unmodified. (We hope that libvirt-lxc will add support for
7483 this feature soon, too.)
7485 * systemd-nspawn gained a new --network-macvlan= setting to
7486 set up a private macvlan interface for the
7487 container. Similarly, systemd-networkd gained a new
7488 Kind=macvlan setting in .netdev files.
7490 * systemd-networkd now supports configuring local addresses
7493 * A new tool systemd-network-wait-online has been added to
7494 synchronously wait for network connectivity using
7497 * The sd-bus.h bus API gained a new sd_bus_track object for
7498 tracking the lifecycle of bus peers. Note that sd-bus.h is
7499 still not a public API though (unless you specify
7500 --enable-kdbus on the configure command line, which however
7501 voids your warranty and you get no API stability guarantee).
7503 * The $XDG_RUNTIME_DIR runtime directories for each user are
7504 now individual tmpfs instances, which has the benefit of
7505 introducing separate pools for each user, with individual
7506 size limits, and thus making sure that unprivileged clients
7507 can no longer negatively impact the system or other users by
7508 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
7509 RuntimeDirectorySize= has been introduced that allows
7510 controlling the default size limit for all users. It
7511 defaults to 10% of the available physical memory. This is no
7512 replacement for quotas on tmpfs though (which the kernel
7513 still does not support), as /dev/shm and /tmp are still
7514 shared resources used by both the system and unprivileged
7517 * logind will now automatically turn off automatic suspending
7518 on laptop lid close when more than one display is
7519 connected. This was previously expected to be implemented
7520 individually in desktop environments (such as GNOME),
7521 however has been added to logind now, in order to fix a
7522 boot-time race where a desktop environment might not have
7523 been started yet and thus not been able to take an inhibitor
7524 lock at the time where logind already suspends the system
7525 due to a closed lid.
7527 * logind will now wait at least 30s after each system
7528 suspend/resume cycle, and 3min after system boot before
7529 suspending the system due to a closed laptop lid. This
7530 should give USB docking stations and similar enough time to
7531 be probed and configured after system resume and boot in
7532 order to then act as suspend blocker.
7534 * systemd-run gained a new --property= setting which allows
7535 initialization of resource control properties (and others)
7536 for the created scope or service unit. Example: "systemd-run
7537 --property=BlockIOWeight=10 updatedb" may be used to run
7538 updatedb at a low block IO scheduling weight.
7540 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
7541 now also work in --scope mode.
7543 * When systemd is compiled with kdbus support, basic support
7544 for enforced policies is now in place. (Note that enabling
7545 kdbus still voids your warranty and no API compatibility
7548 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
7549 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
7550 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
7551 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
7552 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
7553 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
7554 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
7555 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
7556 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
7557 Zbigniew Jędrzejewski-Szmek
7559 — Berlin, 2014-03-12
7563 * systemd will now relabel /dev after loading the SMACK policy
7564 according to SMACK rules.
7566 * A new unit file option AppArmorProfile= has been added to
7567 set the AppArmor profile for the processes of a unit.
7569 * A new condition check ConditionArchitecture= has been added
7570 to conditionalize units based on the system architecture, as
7571 reported by uname()'s "machine" field.
7573 * systemd-networkd now supports matching on the system
7574 virtualization, architecture, kernel command line, host name
7577 * logind is now a lot more aggressive when suspending the
7578 machine due to a closed laptop lid. Instead of acting only
7579 on the lid close action, it will continuously watch the lid
7580 status and act on it. This is useful for laptops where the
7581 power button is on the outside of the chassis so that it can
7582 be reached without opening the lid (such as the Lenovo
7583 Yoga). On those machines, logind will now immediately
7584 re-suspend the machine if the power button has been
7585 accidentally pressed while the laptop was suspended and in a
7586 backpack or similar.
7588 * logind will now watch SW_DOCK switches and inhibit reaction
7589 to the lid switch if it is pressed. This means that logind
7590 will not suspend the machine anymore if the lid is closed
7591 and the system is docked, if the laptop supports SW_DOCK
7592 notifications via the input layer. Note that ACPI docking
7593 stations do not generate this currently. Also note that this
7594 logic is usually not fully sufficient and Desktop
7595 Environments should take a lid switch inhibitor lock when an
7596 external display is connected, as systemd will not watch
7599 * nspawn will now make use of the devices cgroup controller by
7600 default, and only permit creation of and access to the usual
7601 API device nodes like /dev/null or /dev/random, as well as
7602 access to (but not creation of) the pty devices.
7604 * We will now ship a default .network file for
7605 systemd-networkd that automatically configures DHCP for
7606 network interfaces created by nspawn's --network-veth or
7607 --network-bridge= switches.
7609 * systemd will now understand the usual M, K, G, T suffixes
7610 according to SI conventions (i.e. to the base 1000) when
7611 referring to throughput and hardware metrics. It will stay
7612 with IEC conventions (i.e. to the base 1024) for software
7613 metrics, according to what is customary according to
7614 Wikipedia. We explicitly document which base applies for
7615 each configuration option.
7617 * The DeviceAllow= setting in unit files now supports a syntax
7618 to whitelist an entire group of devices node majors at once,
7619 based on the /proc/devices listing. For example, with the
7620 string "char-pts", it is now possible to whitelist all
7621 current and future pseudo-TTYs at once.
7623 * sd-event learned a new "post" event source. Event sources of
7624 this type are triggered by the dispatching of any event
7625 source of a type that is not "post". This is useful for
7626 implementing clean-up and check event sources that are
7627 triggered by other work being done in the program.
7629 * systemd-networkd is no longer statically enabled, but uses
7630 the usual [Install] sections so that it can be
7631 enabled/disabled using systemctl. It still is enabled by
7634 * When creating a veth interface pair with systemd-nspawn, the
7635 host side will now be prefixed with "vb-" if
7636 --network-bridge= is used, and with "ve-" if --network-veth
7637 is used. This way, it is easy to distinguish these cases on
7638 the host, for example to apply different configuration to
7639 them with systemd-networkd.
7641 * The compatibility libraries for libsystemd-journal.so,
7642 libsystem-id128.so, libsystemd-login.so and
7643 libsystemd-daemon.so do not make use of IFUNC
7644 anymore. Instead, we now build libsystemd.so multiple times
7645 under these alternative names. This means that the footprint
7646 is drastically increased, but given that these are
7647 transitional compatibility libraries, this should not matter
7648 much. This change has been made necessary to support the ARM
7649 platform for these compatibility libraries, as the ARM
7650 toolchain is not really at the same level as the toolchain
7651 for other architectures like x86 and does not support
7652 IFUNC. Please make sure to use --enable-compat-libs only
7653 during a transitional period!
7655 * The .include syntax has been deprecated and is not documented
7656 anymore. Drop-in files in .d directories should be used instead.
7658 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
7659 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
7660 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
7661 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
7662 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
7663 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
7664 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
7665 Zbigniew Jędrzejewski-Szmek
7667 — Berlin, 2014-02-24
7671 * A new component "systemd-networkd" has been added that can
7672 be used to configure local network interfaces statically or
7673 via DHCP. It is capable of bringing up bridges, VLANs, and
7674 bonding. Currently, no hook-ups for interactive network
7675 configuration are provided. Use this for your initrd,
7676 container, embedded, or server setup if you need a simple,
7677 yet powerful, network configuration solution. This
7678 configuration subsystem is quite nifty, as it allows wildcard
7679 hotplug matching in interfaces. For example, with a single
7680 configuration snippet, you can configure that all Ethernet
7681 interfaces showing up are automatically added to a bridge,
7682 or similar. It supports link-sensing and more.
7684 * A new tool "systemd-socket-proxyd" has been added which can
7685 act as a bidirectional proxy for TCP sockets. This is
7686 useful for adding socket activation support to services that
7687 do not actually support socket activation, including virtual
7688 machines and the like.
7690 * Add a new tool to save/restore rfkill state on
7693 * Save/restore state of keyboard backlights in addition to
7694 display backlights on shutdown/boot.
7696 * udev learned a new SECLABEL{} construct to label device
7697 nodes with a specific security label when they appear. For
7698 now, only SECLABEL{selinux} is supported, but the syntax is
7699 prepared for additional security frameworks.
7701 * udev gained a new scheme to configure link-level attributes
7702 from files in /etc/systemd/network/*.link. These files can
7703 match against MAC address, device path, driver name and type,
7704 and will apply attributes like the naming policy, link speed,
7705 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
7706 address assignment policy (randomized, ...).
7708 * The configuration of network interface naming rules for
7709 "permanent interface names" has changed: a new NamePolicy=
7710 setting in the [Link] section of .link files determines the
7711 priority of possible naming schemes (onboard, slot, MAC,
7712 path). The default value of this setting is determined by
7713 /usr/lib/net/links/99-default.link. Old
7714 80-net-name-slot.rules udev configuration file has been
7715 removed, so local configuration overriding this file should
7716 be adapted to override 99-default.link instead.
7718 * When the User= switch is used in a unit file, also
7719 initialize $SHELL= based on the user database entry.
7721 * systemd no longer depends on libdbus. All communication is
7722 now done with sd-bus, systemd's low-level bus library
7725 * kdbus support has been added to PID 1 itself. When kdbus is
7726 enabled, this causes PID 1 to set up the system bus and
7727 enable support for a new ".busname" unit type that
7728 encapsulates bus name activation on kdbus. It works a little
7729 bit like ".socket" units, except for bus names. A new
7730 generator has been added that converts classic dbus1 service
7731 activation files automatically into native systemd .busname
7734 * sd-bus: add a light-weight vtable implementation that allows
7735 defining objects on the bus with a simple static const
7736 vtable array of its methods, signals and properties.
7738 * systemd will not generate or install static dbus
7739 introspection data anymore to /usr/share/dbus-1/interfaces,
7740 as the precise format of these files is unclear, and
7741 nothing makes use of it.
7743 * A proxy daemon is now provided to proxy clients connecting
7744 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
7745 compatibility with classic D-Bus.
7747 * A bus driver implementation has been added that supports the
7748 classic D-Bus bus driver calls on kdbus, also for
7749 compatibility purposes.
7751 * A new API "sd-event.h" has been added that implements a
7752 minimal event loop API built around epoll. It provides a
7753 couple of features that direct epoll usage is lacking:
7754 prioritization of events, scales to large numbers of timer
7755 events, per-event timer slack (accuracy), system-wide
7756 coalescing of timer events, exit handlers, watchdog
7757 supervision support using systemd's sd_notify() API, child
7760 * A new API "sd-rntl.h" has been added that provides an API
7761 around the route netlink interface of the kernel, similar in
7762 style to "sd-bus.h".
7764 * A new API "sd-dhcp-client.h" has been added that provides a
7765 small DHCPv4 client-side implementation. This is used by
7768 * There is a new kernel command line option
7769 "systemd.restore_state=0|1". When set to "0", none of the
7770 systemd tools will restore saved runtime state to hardware
7771 devices. More specifically, the rfkill and backlight states
7774 * The FsckPassNo= compatibility option in mount/service units
7775 has been removed. The fstab generator will now add the
7776 necessary dependencies automatically, and does not require
7777 PID1's support for that anymore.
7779 * journalctl gained a new switch, --list-boots, that lists
7780 recent boots with their times and boot IDs.
7782 * The various tools like systemctl, loginctl, timedatectl,
7783 busctl, systemd-run, ... have gained a new switch "-M" to
7784 connect to a specific, local OS container (as direct
7785 connection, without requiring SSH). This works on any
7786 container that is registered with machined, such as those
7787 created by libvirt-lxc or nspawn.
7789 * systemd-run and systemd-analyze also gained support for "-H"
7790 to connect to remote hosts via SSH. This is particularly
7791 useful for systemd-run because it enables queuing of jobs
7792 onto remote systems.
7794 * machinectl gained a new command "login" to open a getty
7795 login in any local container. This works with any container
7796 that is registered with machined (such as those created by
7797 libvirt-lxc or nspawn), and which runs systemd inside.
7799 * machinectl gained a new "reboot" command that may be used to
7800 trigger a reboot on a specific container that is registered
7801 with machined. This works on any container that runs an init
7802 system of some kind.
7804 * systemctl gained a new "list-timers" command to print a nice
7805 listing of installed timer units with the times they elapse
7808 * Alternative reboot() parameters may now be specified on the
7809 "systemctl reboot" command line and are passed to the
7810 reboot() system call.
7812 * systemctl gained a new --job-mode= switch to configure the
7813 mode to queue a job with. This is a more generic version of
7814 --fail, --irreversible, and --ignore-dependencies, which are
7815 still available but not advertised anymore.
7817 * /etc/systemd/system.conf gained new settings to configure
7818 various default timeouts of units, as well as the default
7819 start limit interval and burst. These may still be overridden
7822 * PID1 will now export on the bus profile data of the security
7823 policy upload process (such as the SELinux policy upload to
7826 * journald: when forwarding logs to the console, include
7827 timestamps (following the setting in
7828 /sys/module/printk/parameters/time).
7830 * OnCalendar= in timer units now understands the special
7831 strings "yearly" and "annually". (Both are equivalent)
7833 * The accuracy of timer units is now configurable with the new
7834 AccuracySec= setting. It defaults to 1min.
7836 * A new dependency type JoinsNamespaceOf= has been added that
7837 allows running two services within the same /tmp and network
7838 namespace, if PrivateNetwork= or PrivateTmp= are used.
7840 * A new command "cat" has been added to systemctl. It outputs
7841 the original unit file of a unit, and concatenates the
7842 contents of additional "drop-in" unit file snippets, so that
7843 the full configuration is shown.
7845 * systemctl now supports globbing on the various "list-xyz"
7846 commands, like "list-units" or "list-sockets", as well as on
7847 those commands which take multiple unit names.
7849 * journalctl's --unit= switch gained support for globbing.
7851 * All systemd daemons now make use of the watchdog logic so
7852 that systemd automatically notices when they hang.
7854 * If the $container_ttys environment variable is set,
7855 getty-generator will automatically spawn a getty for each
7856 listed tty. This is useful for container managers to request
7857 login gettys to be spawned on as many ttys as needed.
7859 * %h, %s, %U specifier support is not available anymore when
7860 used in unit files for PID 1. This is because NSS calls are
7861 not safe from PID 1. They stay available for --user
7862 instances of systemd, and as special case for the root user.
7864 * loginctl gained a new "--no-legend" switch to turn off output
7867 * The "sd-login.h" API gained three new calls:
7868 sd_session_is_remote(), sd_session_get_remote_user(),
7869 sd_session_get_remote_host() to query information about
7872 * The udev hardware database now also carries vendor/product
7873 information of SDIO devices.
7875 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
7876 determine whether watchdog notifications are requested by
7879 * Socket-activated per-connection services now include a
7880 short description of the connection parameters in the
7883 * tmpfiles gained a new "--boot" option. When this is not used,
7884 only lines where the command character is not suffixed with
7885 "!" are executed. When this option is specified, those
7886 options are executed too. This partitions tmpfiles
7887 directives into those that can be safely executed at any
7888 time, and those which should be run only at boot (for
7889 example, a line that creates /run/nologin).
7891 * A new API "sd-resolve.h" has been added which provides a simple
7892 asynchronous wrapper around glibc NSS host name resolution
7893 calls, such as getaddrinfo(). In contrast to glibc's
7894 getaddrinfo_a(), it does not use signals. In contrast to most
7895 other asynchronous name resolution libraries, this one does
7896 not reimplement DNS, but reuses NSS, so that alternate
7897 host name resolution systems continue to work, such as mDNS,
7898 LDAP, etc. This API is based on libasyncns, but it has been
7899 cleaned up for inclusion in systemd.
7901 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
7902 "sd-daemon.h" are no longer found in individual libraries
7903 libsystemd-journal.so, libsystemd-login.so,
7904 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
7905 merged them into a single library, libsystemd.so, which
7906 provides all symbols. The reason for this is cyclic
7907 dependencies, as these libraries tend to use each other's
7908 symbols. So far, we have managed to workaround that by linking
7909 a copy of a good part of our code into each of these
7910 libraries again and again, which, however, makes certain
7911 things hard to do, like sharing static variables. Also, it
7912 substantially increases footprint. With this change, there
7913 is only one library for the basic APIs systemd
7914 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
7915 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
7916 library as well, however are subject to the --enable-kdbus
7917 switch (see below). Note that "sd-dhcp-client.h" is not part
7918 of this library (this is because it only consumes, never
7919 provides, services of/to other APIs). To make the transition
7920 easy from the separate libraries to the unified one, we
7921 provide the --enable-compat-libs compile-time switch which
7922 will generate stub libraries that are compatible with the
7923 old ones but redirect all calls to the new one.
7925 * All of the kdbus logic and the new APIs "sd-bus.h",
7926 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
7927 and "sd-utf8.h" are compile-time optional via the
7928 "--enable-kdbus" switch, and they are not compiled in by
7929 default. To make use of kdbus, you have to explicitly enable
7930 the switch. Note however, that neither the kernel nor the
7931 userspace API for all of this is considered stable yet. We
7932 want to maintain the freedom to still change the APIs for
7933 now. By specifying this build-time switch, you acknowledge
7934 that you are aware of the instability of the current
7937 * Also, note that while kdbus is pretty much complete,
7938 it lacks one thing: proper policy support. This means you
7939 can build a fully working system with all features; however,
7940 it will be highly insecure. Policy support will be added in
7941 one of the next releases, at the same time that we will
7942 declare the APIs stable.
7944 * When the kernel command line argument "kdbus" is specified,
7945 systemd will automatically load the kdbus.ko kernel module. At
7946 this stage of development, it is only useful for testing kdbus
7947 and should not be used in production. Note: if "--enable-kdbus"
7948 is specified, and the kdbus.ko kernel module is available, and
7949 "kdbus" is added to the kernel command line, the entire system
7950 runs with kdbus instead of dbus-daemon, with the above mentioned
7951 problem of missing the system policy enforcement. Also a future
7952 version of kdbus.ko or a newer systemd will not be compatible with
7953 each other, and will unlikely be able to boot the machine if only
7954 one of them is updated.
7956 * systemctl gained a new "import-environment" command which
7957 uploads the caller's environment (or parts thereof) into the
7958 service manager so that it is inherited by services started
7959 by the manager. This is useful to upload variables like
7960 $DISPLAY into the user service manager.
7962 * A new PrivateDevices= switch has been added to service units
7963 which allows running a service with a namespaced /dev
7964 directory that does not contain any device nodes for
7965 physical devices. More specifically, it only includes devices
7966 such as /dev/null, /dev/urandom, and /dev/zero which are API
7969 * logind has been extended to support behaviour like VT
7970 switching on seats that do not support a VT. This makes
7971 multi-session available on seats that are not the first seat
7972 (seat0), and on systems where kernel support for VTs has
7973 been disabled at compile-time.
7975 * If a process holds a delay lock for system sleep or shutdown
7976 and fails to release it in time, we will now log its
7977 identity. This makes it easier to identify processes that
7978 cause slow suspends or power-offs.
7980 * When parsing /etc/crypttab, support for a new key-slot=
7981 option as supported by Debian is added. It allows indicating
7982 which LUKS slot to use on disk, speeding up key loading.
7984 * The sd_journal_sendv() API call has been checked and
7985 officially declared to be async-signal-safe so that it may
7986 be invoked from signal handlers for logging purposes.
7988 * Boot-time status output is now enabled automatically after a
7989 short timeout if boot does not progress, in order to give
7990 the user an indication what she or he is waiting for.
7992 * The boot-time output has been improved to show how much time
7993 remains until jobs expire.
7995 * The KillMode= switch in service units gained a new possible
7996 value "mixed". If set, and the unit is shut down, then the
7997 initial SIGTERM signal is sent only to the main daemon
7998 process, while the following SIGKILL signal is sent to
7999 all remaining processes of the service.
8001 * When a scope unit is registered, a new property "Controller"
8002 may be set. If set to a valid bus name, systemd will send a
8003 RequestStop() signal to this name when it would like to shut
8004 down the scope. This may be used to hook manager logic into
8005 the shutdown logic of scope units. Also, scope units may now
8006 be put in a special "abandoned" state, in which case the
8007 manager process which created them takes no further
8008 responsibilities for it.
8010 * When reading unit files, systemd will now verify
8011 the access mode of these files, and warn about certain
8012 suspicious combinations. This has been added to make it
8013 easier to track down packaging bugs where unit files are
8014 marked executable or world-writable.
8016 * systemd-nspawn gained a new "--setenv=" switch to set
8017 container-wide environment variables. The similar option in
8018 systemd-activate was renamed from "--environment=" to
8019 "--setenv=" for consistency.
8021 * systemd-nspawn has been updated to create a new kdbus domain
8022 for each container that is invoked, thus allowing each
8023 container to have its own set of system and user buses,
8024 independent of the host.
8026 * systemd-nspawn gained a new --drop-capability= switch to run
8027 the container with less capabilities than the default. Both
8028 --drop-capability= and --capability= now take the special
8029 string "all" for dropping or keeping all capabilities.
8031 * systemd-nspawn gained new switches for executing containers
8032 with specific SELinux labels set.
8034 * systemd-nspawn gained a new --quiet switch to not generate
8035 any additional output but the container's own console
8038 * systemd-nspawn gained a new --share-system switch to run a
8039 container without PID namespacing enabled.
8041 * systemd-nspawn gained a new --register= switch to control
8042 whether the container is registered with systemd-machined or
8043 not. This is useful for containers that do not run full
8044 OS images, but only specific apps.
8046 * systemd-nspawn gained a new --keep-unit which may be used
8047 when invoked as the only program from a service unit, and
8048 results in registration of the unit service itself in
8049 systemd-machined, instead of a newly opened scope unit.
8051 * systemd-nspawn gained a new --network-interface= switch for
8052 moving arbitrary interfaces to the container. The new
8053 --network-veth switch creates a virtual Ethernet connection
8054 between host and container. The new --network-bridge=
8055 switch then allows assigning the host side of this virtual
8056 Ethernet connection to a bridge device.
8058 * systemd-nspawn gained a new --personality= switch for
8059 setting the kernel personality for the container. This is
8060 useful when running a 32-bit container on a 64-bit host. A
8061 similar option Personality= is now also available for service
8064 * logind will now also track a "Desktop" identifier for each
8065 session which encodes the desktop environment of it. This is
8066 useful for desktop environments that want to identify
8067 multiple running sessions of itself easily.
8069 * A new SELinuxContext= setting for service units has been
8070 added that allows setting a specific SELinux execution
8071 context for a service.
8073 * Most systemd client tools will now honour $SYSTEMD_LESS for
8074 settings of the "less" pager. By default, these tools will
8075 override $LESS to allow certain operations to work, such as
8076 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
8077 influence this logic.
8079 * systemd's "seccomp" hook-up has been changed to make use of
8080 the libseccomp library instead of using its own
8081 implementation. This has benefits for portability among
8084 * For usage together with SystemCallFilter=, a new
8085 SystemCallErrorNumber= setting has been introduced that
8086 allows configuration of a system error number to be returned
8087 on filtered system calls, instead of immediately killing the
8088 process. Also, SystemCallArchitectures= has been added to
8089 limit access to system calls of a particular architecture
8090 (in order to turn off support for unused secondary
8091 architectures). There is also a global
8092 SystemCallArchitectures= setting in system.conf now to turn
8093 off support for non-native system calls system-wide.
8095 * systemd requires a kernel with a working name_to_handle_at(),
8096 please see the kernel config requirements in the README file.
8098 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
8099 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
8100 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
8101 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
8102 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
8103 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
8104 Elia Pinto, Florian Weimer, George McCollister, Goffredo
8105 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
8106 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
8107 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
8108 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
8109 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
8110 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
8111 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
8112 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
8113 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
8114 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
8115 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
8116 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
8117 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
8118 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
8119 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
8120 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
8121 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
8123 — Berlin, 2014-02-20
8127 * logind has gained support for facilitating privileged input
8128 and drm device access for unprivileged clients. This work is
8129 useful to allow Wayland display servers (and similar
8130 programs, such as kmscon) to run under the user's ID and
8131 access input and drm devices which are normally
8132 protected. When this is used (and the kernel is new enough)
8133 logind will "mute" IO on the file descriptors passed to
8134 Wayland as long as it is in the background and "unmute" it
8135 if it returns into the foreground. This allows secure
8136 session switching without allowing background sessions to
8137 eavesdrop on input and display data. This also introduces
8138 session switching support if VT support is turned off in the
8139 kernel, and on seats that are not seat0.
8141 * A new kernel command line option luks.options= is understood
8142 now which allows specifying LUKS options for usage for LUKS
8143 encrypted partitions specified with luks.uuid=.
8145 * tmpfiles.d(5) snippets may now use specifier expansion in
8146 path names. More specifically %m, %b, %H, %v, are now
8147 replaced by the local machine id, boot id, hostname, and
8148 kernel version number.
8150 * A new tmpfiles.d(5) command "m" has been introduced which
8151 may be used to change the owner/group/access mode of a file
8152 or directory if it exists, but do nothing if it does not.
8154 * This release removes high-level support for the
8155 MemorySoftLimit= cgroup setting. The underlying kernel
8156 cgroup attribute memory.soft_limit= is currently badly
8157 designed and likely to be removed from the kernel API in its
8158 current form, hence we should not expose it for now.
8160 * The memory.use_hierarchy cgroup attribute is now enabled for
8161 all cgroups systemd creates in the memory cgroup
8162 hierarchy. This option is likely to be come the built-in
8163 default in the kernel anyway, and the non-hierarchical mode
8164 never made much sense in the intrinsically hierarchical
8167 * A new field _SYSTEMD_SLICE= is logged along with all journal
8168 messages containing the slice a message was generated
8169 from. This is useful to allow easy per-customer filtering of
8170 logs among other things.
8172 * systemd-journald will no longer adjust the group of journal
8173 files it creates to the "systemd-journal" group. Instead we
8174 rely on the journal directory to be owned by the
8175 "systemd-journal" group, and its setgid bit set, so that the
8176 kernel file system layer will automatically enforce that
8177 journal files inherit this group assignment. The reason for
8178 this change is that we cannot allow NSS look-ups from
8179 journald which would be necessary to resolve
8180 "systemd-journal" to a numeric GID, because this might
8181 create deadlocks if NSS involves synchronous queries to
8182 other daemons (such as nscd, or sssd) which in turn are
8183 logging clients of journald and might block on it, which
8184 would then dead lock. A tmpfiles.d(5) snippet included in
8185 systemd will make sure the setgid bit and group are
8186 properly set on the journal directory if it exists on every
8187 boot. However, we recommend adjusting it manually after
8188 upgrades too (or from RPM scriptlets), so that the change is
8189 not delayed until next reboot.
8191 * Backlight and random seed files in /var/lib/ have moved into
8192 the /var/lib/systemd/ directory, in order to centralize all
8193 systemd generated files in one directory.
8195 * Boot time performance measurements (as displayed by
8196 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
8197 performance information if that's available to determine how
8198 much time BIOS and boot loader initialization required. With
8199 a sufficiently new BIOS you hence no longer need to boot
8200 with Gummiboot to get access to such information.
8202 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
8203 Cristian Rodríguez, Dave Reisner, David Herrmann, David
8204 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
8205 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
8206 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
8207 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
8208 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
8210 — Berlin, 2013-10-02
8214 * The Restart= option for services now understands a new
8215 on-watchdog setting, which will restart the service
8216 automatically if the service stops sending out watchdog keep
8217 alive messages (as configured with WatchdogSec=).
8219 * The getty generator (which is responsible for bringing up a
8220 getty on configured serial consoles) will no longer only
8221 start a getty on the primary kernel console but on all
8222 others, too. This makes the order in which console= is
8223 specified on the kernel command line less important.
8225 * libsystemd-logind gained a new sd_session_get_vt() call to
8226 retrieve the VT number of a session.
8228 * If the option "tries=0" is set for an entry of /etc/crypttab
8229 its passphrase is queried indefinitely instead of any
8230 maximum number of tries.
8232 * If a service with a configure PID file terminates its PID
8233 file will now be removed automatically if it still exists
8234 afterwards. This should put an end to stale PID files.
8236 * systemd-run will now also take relative binary path names
8237 for execution and no longer insists on absolute paths.
8239 * InaccessibleDirectories= and ReadOnlyDirectories= now take
8240 paths that are optionally prefixed with "-" to indicate that
8241 it should not be considered a failure if they do not exist.
8243 * journalctl -o (and similar commands) now understands a new
8244 output mode "short-precise", it is similar to "short" but
8245 shows timestamps with usec accuracy.
8247 * The option "discard" (as known from Debian) is now
8248 synonymous to "allow-discards" in /etc/crypttab. In fact,
8249 "discard" is preferred now (since it is easier to remember
8252 * Some licensing clean-ups were made, so that more code is now
8253 LGPL-2.1 licensed than before.
8255 * A minimal tool to save/restore the display backlight
8256 brightness across reboots has been added. It will store the
8257 backlight setting as late as possible at shutdown, and
8258 restore it as early as possible during reboot.
8260 * A logic to automatically discover and enable home and swap
8261 partitions on GPT disks has been added. With this in place
8262 /etc/fstab becomes optional for many setups as systemd can
8263 discover certain partitions located on the root disk
8264 automatically. Home partitions are recognized under their
8265 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
8266 partitions are recognized under their GPT type ID
8267 0657fd6da4ab43c484e50933c84b4f4f.
8269 * systemd will no longer pass any environment from the kernel
8270 or initrd to system services. If you want to set an
8271 environment for all services, do so via the kernel command
8272 line systemd.setenv= assignment.
8274 * The systemd-sysctl tool no longer natively reads the file
8275 /etc/sysctl.conf. If desired, the file should be symlinked
8276 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
8277 legacy support by a symlink rather than built-in code, it
8278 also makes the otherwise hidden order of application of the
8279 different files visible. (Note that this partly reverts to a
8280 pre-198 application order of sysctl knobs!)
8282 * The "systemctl set-log-level" and "systemctl dump" commands
8283 have been moved to systemd-analyze.
8285 * systemd-run learned the new --remain-after-exit switch,
8286 which causes the scope unit not to be cleaned up
8287 automatically after the process terminated.
8289 * tmpfiles learned a new --exclude-prefix= switch to exclude
8290 certain paths from operation.
8292 * journald will now automatically flush all messages to disk
8293 as soon as a message at the log level CRIT, ALERT or EMERG
8296 Contributions from: Andrew Cook, Brandon Philips, Christian
8297 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
8298 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
8299 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
8300 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
8301 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
8302 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
8303 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
8304 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
8305 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
8306 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
8307 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
8308 William Giokas, Zbigniew Jędrzejewski-Szmek
8310 — Berlin, 2013-09-13
8314 * The documentation has been updated to cover the various new
8315 concepts introduced with 205.
8317 * Unit files now understand the new %v specifier which
8318 resolves to the kernel version string as returned by "uname
8321 * systemctl now supports filtering the unit list output by
8322 load state, active state and sub state, using the new
8325 * "systemctl status" will now show the results of the
8326 condition checks (like ConditionPathExists= and similar) of
8327 the last start attempts of the unit. They are also logged to
8330 * "journalctl -b" may now be used to look for boot output of a
8331 specific boot. Try "journalctl -b -1" for the previous boot,
8332 but the syntax is substantially more powerful.
8334 * "journalctl --show-cursor" has been added which prints the
8335 cursor string the last shown log line. This may then be used
8336 with the new "journalctl --after-cursor=" switch to continue
8337 browsing logs from that point on.
8339 * "journalctl --force" may now be used to force regeneration
8342 * Creation of "dead" device nodes has been moved from udev
8343 into kmod and tmpfiles. Previously, udev would read the kmod
8344 databases to pre-generate dead device nodes based on meta
8345 information contained in kernel modules, so that these would
8346 be auto-loaded on access rather then at boot. As this
8347 does not really have much to do with the exposing actual
8348 kernel devices to userspace this has always been slightly
8349 alien in the udev codebase. Following the new scheme kmod
8350 will now generate a runtime snippet for tmpfiles from the
8351 module meta information and it now is tmpfiles' job to the
8352 create the nodes. This also allows overriding access and
8353 other parameters for the nodes using the usual tmpfiles
8354 facilities. As side effect this allows us to remove the
8355 CAP_SYS_MKNOD capability bit from udevd entirely.
8357 * logind's device ACLs may now be applied to these "dead"
8358 devices nodes too, thus finally allowing managed access to
8359 devices such as /dev/snd/sequencer without loading the
8360 backing module right-away.
8362 * A new RPM macro has been added that may be used to apply
8363 tmpfiles configuration during package installation.
8365 * systemd-detect-virt and ConditionVirtualization= now can
8366 detect User-Mode-Linux machines (UML).
8368 * journald will now implicitly log the effective capabilities
8369 set of processes in the message metadata.
8371 * systemd-cryptsetup has gained support for TrueCrypt volumes.
8373 * The initrd interface has been simplified (more specifically,
8374 support for passing performance data via environment
8375 variables and fsck results via files in /run has been
8376 removed). These features were non-essential, and are
8377 nowadays available in a much nicer way by having systemd in
8378 the initrd serialize its state and have the hosts systemd
8379 deserialize it again.
8381 * The udev "keymap" data files and tools to apply keyboard
8382 specific mappings of scan to key codes, and force-release
8383 scan code lists have been entirely replaced by a udev
8384 "keyboard" builtin and a hwdb data file.
8386 * systemd will now honour the kernel's "quiet" command line
8387 argument also during late shutdown, resulting in a
8388 completely silent shutdown when used.
8390 * There's now an option to control the SO_REUSEPORT socket
8391 option in .socket units.
8393 * Instance units will now automatically get a per-template
8394 subslice of system.slice unless something else is explicitly
8395 configured. For example, instances of sshd@.service will now
8396 implicitly be placed in system-sshd.slice rather than
8397 system.slice as before.
8399 * Test coverage support may now be enabled at build time.
8401 Contributions from: Dave Reisner, Frederic Crozat, Harald
8402 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
8403 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
8404 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
8405 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
8406 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
8407 Giokas, Zbigniew Jędrzejewski-Szmek
8409 — Berlin, 2013-07-23
8413 * Two new unit types have been introduced:
8415 Scope units are very similar to service units, however, are
8416 created out of pre-existing processes — instead of PID 1
8417 forking off the processes. By using scope units it is
8418 possible for system services and applications to group their
8419 own child processes (worker processes) in a powerful way
8420 which then maybe used to organize them, or kill them
8421 together, or apply resource limits on them.
8423 Slice units may be used to partition system resources in an
8424 hierarchical fashion and then assign other units to them. By
8425 default there are now three slices: system.slice (for all
8426 system services), user.slice (for all user sessions),
8427 machine.slice (for VMs and containers).
8429 Slices and scopes have been introduced primarily in
8430 context of the work to move cgroup handling to a
8431 single-writer scheme, where only PID 1
8432 creates/removes/manages cgroups.
8434 * There's a new concept of "transient" units. In contrast to
8435 normal units these units are created via an API at runtime,
8436 not from configuration from disk. More specifically this
8437 means it is now possible to run arbitrary programs as
8438 independent services, with all execution parameters passed
8439 in via bus APIs rather than read from disk. Transient units
8440 make systemd substantially more dynamic then it ever was,
8441 and useful as a general batch manager.
8443 * logind has been updated to make use of scope and slice units
8444 for managing user sessions. As a user logs in he will get
8445 his own private slice unit, to which all sessions are added
8446 as scope units. We also added support for automatically
8447 adding an instance of user@.service for the user into the
8448 slice. Effectively logind will no longer create cgroup
8449 hierarchies on its own now, it will defer entirely to PID 1
8450 for this by means of scope, service and slice units. Since
8451 user sessions this way become entities managed by PID 1
8452 the output of "systemctl" is now a lot more comprehensive.
8454 * A new mini-daemon "systemd-machined" has been added which
8455 may be used by virtualization managers to register local
8456 VMs/containers. nspawn has been updated accordingly, and
8457 libvirt will be updated shortly. machined will collect a bit
8458 of meta information about the VMs/containers, and assign
8459 them their own scope unit (see above). The collected
8460 meta-data is then made available via the "machinectl" tool,
8461 and exposed in "ps" and similar tools. machined/machinectl
8462 is compile-time optional.
8464 * As discussed earlier, the low-level cgroup configuration
8465 options ControlGroup=, ControlGroupModify=,
8466 ControlGroupPersistent=, ControlGroupAttribute= have been
8467 removed. Please use high-level attribute settings instead as
8468 well as slice units.
8470 * A new bus call SetUnitProperties() has been added to alter
8471 various runtime parameters of a unit. This is primarily
8472 useful to alter cgroup parameters dynamically in a nice way,
8473 but will be extended later on to make more properties
8474 modifiable at runtime. systemctl gained a new set-properties
8475 command that wraps this call.
8477 * A new tool "systemd-run" has been added which can be used to
8478 run arbitrary command lines as transient services or scopes,
8479 while configuring a number of settings via the command
8480 line. This tool is currently very basic, however already
8481 very useful. We plan to extend this tool to even allow
8482 queuing of execution jobs with time triggers from the
8483 command line, similar in fashion to "at".
8485 * nspawn will now inform the user explicitly that kernels with
8486 audit enabled break containers, and suggest the user to turn
8489 * Support for detecting the IMA and AppArmor security
8490 frameworks with ConditionSecurity= has been added.
8492 * journalctl gained a new "-k" switch for showing only kernel
8493 messages, mimicking dmesg output; in addition to "--user"
8494 and "--system" switches for showing only user's own logs
8497 * systemd-delta can now show information about drop-in
8498 snippets extending unit files.
8500 * libsystemd-bus has been substantially updated but is still
8501 not available as public API.
8503 * systemd will now look for the "debug" argument on the kernel
8504 command line and enable debug logging, similar to what
8505 "systemd.log_level=debug" already did before.
8507 * "systemctl set-default", "systemctl get-default" has been
8508 added to configure the default.target symlink, which
8509 controls what to boot into by default.
8511 * "systemctl set-log-level" has been added as a convenient
8512 way to raise and lower systemd logging threshold.
8514 * "systemd-analyze plot" will now show the time the various
8515 generators needed for execution, as well as information
8516 about the unit file loading.
8518 * libsystemd-journal gained a new sd_journal_open_files() call
8519 for opening specific journal files. journactl also gained a
8520 new switch to expose this new functionality. Previously we
8521 only supported opening all files from a directory, or all
8522 files from the system, as opening individual files only is
8523 racy due to journal file rotation.
8525 * systemd gained the new DefaultEnvironment= setting in
8526 /etc/systemd/system.conf to set environment variables for
8529 * If a privileged process logs a journal message with the
8530 OBJECT_PID= field set, then journald will automatically
8531 augment this with additional OBJECT_UID=, OBJECT_GID=,
8532 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
8533 system services want to log events about specific client
8534 processes. journactl/systemctl has been updated to make use
8535 of this information if all log messages regarding a specific
8538 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
8539 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
8540 Reisner, David Coppa, David King, David Strauss, Eelco
8541 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
8542 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
8543 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
8544 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
8545 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
8546 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
8547 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
8548 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
8549 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
8550 Łukasz Stelmach, 장동준
8554 * The Python bindings gained some minimal support for the APIs
8555 exposed by libsystemd-logind.
8557 * ConditionSecurity= gained support for detecting SMACK. Since
8558 this condition already supports SELinux and AppArmor we only
8559 miss IMA for this. Patches welcome!
8561 Contributions from: Karol Lewandowski, Lennart Poettering,
8562 Zbigniew Jędrzejewski-Szmek
8566 * systemd-nspawn will now create /etc/resolv.conf if
8567 necessary, before bind-mounting the host's file onto it.
8569 * systemd-nspawn will now store meta information about a
8570 container on the container's cgroup as extended attribute
8571 fields, including the root directory.
8573 * The cgroup hierarchy has been reworked in many ways. All
8574 objects any of the components systemd creates in the cgroup
8575 tree are now suffixed. More specifically, user sessions are
8576 now placed in cgroups suffixed with ".session", users in
8577 cgroups suffixed with ".user", and nspawn containers in
8578 cgroups suffixed with ".nspawn". Furthermore, all cgroup
8579 names are now escaped in a simple scheme to avoid collision
8580 of userspace object names with kernel filenames. This work
8581 is preparation for making these objects relocatable in the
8582 cgroup tree, in order to allow easy resource partitioning of
8583 these objects without causing naming conflicts.
8585 * systemctl list-dependencies gained the new switches
8586 --plain, --reverse, --after and --before.
8588 * systemd-inhibit now shows the process name of processes that
8589 have taken an inhibitor lock.
8591 * nss-myhostname will now also resolve "localhost"
8592 implicitly. This makes /etc/hosts an optional file and
8593 nicely handles that on IPv6 ::1 maps to both "localhost" and
8596 * libsystemd-logind.so gained a new call
8597 sd_get_machine_names() to enumerate running containers and
8598 VMs (currently only supported by very new libvirt and
8599 nspawn). sd_login_monitor can now be used to watch
8600 VMs/containers coming and going.
8602 * .include is not allowed recursively anymore, and only in
8603 unit files. Usually it is better to use drop-in snippets in
8604 .d/*.conf anyway, as introduced with systemd 198.
8606 * systemd-analyze gained a new "critical-chain" command that
8607 determines the slowest chain of units run during system
8608 boot-up. It is very useful for tracking down where
8609 optimizing boot time is the most beneficial.
8611 * systemd will no longer allow manipulating service paths in
8612 the name=systemd:/system cgroup tree using ControlGroup= in
8613 units. (But is still fine with it in all other dirs.)
8615 * There's a new systemd-nspawn@.service service file that may
8616 be used to easily run nspawn containers as system
8617 services. With the container's root directory in
8618 /var/lib/container/foobar it is now sufficient to run
8619 "systemctl start systemd-nspawn@foobar.service" to boot it.
8621 * systemd-cgls gained a new parameter "--machine" to list only
8622 the processes within a certain container.
8624 * ConditionSecurity= now can check for "apparmor". We still
8625 are lacking checks for SMACK and IMA for this condition
8626 check though. Patches welcome!
8628 * A new configuration file /etc/systemd/sleep.conf has been
8629 added that may be used to configure which kernel operation
8630 systemd is supposed to execute when "suspend", "hibernate"
8631 or "hybrid-sleep" is requested. This makes the new kernel
8632 "freeze" state accessible to the user.
8634 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
8635 the passed argument if applicable.
8637 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
8638 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
8639 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
8640 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
8641 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
8642 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
8643 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
8648 * The output of 'systemctl list-jobs' got some polishing. The
8649 '--type=' argument may now be passed more than once. A new
8650 command 'systemctl list-sockets' has been added which shows
8651 a list of kernel sockets systemd is listening on with the
8652 socket units they belong to, plus the units these socket
8655 * The experimental libsystemd-bus library got substantial
8656 updates to work in conjunction with the (also experimental)
8657 kdbus kernel project. It works well enough to exchange
8658 messages with some sophistication. Note that kdbus is not
8659 ready yet, and the library is mostly an elaborate test case
8660 for now, and not installable.
8662 * systemd gained a new unit 'systemd-static-nodes.service'
8663 that generates static device nodes earlier during boot, and
8664 can run in conjunction with udev.
8666 * libsystemd-login gained a new call sd_pid_get_user_unit()
8667 to retrieve the user systemd unit a process is running
8668 in. This is useful for systems where systemd is used as
8671 * systemd-nspawn now places all containers in the new /machine
8672 top-level cgroup directory in the name=systemd
8673 hierarchy. libvirt will soon do the same, so that we get a
8674 uniform separation of /system, /user and /machine for system
8675 services, user processes and containers/virtual
8676 machines. This new cgroup hierarchy is also useful to stick
8677 stable names to specific container instances, which can be
8678 recognized later this way (this name may be controlled
8679 via systemd-nspawn's new -M switch). libsystemd-login also
8680 gained a new call sd_pid_get_machine_name() to retrieve the
8681 name of the container/VM a specific process belongs to.
8683 * bootchart can now store its data in the journal.
8685 * libsystemd-journal gained a new call
8686 sd_journal_add_conjunction() for AND expressions to the
8687 matching logic. This can be used to express more complex
8688 logical expressions.
8690 * journactl can now take multiple --unit= and --user-unit=
8693 * The cryptsetup logic now understands the "luks.key=" kernel
8694 command line switch for specifying a file to read the
8695 decryption key from. Also, if a configured key file is not
8696 found the tool will now automatically fall back to prompting
8699 * Python systemd.journal module was updated to wrap recently
8700 added functions from libsystemd-journal. The interface was
8701 changed to bring the low level interface in s.j._Reader
8702 closer to the C API, and the high level interface in
8703 s.j.Reader was updated to wrap and convert all data about
8706 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
8707 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
8708 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
8709 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
8710 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
8711 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
8715 * journalctl --update-catalog now understands a new --root=
8716 option to operate on catalogs found in a different root
8719 * During shutdown after systemd has terminated all running
8720 services a final killing loop kills all remaining left-over
8721 processes. We will now print the name of these processes
8722 when we send SIGKILL to them, since this usually indicates a
8725 * If /etc/crypttab refers to password files stored on
8726 configured mount points automatic dependencies will now be
8727 generated to ensure the specific mount is established first
8728 before the key file is attempted to be read.
8730 * 'systemctl status' will now show information about the
8731 network sockets a socket unit is listening on.
8733 * 'systemctl status' will also shown information about any
8734 drop-in configuration file for units. (Drop-In configuration
8735 files in this context are files such as
8736 /etc/systemd/systemd/foobar.service.d/*.conf)
8738 * systemd-cgtop now optionally shows summed up CPU times of
8739 cgroups. Press '%' while running cgtop to switch between
8740 percentage and absolute mode. This is useful to determine
8741 which cgroups use up the most CPU time over the entire
8742 runtime of the system. systemd-cgtop has also been updated
8743 to be 'pipeable' for processing with further shell tools.
8745 * 'hostnamectl set-hostname' will now allow setting of FQDN
8748 * The formatting and parsing of time span values has been
8749 changed. The parser now understands fractional expressions
8750 such as "5.5h". The formatter will now output fractional
8751 expressions for all time spans under 1min, i.e. "5.123456s"
8752 rather than "5s 123ms 456us". For time spans under 1s
8753 millisecond values are shown, for those under 1ms
8754 microsecond values are shown. This should greatly improve
8755 all time-related output of systemd.
8757 * libsystemd-login and libsystemd-journal gained new
8758 functions for querying the poll() events mask and poll()
8759 timeout value for integration into arbitrary event
8762 * localectl gained the ability to list available X11 keymaps
8763 (models, layouts, variants, options).
8765 * 'systemd-analyze dot' gained the ability to filter for
8766 specific units via shell-style globs, to create smaller,
8767 more useful graphs. I.e. it is now possible to create simple
8768 graphs of all the dependencies between only target units, or
8769 of all units that Avahi has dependencies with.
8771 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
8772 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
8773 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
8774 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
8775 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
8776 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
8777 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
8781 * The boot-time readahead implementation for rotating media
8782 will now read the read-ahead data in multiple passes which
8783 consist of all read requests made in equidistant time
8784 intervals. This means instead of strictly reading read-ahead
8785 data in its physical order on disk we now try to find a
8786 middle ground between physical and access time order.
8788 * /etc/os-release files gained a new BUILD_ID= field for usage
8789 on operating systems that provide continuous builds of OS
8792 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
8793 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
8794 William Douglas, Zbigniew Jędrzejewski-Szmek
8798 * systemd-python gained an API exposing libsystemd-daemon.
8800 * The SMACK setup logic gained support for uploading CIPSO
8803 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
8804 ReadOnlyDirectories= and InaccessibleDirectories= has
8805 changed. The private /tmp and /var/tmp directories are now
8806 shared by all processes of a service (which means
8807 ExecStartPre= may now leave data in /tmp that ExecStart= of
8808 the same service can still access). When a service is
8809 stopped its temporary directories are immediately deleted
8810 (normal clean-up with tmpfiles is still done in addition to
8813 * By default, systemd will now set a couple of sysctl
8814 variables in the kernel: the safe sysrq options are turned
8815 on, IP route verification is turned on, and source routing
8816 disabled. The recently added hardlink and softlink
8817 protection of the kernel is turned on. These settings should
8818 be reasonably safe, and good defaults for all new systems.
8820 * The predictable network naming logic may now be turned off
8821 with a new kernel command line switch: net.ifnames=0.
8823 * A new libsystemd-bus module has been added that implements a
8824 pretty complete D-Bus client library. For details see:
8826 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
8828 * journald will now explicitly flush the journal files to disk
8829 at the latest 5min after each write. The file will then also
8830 be marked offline until the next write. This should increase
8831 reliability in case of a crash. The synchronization delay
8832 can be configured via SyncIntervalSec= in journald.conf.
8834 * There's a new remote-fs-setup.target unit that can be used
8835 to pull in specific services when at least one remote file
8836 system is to be mounted.
8838 * There are new targets timers.target and paths.target as
8839 canonical targets to pull user timer and path units in
8840 from. This complements sockets.target with a similar
8841 purpose for socket units.
8843 * libudev gained a new call udev_device_set_attribute_value()
8844 to set sysfs attributes of a device.
8846 * The udev daemon now sets the default number of worker
8847 processes executed in parallel based on the number of available
8848 CPUs instead of the amount of available RAM. This is supposed
8849 to provide a more reliable default and limit a too aggressive
8850 parallelism for setups with 1000s of devices connected.
8852 Contributions from: Auke Kok, Colin Walters, Cristian
8853 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
8854 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
8855 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
8856 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
8857 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
8858 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
8859 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
8860 Zbigniew Jędrzejewski-Szmek
8864 * Configuration of unit files may now be extended via drop-in
8865 files without having to edit/override the unit files
8866 themselves. More specifically, if the administrator wants to
8867 change one value for a service file foobar.service he can
8868 now do so by dropping in a configuration snippet into
8869 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
8870 will load all these snippets and apply them on top of the
8871 main unit configuration file, possibly extending or
8872 overriding its settings. Using these drop-in snippets is
8873 generally nicer than the two earlier options for changing
8874 unit files locally: copying the files from
8875 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
8876 them there; or creating a new file in /etc/systemd/system/
8877 that incorporates the original one via ".include". Drop-in
8878 snippets into these .d/ directories can be placed in any
8879 directory systemd looks for units in, and the usual
8880 overriding semantics between /usr/lib, /etc and /run apply
8883 * Most unit file settings which take lists of items can now be
8884 reset by assigning the empty string to them. For example,
8885 normally, settings such as Environment=FOO=BAR append a new
8886 environment variable assignment to the environment block,
8887 each time they are used. By assigning Environment= the empty
8888 string the environment block can be reset to empty. This is
8889 particularly useful with the .d/*.conf drop-in snippets
8890 mentioned above, since this adds the ability to reset list
8891 settings from vendor unit files via these drop-ins.
8893 * systemctl gained a new "list-dependencies" command for
8894 listing the dependencies of a unit recursively.
8896 * Inhibitors are now honored and listed by "systemctl
8897 suspend", "systemctl poweroff" (and similar) too, not only
8898 GNOME. These commands will also list active sessions by
8901 * Resource limits (as exposed by the various control group
8902 controllers) can now be controlled dynamically at runtime
8903 for all units. More specifically, you can now use a command
8904 like "systemctl set-cgroup-attr foobar.service cpu.shares
8905 2000" to alter the CPU shares a specific service gets. These
8906 settings are stored persistently on disk, and thus allow the
8907 administrator to easily adjust the resource usage of
8908 services with a few simple commands. This dynamic resource
8909 management logic is also available to other programs via the
8910 bus. Almost any kernel cgroup attribute and controller is
8913 * systemd-vconsole-setup will now copy all font settings to
8914 all allocated VTs, where it previously applied them only to
8917 * libsystemd-login gained the new sd_session_get_tty() API
8920 * This release drops support for a few legacy or
8921 distribution-specific LSB facility names when parsing init
8922 scripts: $x-display-manager, $mail-transfer-agent,
8923 $mail-transport-agent, $mail-transfer-agent, $smtp,
8924 $null. Also, the mail-transfer-agent.target unit backing
8925 this has been removed. Distributions which want to retain
8926 compatibility with this should carry the burden for
8927 supporting this themselves and patch support for these back
8928 in, if they really need to. Also, the facilities $syslog and
8929 $local_fs are now ignored, since systemd does not support
8930 early-boot LSB init scripts anymore, and these facilities
8931 are implied anyway for normal services. syslog.target has
8934 * There are new bus calls on PID1's Manager object for
8935 cancelling jobs, and removing snapshot units. Previously,
8936 both calls were only available on the Job and Snapshot
8939 * systemd-journal-gatewayd gained SSL support.
8941 * The various "environment" files, such as /etc/locale.conf
8942 now support continuation lines with a backslash ("\") as
8943 last character in the line, similarly in style (but different)
8944 to how this is supported in shells.
8946 * For normal user processes the _SYSTEMD_USER_UNIT= field is
8947 now implicitly appended to every log entry logged. systemctl
8948 has been updated to filter by this field when operating on a
8949 user systemd instance.
8951 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
8952 CAP_AUDIT_CONTROL capabilities to the capabilities set for
8953 the container. This makes it easier to boot unmodified
8954 Fedora systems in a container, which however still requires
8955 audit=0 to be passed on the kernel command line. Auditing in
8956 kernel and userspace is unfortunately still too broken in
8957 context of containers, hence we recommend compiling it out
8958 of the kernel or using audit=0. Hopefully this will be fixed
8959 one day for good in the kernel.
8961 * nspawn gained the new --bind= and --bind-ro= parameters to
8962 bind mount specific directories from the host into the
8965 * nspawn will now mount its own devpts file system instance
8966 into the container, in order not to leak pty devices from
8967 the host into the container.
8969 * systemd will now read the firmware boot time performance
8970 information from the EFI variables, if the used boot loader
8971 supports this, and takes it into account for boot performance
8972 analysis via "systemd-analyze". This is currently supported
8973 only in conjunction with Gummiboot, but could be supported
8974 by other boot loaders too. For details see:
8976 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
8978 * A new generator has been added that automatically mounts the
8979 EFI System Partition (ESP) to /boot, if that directory
8980 exists, is empty, and no other file system has been
8981 configured to be mounted there.
8983 * logind will now send out PrepareForSleep(false) out
8984 unconditionally, after coming back from suspend. This may be
8985 used by applications as asynchronous notification for
8986 system resume events.
8988 * "systemctl unlock-sessions" has been added, that allows
8989 unlocking the screens of all user sessions at once, similar
8990 to how "systemctl lock-sessions" already locked all users
8991 sessions. This is backed by a new D-Bus call UnlockSessions().
8993 * "loginctl seat-status" will now show the master device of a
8994 seat. (i.e. the device of a seat that needs to be around for
8995 the seat to be considered available, usually the graphics
8998 * tmpfiles gained a new "X" line type, that allows
8999 configuration of files and directories (with wildcards) that
9000 shall be excluded from automatic cleanup ("aging").
9002 * udev default rules set the device node permissions now only
9003 at "add" events, and do not change them any longer with a
9004 later "change" event.
9006 * The log messages for lid events and power/sleep keypresses
9007 now carry a message ID.
9009 * We now have a substantially larger unit test suite, but this
9010 continues to be work in progress.
9012 * udevadm hwdb gained a new --root= parameter to change the
9013 root directory to operate relative to.
9015 * logind will now issue a background sync() request to the kernel
9016 early at shutdown, so that dirty buffers are flushed to disk early
9017 instead of at the last moment, in order to optimize shutdown
9020 * A new bootctl tool has been added that is an interface for
9021 certain boot loader operations. This is currently a preview
9022 and is likely to be extended into a small mechanism daemon
9023 like timedated, localed, hostnamed, and can be used by
9024 graphical UIs to enumerate available boot options, and
9025 request boot into firmware operations.
9027 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
9028 the rest of the package. It also has been updated to work
9029 correctly in initrds.
9031 * polkit previously has been runtime optional, and is now also
9032 compile time optional via a configure switch.
9034 * systemd-analyze has been reimplemented in C. Also "systemctl
9035 dot" has moved into systemd-analyze.
9037 * "systemctl status" with no further parameters will now print
9038 the status of all active or failed units.
9040 * Operations such as "systemctl start" can now be executed
9041 with a new mode "--irreversible" which may be used to queue
9042 operations that cannot accidentally be reversed by a later
9043 job queuing. This is by default used to make shutdown
9044 requests more robust.
9046 * The Python API of systemd now gained a new module for
9047 reading journal files.
9049 * A new tool kernel-install has been added that can install
9050 kernel images according to the Boot Loader Specification:
9052 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
9054 * Boot time console output has been improved to provide
9055 animated boot time output for hanging jobs.
9057 * A new tool systemd-activate has been added which can be used
9058 to test socket activation with, directly from the command
9059 line. This should make it much easier to test and debug
9060 socket activation in daemons.
9062 * journalctl gained a new "--reverse" (or -r) option to show
9063 journal output in reverse order (i.e. newest line first).
9065 * journalctl gained a new "--pager-end" (or -e) option to jump
9066 to immediately jump to the end of the journal in the
9067 pager. This is only supported in conjunction with "less".
9069 * journalctl gained a new "--user-unit=" option, that works
9070 similarly to "--unit=" but filters for user units rather than
9073 * A number of unit files to ease adoption of systemd in
9074 initrds has been added. This moves some minimal logic from
9075 the various initrd implementations into systemd proper.
9077 * The journal files are now owned by a new group
9078 "systemd-journal", which exists specifically to allow access
9079 to the journal, and nothing else. Previously, we used the
9080 "adm" group for that, which however possibly covers more
9081 than just journal/log file access. This new group is now
9082 already used by systemd-journal-gatewayd to ensure this
9083 daemon gets access to the journal files and as little else
9084 as possible. Note that "make install" will also set FS ACLs
9085 up for /var/log/journal to give "adm" and "wheel" read
9086 access to it, in addition to "systemd-journal" which owns
9087 the journal files. We recommend that packaging scripts also
9088 add read access to "adm" + "wheel" to /var/log/journal, and
9089 all existing/future journal files. To normal users and
9090 administrators little changes, however packagers need to
9091 ensure to create the "systemd-journal" system group at
9092 package installation time.
9094 * The systemd-journal-gatewayd now runs as unprivileged user
9095 systemd-journal-gateway:systemd-journal-gateway. Packaging
9096 scripts need to create these system user/group at
9099 * timedated now exposes a new boolean property CanNTP that
9100 indicates whether a local NTP service is available or not.
9102 * systemd-detect-virt will now also detect xen PVs
9104 * The pstore file system is now mounted by default, if it is
9107 * In addition to the SELinux and IMA policies we will now also
9108 load SMACK policies at early boot.
9110 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
9111 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
9112 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
9113 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
9114 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
9115 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
9116 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
9117 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
9118 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
9119 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
9120 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
9121 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
9122 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
9123 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
9127 * Timer units now support calendar time events in addition to
9128 monotonic time events. That means you can now trigger a unit
9129 based on a calendar time specification such as "Thu,Fri
9130 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
9131 or fifth day of any month of the year 2013, given that it is
9132 a thursday or friday. This brings timer event support
9133 considerably closer to cron's capabilities. For details on
9134 the supported calendar time specification language see
9137 * udev now supports a number of different naming policies for
9138 network interfaces for predictable names, and a combination
9139 of these policies is now the default. Please see this wiki
9140 document for details:
9142 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
9144 * Auke Kok's bootchart implementation has been added to the
9145 systemd tree. It is an optional component that can graph the
9146 boot in quite some detail. It is one of the best bootchart
9147 implementations around and minimal in its code and
9150 * nss-myhostname has been integrated into the systemd source
9151 tree. nss-myhostname guarantees that the local hostname
9152 always stays resolvable via NSS. It has been a weak
9153 requirement of systemd-hostnamed since a long time, and
9154 since its code is actually trivial we decided to just
9155 include it in systemd's source tree. It can be turned off
9156 with a configure switch.
9158 * The read-ahead logic is now capable of properly detecting
9159 whether a btrfs file system is on SSD or rotating media, in
9160 order to optimize the read-ahead scheme. Previously, it was
9161 only capable of detecting this on traditional file systems
9164 * In udev, additional device properties are now read from the
9165 IAB in addition to the OUI database. Also, Bluetooth company
9166 identities are attached to the devices as well.
9168 * In service files %U may be used as specifier that is
9169 replaced by the configured user name of the service.
9171 * nspawn may now be invoked without a controlling TTY. This
9172 makes it suitable for invocation as its own service. This
9173 may be used to set up a simple containerized server system
9174 using only core OS tools.
9176 * systemd and nspawn can now accept socket file descriptors
9177 when they are started for socket activation. This enables
9178 implementation of socket activated nspawn
9179 containers. i.e. think about autospawning an entire OS image
9180 when the first SSH or HTTP connection is received. We expect
9181 that similar functionality will also be added to libvirt-lxc
9184 * journalctl will now suppress ANSI color codes when
9185 presenting log data.
9187 * systemctl will no longer show control group information for
9188 a unit if the control group is empty anyway.
9190 * logind can now automatically suspend/hibernate/shutdown the
9193 * /etc/machine-info and hostnamed now also expose the chassis
9194 type of the system. This can be used to determine whether
9195 the local system is a laptop, desktop, handset or
9196 tablet. This information may either be configured by the
9197 user/vendor or is automatically determined from ACPI and DMI
9198 information if possible.
9200 * A number of polkit actions are now bound together with "imply"
9201 rules. This should simplify creating UIs because many actions
9202 will now authenticate similar ones as well.
9204 * Unit files learnt a new condition ConditionACPower= which
9205 may be used to conditionalize a unit depending on whether an
9206 AC power source is connected or not, of whether the system
9207 is running on battery power.
9209 * systemctl gained a new "is-failed" verb that may be used in
9210 shell scripts and suchlike to check whether a specific unit
9211 is in the "failed" state.
9213 * The EnvironmentFile= setting in unit files now supports file
9214 globbing, and can hence be used to easily read a number of
9215 environment files at once.
9217 * systemd will no longer detect and recognize specific
9218 distributions. All distribution-specific #ifdeffery has been
9219 removed, systemd is now fully generic and
9220 distribution-agnostic. Effectively, not too much is lost as
9221 a lot of the code is still accessible via explicit configure
9222 switches. However, support for some distribution specific
9223 legacy configuration file formats has been dropped. We
9224 recommend distributions to simply adopt the configuration
9225 files everybody else uses now and convert the old
9226 configuration from packaging scripts. Most distributions
9227 already did that. If that's not possible or desirable,
9228 distributions are welcome to forward port the specific
9229 pieces of code locally from the git history.
9231 * When logging a message about a unit systemd will now always
9232 log the unit name in the message meta data.
9234 * localectl will now also discover system locale data that is
9235 not stored in locale archives, but directly unpacked.
9237 * logind will no longer unconditionally use framebuffer
9238 devices as seat masters, i.e. as devices that are required
9239 to be existing before a seat is considered preset. Instead,
9240 it will now look for all devices that are tagged as
9241 "seat-master" in udev. By default, framebuffer devices will
9242 be marked as such, but depending on local systems, other
9243 devices might be marked as well. This may be used to
9244 integrate graphics cards using closed source drivers (such
9245 as NVidia ones) more nicely into logind. Note however, that
9246 we recommend using the open source NVidia drivers instead,
9247 and no udev rules for the closed-source drivers will be
9248 shipped from us upstream.
9250 Contributions from: Adam Williamson, Alessandro Crismani, Auke
9251 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
9252 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
9253 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
9254 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
9255 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
9256 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
9257 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
9258 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
9259 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
9260 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
9261 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
9266 * udev gained support for loading additional device properties
9267 from an indexed database that is keyed by vendor/product IDs
9268 and similar device identifiers. For the beginning this
9269 "hwdb" is populated with data from the well-known PCI and
9270 USB database, but also includes PNP, ACPI and OID data. In
9271 the longer run this indexed database shall grow into
9272 becoming the one central database for non-essential
9273 userspace device metadata. Previously, data from the PCI/USB
9274 database was only attached to select devices, since the
9275 lookup was a relatively expensive operation due to O(n) time
9276 complexity (with n being the number of entries in the
9277 database). Since this is now O(1), we decided to add in this
9278 data for all devices where this is available, by
9279 default. Note that the indexed database needs to be rebuilt
9280 when new data files are installed. To achieve this you need
9281 to update your packaging scripts to invoke "udevadm hwdb
9282 --update" after installation of hwdb data files. For
9283 RPM-based distributions we introduced the new
9284 %udev_hwdb_update macro for this purpose.
9286 * The Journal gained support for the "Message Catalog", an
9287 indexed database to link up additional information with
9288 journal entries. For further details please check:
9290 https://www.freedesktop.org/wiki/Software/systemd/catalog
9292 The indexed message catalog database also needs to be
9293 rebuilt after installation of message catalog files. Use
9294 "journalctl --update-catalog" for this. For RPM-based
9295 distributions we introduced the %journal_catalog_update
9296 macro for this purpose.
9298 * The Python Journal bindings gained support for the standard
9299 Python logging framework.
9301 * The Journal API gained new functions for checking whether
9302 the underlying file system of a journal file is capable of
9303 properly reporting file change notifications, or whether
9304 applications that want to reflect journal changes "live"
9305 need to recheck journal files continuously in appropriate
9308 * It is now possible to set the "age" field for tmpfiles
9309 entries to 0, indicating that files matching this entry
9310 shall always be removed when the directories are cleaned up.
9312 * coredumpctl gained a new "gdb" verb which invokes gdb
9313 right-away on the selected coredump.
9315 * There's now support for "hybrid sleep" on kernels that
9316 support this, in addition to "suspend" and "hibernate". Use
9317 "systemctl hybrid-sleep" to make use of this.
9319 * logind's HandleSuspendKey= setting (and related settings)
9320 now gained support for a new "lock" setting to simply
9321 request the screen lock on all local sessions, instead of
9322 actually executing a suspend or hibernation.
9324 * systemd will now mount the EFI variables file system by
9327 * Socket units now gained support for configuration of the
9328 SMACK security label.
9330 * timedatectl will now output the time of the last and next
9331 daylight saving change.
9333 * We dropped support for various legacy and distro-specific
9334 concepts, such as insserv, early-boot SysV services
9335 (i.e. those for non-standard runlevels such as 'b' or 'S')
9336 or ArchLinux /etc/rc.conf support. We recommend the
9337 distributions who still need support this to either continue
9338 to maintain the necessary patches downstream, or find a
9339 different solution. (Talk to us if you have questions!)
9341 * Various systemd components will now bypass polkit checks for
9342 root and otherwise handle properly if polkit is not found to
9343 be around. This should fix most issues for polkit-less
9344 systems. Quite frankly this should have been this way since
9345 day one. It is absolutely our intention to make systemd work
9346 fine on polkit-less systems, and we consider it a bug if
9347 something does not work as it should if polkit is not around.
9349 * For embedded systems it is now possible to build udev and
9350 systemd without blkid and/or kmod support.
9352 * "systemctl switch-root" is now capable of switching root
9353 more than once. I.e. in addition to transitions from the
9354 initrd to the host OS it is now possible to transition to
9355 further OS images from the host. This is useful to implement
9356 offline updating tools.
9358 * Various other additions have been made to the RPM macros
9359 shipped with systemd. Use %udev_rules_update() after
9360 installing new udev rules files. %_udevhwdbdir,
9361 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
9362 %_sysctldir are now available which resolve to the right
9363 directories for packages to place various data files in.
9365 * journalctl gained the new --full switch (in addition to
9366 --all, to disable ellipsation for long messages.
9368 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
9369 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
9370 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
9371 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
9372 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
9373 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
9374 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
9375 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
9376 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
9380 * journalctl gained new --since= and --until= switches to
9381 filter by time. It also now supports nice filtering for
9382 units via --unit=/-u.
9384 * Type=oneshot services may use ExecReload= and do the
9387 * The journal daemon now supports time-based rotation and
9388 vacuuming, in addition to the usual disk-space based
9391 * The journal will now index the available field values for
9392 each field name. This enables clients to show pretty drop
9393 downs of available match values when filtering. The bash
9394 completion of journalctl has been updated
9395 accordingly. journalctl gained a new switch -F to list all
9396 values a certain field takes in the journal database.
9398 * More service events are now written as structured messages
9399 to the journal, and made recognizable via message IDs.
9401 * The timedated, localed and hostnamed mini-services which
9402 previously only provided support for changing time, locale
9403 and hostname settings from graphical DEs such as GNOME now
9404 also have a minimal (but very useful) text-based client
9405 utility each. This is probably the nicest way to changing
9406 these settings from the command line now, especially since
9407 it lists available options and is fully integrated with bash
9410 * There's now a new tool "systemd-coredumpctl" to list and
9411 extract coredumps from the journal.
9413 * We now install a README each in /var/log/ and
9414 /etc/rc.d/init.d explaining where the system logs and init
9415 scripts went. This hopefully should help folks who go to
9416 that dirs and look into the otherwise now empty void and
9417 scratch their heads.
9419 * When user-services are invoked (by systemd --user) the
9420 $MANAGERPID env var is set to the PID of systemd.
9422 * SIGRTMIN+24 when sent to a --user instance will now result
9423 in immediate termination of systemd.
9425 * gatewayd received numerous feature additions such as a
9426 "follow" mode, for live syncing and filtering.
9428 * browse.html now allows filtering and showing detailed
9429 information on specific entries. Keyboard navigation and
9430 mouse screen support has been added.
9432 * gatewayd/journalctl now supports HTML5/JSON
9433 Server-Sent-Events as output.
9435 * The SysV init script compatibility logic will now
9436 heuristically determine whether a script supports the
9437 "reload" verb, and only then make this available as
9440 * "systemctl status --follow" has been removed, use "journalctl
9443 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
9444 have been removed since they are hardly useful to be
9447 * And I'd like to take the opportunity to specifically mention
9448 Zbigniew for his great contributions. Zbigniew, you rock!
9450 Contributions from: Andrew Eikum, Christian Hesse, Colin
9451 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
9452 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
9453 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
9454 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
9455 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
9456 Jędrzejewski-Szmek, Сковорода Никита Андреевич
9460 * If /etc/vconsole.conf is non-existent or empty we will no
9461 longer load any console font or key map at boot by
9462 default. Instead the kernel defaults will be left
9463 intact. This is definitely the right thing to do, as no
9464 configuration should mean no configuration, and hard-coding
9465 font names that are different on all archs is probably a bad
9466 idea. Also, the kernel default key map and font should be
9467 good enough for most cases anyway, and mostly identical to
9468 the userspace fonts/key maps we previously overloaded them
9469 with. If distributions want to continue to default to a
9470 non-kernel font or key map they should ship a default
9471 /etc/vconsole.conf with the appropriate contents.
9473 Contributions from: Colin Walters, Daniel J Walsh, Dave
9474 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
9475 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
9479 * journalctl gained a new --cursor= switch to show entries
9480 starting from the specified location in the journal.
9482 * We now enforce a size limit on journal entry fields exported
9483 with "-o json" in journalctl. Fields larger than 4K will be
9484 assigned null. This can be turned off with --all.
9486 * An (optional) journal gateway daemon is now available as
9487 "systemd-journal-gatewayd.service". This service provides
9488 access to the journal via HTTP and JSON. This functionality
9489 will be used to implement live log synchronization in both
9490 pull and push modes, but has various other users too, such
9491 as easy log access for debugging of embedded devices. Right
9492 now it is already useful to retrieve the journal via HTTP:
9494 # systemctl start systemd-journal-gatewayd.service
9495 # wget http://localhost:19531/entries
9497 This will download the journal contents in a
9498 /var/log/messages compatible format. The same as JSON:
9500 # curl -H"Accept: application/json" http://localhost:19531/entries
9502 This service is also accessible via a web browser where a
9503 single static HTML5 app is served that uses the JSON logic
9504 to enable the user to do some basic browsing of the
9505 journal. This will be extended later on. Here's an example
9506 screenshot of this app in its current state:
9508 http://0pointer.de/public/journal-gatewayd
9510 Contributions from: Kay Sievers, Lennart Poettering, Robert
9511 Milasan, Tom Gundersen
9515 * The bash completion logic is now available for journalctl
9518 * We do not mount the "cpuset" controller anymore together with
9519 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
9520 started if no parameters are assigned to it. "cpuset" hence
9521 broke code that assumed it could create "cpu" groups and
9524 * journalctl -f will now subscribe to terminal size changes,
9525 and line break accordingly.
9527 Contributions from: Dave Reisner, Kay Sievers, Lennart
9528 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
9532 * nspawn will now create a symlink /etc/localtime in the
9533 container environment, copying the host's timezone
9534 setting. Previously this has been done via a bind mount, but
9535 since symlinks cannot be bind mounted this has now been
9536 changed to create/update the appropriate symlink.
9538 * journalctl -n's line number argument is now optional, and
9539 will default to 10 if omitted.
9541 * journald will now log the maximum size the journal files may
9542 take up on disk. This is particularly useful if the default
9543 built-in logic of determining this parameter from the file
9544 system size is used. Use "systemctl status
9545 systemd-journald.service" to see this information.
9547 * The multi-seat X wrapper tool has been stripped down. As X
9548 is now capable of enumerating graphics devices via udev in a
9549 seat-aware way the wrapper is not strictly necessary
9550 anymore. A stripped down temporary stop-gap is still shipped
9551 until the upstream display managers have been updated to
9552 fully support the new X logic. Expect this wrapper to be
9553 removed entirely in one of the next releases.
9555 * HandleSleepKey= in logind.conf has been split up into
9556 HandleSuspendKey= and HandleHibernateKey=. The old setting
9557 is not available anymore. X11 and the kernel are
9558 distinguishing between these keys and we should too. This
9559 also means the inhibition lock for these keys has been split
9562 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
9563 Poettering, Lukas Nykryn, Václav Pavlín
9567 * Whenever a unit changes state we will now log this to the
9568 journal and show along the unit's own log output in
9571 * ConditionPathIsMountPoint= can now properly detect bind
9572 mount points too. (Previously, a bind mount of one file
9573 system to another place in the same file system could not be
9574 detected as mount, since they shared struct stat's st_dev
9577 * We will now mount the cgroup controllers cpu, cpuacct,
9578 cpuset and the controllers net_cls, net_prio together by
9581 * nspawn containers will now have a virtualized boot
9582 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
9583 over with a randomized ID at container initialization). This
9584 has the effect of making "journalctl -b" do the right thing
9587 * The JSON output journal serialization has been updated not
9588 to generate "endless" list objects anymore, but rather one
9589 JSON object per line. This is more in line how most JSON
9590 parsers expect JSON objects. The new output mode
9591 "json-pretty" has been added to provide similar output, but
9592 neatly aligned for readability by humans.
9594 * We dropped all explicit sync() invocations in the shutdown
9595 code. The kernel does this implicitly anyway in the kernel
9596 reboot() syscall. halt(8)'s -n option is now a compatibility
9599 * We now support virtualized reboot() in containers, as
9600 supported by newer kernels. We will fall back to exit() if
9601 CAP_SYS_REBOOT is not available to the container. Also,
9602 nspawn makes use of this now and will actually reboot the
9603 container if the containerized OS asks for that.
9605 * journalctl will only show local log output by default
9606 now. Use --merge (-m) to show remote log output, too.
9608 * libsystemd-journal gained the new sd_journal_get_usage()
9609 call to determine the current disk usage of all journal
9610 files. This is exposed in the new "journalctl --disk-usage"
9613 * journald gained a new configuration setting SplitMode= in
9614 journald.conf which may be used to control how user journals
9615 are split off. See journald.conf(5) for details.
9617 * A new condition type ConditionFileNotEmpty= has been added.
9619 * tmpfiles' "w" lines now support file globbing, to write
9620 multiple files at once.
9622 * We added Python bindings for the journal submission
9623 APIs. More Python APIs for a number of selected APIs will
9624 likely follow. Note that we intend to add native bindings
9625 only for the Python language, as we consider it common
9626 enough to deserve bindings shipped within systemd. There are
9627 various projects outside of systemd that provide bindings
9628 for languages such as PHP or Lua.
9630 * Many conditions will now resolve specifiers such as %i. In
9631 addition, PathChanged= and related directives of .path units
9632 now support specifiers as well.
9634 * There's now a new RPM macro definition for the system preset
9637 * journald will now warn if it ca not forward a message to the
9638 syslog daemon because its socket is full.
9640 * timedated will no longer write or process /etc/timezone,
9641 except on Debian. As we do not support late mounted /usr
9642 anymore /etc/localtime always being a symlink is now safe,
9643 and hence the information in /etc/timezone is not necessary
9646 * logind will now always reserve one VT for a text getty (VT6
9647 by default). Previously if more than 6 X sessions where
9648 started they took up all the VTs with auto-spawned gettys,
9649 so that no text gettys were available anymore.
9651 * udev will now automatically inform the btrfs kernel logic
9652 about btrfs RAID components showing up. This should make
9653 simple hotplug based btrfs RAID assembly work.
9655 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
9656 (but not for its children which will stay at the kernel
9657 default). This should allow setups with a lot more listening
9660 * systemd will now always pass the configured timezone to the
9661 kernel at boot. timedated will do the same when the timezone
9664 * logind's inhibition logic has been updated. By default,
9665 logind will now handle the lid switch, the power and sleep
9666 keys all the time, even in graphical sessions. If DEs want
9667 to handle these events on their own they should take the new
9668 handle-power-key, handle-sleep-key and handle-lid-switch
9669 inhibitors during their runtime. A simple way to achieve
9670 that is to invoke the DE wrapped in an invocation of:
9672 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
9674 * Access to unit operations is now checked via SELinux taking
9675 the unit file label and client process label into account.
9677 * systemd will now notify the administrator in the journal
9678 when he over-mounts a non-empty directory.
9680 * There are new specifiers that are resolved in unit files,
9681 for the host name (%H), the machine ID (%m) and the boot ID
9684 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
9685 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
9686 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
9687 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
9688 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
9689 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
9690 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
9694 * Support for reading structured kernel messages from
9695 /dev/kmsg has now been added and is enabled by default.
9697 * Support for reading kernel messages from /proc/kmsg has now
9698 been removed. If you want kernel messages in the journal
9699 make sure to run a recent kernel (>= 3.5) that supports
9700 reading structured messages from /dev/kmsg (see
9701 above). /proc/kmsg is now exclusive property of classic
9702 syslog daemons again.
9704 * The libudev API gained the new
9705 udev_device_new_from_device_id() call.
9707 * The logic for file system namespace (ReadOnlyDirectory=,
9708 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
9709 require pivot_root() anymore. This means fewer temporary
9710 directories are created below /tmp for this feature.
9712 * nspawn containers will now see and receive all submounts
9713 made on the host OS below the root file system of the
9716 * Forward Secure Sealing is now supported for Journal files,
9717 which provide cryptographical sealing of journal files so
9718 that attackers cannot alter log history anymore without this
9719 being detectable. Lennart will soon post a blog story about
9720 this explaining it in more detail.
9722 * There are two new service settings RestartPreventExitStatus=
9723 and SuccessExitStatus= which allow configuration of exit
9724 status (exit code or signal) which will be excepted from the
9725 restart logic, resp. consider successful.
9727 * journalctl gained the new --verify switch that can be used
9728 to check the integrity of the structure of journal files and
9729 (if Forward Secure Sealing is enabled) the contents of
9732 * nspawn containers will now be run with /dev/stdin, /dev/fd/
9733 and similar symlinks pre-created. This makes running shells
9734 as container init process a lot more fun.
9736 * The fstab support can now handle PARTUUID= and PARTLABEL=
9739 * A new ConditionHost= condition has been added to match
9740 against the hostname (with globs) and machine ID. This is
9741 useful for clusters where a single OS image is used to
9742 provision a large number of hosts which shall run slightly
9743 different sets of services.
9745 * Services which hit the restart limit will now be placed in a
9748 Contributions from: Bertram Poettering, Dave Reisner, Huang
9749 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
9750 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
9754 * When running in --user mode systemd will now become a
9755 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
9756 tree a lot more organized.
9758 * A new PartOf= unit dependency type has been introduced that
9759 may be used to group services in a natural way.
9761 * "systemctl enable" may now be used to enable instances of
9764 * journalctl now prints error log levels in red, and
9765 warning/notice log levels in bright white. It also supports
9766 filtering by log level now.
9768 * cgtop gained a new -n switch (similar to top), to configure
9769 the maximum number of iterations to run for. It also gained
9770 -b, to run in batch mode (accepting no input).
9772 * The suffix ".service" may now be omitted on most systemctl
9773 command lines involving service unit names.
9775 * There's a new bus call in logind to lock all sessions, as
9776 well as a loginctl verb for it "lock-sessions".
9778 * libsystemd-logind.so gained a new call sd_journal_perror()
9779 that works similar to libc perror() but logs to the journal
9780 and encodes structured information about the error number.
9782 * /etc/crypttab entries now understand the new keyfile-size=
9785 * shutdown(8) now can send a (configurable) wall message when
9786 a shutdown is cancelled.
9788 * The mount propagation mode for the root file system will now
9789 default to "shared", which is useful to make containers work
9790 nicely out-of-the-box so that they receive new mounts from
9791 the host. This can be undone locally by running "mount
9792 --make-rprivate /" if needed.
9794 * The prefdm.service file has been removed. Distributions
9795 should maintain this unit downstream if they intend to keep
9796 it around. However, we recommend writing normal unit files
9797 for display managers instead.
9799 * Since systemd is a crucial part of the OS we will now
9800 default to a number of compiler switches that improve
9801 security (hardening) such as read-only relocations, stack
9802 protection, and suchlike.
9804 * The TimeoutSec= setting for services is now split into
9805 TimeoutStartSec= and TimeoutStopSec= to allow configuration
9806 of individual time outs for the start and the stop phase of
9809 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
9810 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
9811 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
9812 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
9813 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
9814 Gundersen, Zbigniew Jędrzejewski-Szmek
9818 * The journal and id128 C APIs are now fully documented as man
9821 * Extra safety checks have been added when transitioning from
9822 the initial RAM disk to the main system to avoid accidental
9825 * /etc/crypttab entries now understand the new keyfile-offset=
9828 * systemctl -t can now be used to filter by unit load state.
9830 * The journal C API gained the new sd_journal_wait() call to
9831 make writing synchronous journal clients easier.
9833 * journalctl gained the new -D switch to show journals from a
9836 * journalctl now displays a special marker between log
9837 messages of two different boots.
9839 * The journal is now explicitly flushed to /var via a service
9840 systemd-journal-flush.service, rather than implicitly simply
9841 by seeing /var/log/journal to be writable.
9843 * journalctl (and the journal C APIs) can now match for much
9844 more complex expressions, with alternatives and
9847 * When transitioning from the initial RAM disk to the main
9848 system we will now kill all processes in a killing spree to
9849 ensure no processes stay around by accident.
9851 * Three new specifiers may be used in unit files: %u, %h, %s
9852 resolve to the user name, user home directory resp. user
9853 shell. This is useful for running systemd user instances.
9855 * We now automatically rotate journal files if their data
9856 object hash table gets a fill level > 75%. We also size the
9857 hash table based on the configured maximum file size. This
9858 together should lower hash collisions drastically and thus
9859 speed things up a bit.
9861 * journalctl gained the new "--header" switch to introspect
9862 header data of journal files.
9864 * A new setting SystemCallFilters= has been added to services
9865 which may be used to apply blacklists or whitelists to
9866 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
9868 * nspawn gained a new --link-journal= switch (and quicker: -j)
9869 to link the container journal with the host. This makes it
9870 very easy to centralize log viewing on the host for all
9871 guests while still keeping the journal files separated.
9873 * Many bugfixes and optimizations
9875 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
9876 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
9877 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
9882 * Several tools now understand kernel command line arguments,
9883 which are only read when run in an initial RAM disk. They
9884 usually follow closely their normal counterparts, but are
9887 * There's a new tool to analyze the readahead files that are
9888 automatically generated at boot. Use:
9890 /usr/lib/systemd/systemd-readahead analyze /.readahead
9892 * We now provide an early debug shell on tty9 if this enabled. Use:
9894 systemctl enable debug-shell.service
9896 * All plymouth related units have been moved into the Plymouth
9897 package. Please make sure to upgrade your Plymouth version
9900 * systemd-tmpfiles now supports getting passed the basename of
9901 a configuration file only, in which case it will look for it
9902 in all appropriate directories automatically.
9904 * udevadm info now takes a /dev or /sys path as argument, and
9905 does the right thing. Example:
9907 udevadm info /dev/sda
9908 udevadm info /sys/class/block/sda
9910 * systemctl now prints a warning if a unit is stopped but a
9911 unit that might trigger it continues to run. Example: a
9912 service is stopped but the socket that activates it is left
9915 * "systemctl status" will now mention if the log output was
9916 shortened due to rotation since a service has been started.
9918 * The journal API now exposes functions to determine the
9919 "cutoff" times due to rotation.
9921 * journald now understands SIGUSR1 and SIGUSR2 for triggering
9922 immediately flushing of runtime logs to /var if possible,
9923 resp. for triggering immediate rotation of the journal
9926 * It is now considered an error if a service is attempted to
9927 be stopped that is not loaded.
9929 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
9931 * systemd-analyze now supports Python 3
9933 * tmpfiles now supports cleaning up directories via aging
9934 where the first level dirs are always kept around but
9935 directories beneath it automatically aged. This is enabled
9936 by prefixing the age field with '~'.
9938 * Seat objects now expose CanGraphical, CanTTY properties
9939 which is required to deal with very fast bootups where the
9940 display manager might be running before the graphics drivers
9941 completed initialization.
9943 * Seat objects now expose a State property.
9945 * We now include RPM macros for service enabling/disabling
9946 based on the preset logic. We recommend RPM based
9947 distributions to make use of these macros if possible. This
9948 makes it simpler to reuse RPM spec files across
9951 * We now make sure that the collected systemd unit name is
9952 always valid when services log to the journal via
9955 * There's a new man page kernel-command-line(7) detailing all
9956 command line options we understand.
9958 * The fstab generator may now be disabled at boot by passing
9959 fstab=0 on the kernel command line.
9961 * A new kernel command line option modules-load= is now understood
9962 to load a specific kernel module statically, early at boot.
9964 * Unit names specified on the systemctl command line are now
9965 automatically escaped as needed. Also, if file system or
9966 device paths are specified they are automatically turned
9967 into the appropriate mount or device unit names. Example:
9969 systemctl status /home
9970 systemctl status /dev/sda
9972 * The SysVConsole= configuration option has been removed from
9973 system.conf parsing.
9975 * The SysV search path is no longer exported on the D-Bus
9978 * The Names= option has been removed from unit file parsing.
9980 * There's a new man page bootup(7) detailing the boot process.
9982 * Every unit and every generator we ship with systemd now
9983 comes with full documentation. The self-explanatory boot is
9986 * A couple of services gained "systemd-" prefixes in their
9987 name if they wrap systemd code, rather than only external
9988 code. Among them fsck@.service which is now
9989 systemd-fsck@.service.
9991 * The HaveWatchdog property has been removed from the D-Bus
9994 * systemd.confirm_spawn= on the kernel command line should now
9997 * There's a new man page crypttab(5) which details all options
9998 we actually understand.
10000 * systemd-nspawn gained a new --capability= switch to pass
10001 additional capabilities to the container.
10003 * timedated will now read known NTP implementation unit names
10004 from /usr/lib/systemd/ntp-units.d/*.list,
10005 systemd-timedated-ntp.target has been removed.
10007 * journalctl gained a new switch "-b" that lists log data of
10008 the current boot only.
10010 * The notify socket is in the abstract namespace again, in
10011 order to support daemons which chroot() at start-up.
10013 * There is a new Storage= configuration option for journald
10014 which allows configuration of where log data should go. This
10015 also provides a way to disable journal logging entirely, so
10016 that data collected is only forwarded to the console, the
10017 kernel log buffer or another syslog implementation.
10019 * Many bugfixes and optimizations
10021 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
10022 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
10023 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
10024 Shawn Landden, Tom Gundersen
10028 * "systemctl help <unit>" now shows the man page if one is
10031 * Several new man pages have been added.
10033 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
10034 MaxLevelConsole= can now be specified in
10035 journald.conf. These options allow reducing the amount of
10036 data stored on disk or forwarded by the log level.
10038 * TimerSlackNSec= can now be specified in system.conf for
10039 PID1. This allows system-wide power savings.
10041 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
10042 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
10047 * logind is now capable of (optionally) handling power and
10048 sleep keys as well as the lid switch.
10050 * journalctl now understands the syntax "journalctl
10051 /usr/bin/avahi-daemon" to get all log output of a specific
10054 * CapabilityBoundingSet= in system.conf now also influences
10055 the capability bound set of usermode helpers of the kernel.
10057 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
10058 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
10059 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
10060 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
10064 * Note that we skipped 139 releases here in order to set the
10065 new version to something that is greater than both udev's
10066 and systemd's most recent version number.
10068 * udev: all udev sources are merged into the systemd source tree now.
10069 All future udev development will happen in the systemd tree. It
10070 is still fully supported to use the udev daemon and tools without
10071 systemd running, like in initramfs or other init systems. Building
10072 udev though, will require the *build* of the systemd tree, but
10073 udev can be properly *run* without systemd.
10075 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
10076 should be used to create dead device nodes as workarounds for broken
10079 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
10080 no longer supported. udev_monitor_new_from_netlink() needs to be
10081 used to subscribe to events.
10083 * udev: when udevd is started by systemd, processes which are left
10084 behind by forking them off of udev rules, are unconditionally cleaned
10085 up and killed now after the event handling has finished. Services or
10086 daemons must be started as systemd services. Services can be
10087 pulled-in by udev to get started, but they can no longer be directly
10088 forked by udev rules.
10090 * udev: the daemon binary is called systemd-udevd now and installed
10091 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
10092 to adapt to that, create symlink, or rename the binary after building
10095 * libudev no longer provides these symbols:
10096 udev_monitor_from_socket()
10097 udev_queue_get_failed_list_entry()
10098 udev_get_{dev,sys,run}_path()
10099 The versions number was bumped and symbol versioning introduced.
10101 * systemd-loginctl and systemd-journalctl have been renamed
10102 to loginctl and journalctl to match systemctl.
10104 * The config files: /etc/systemd/systemd-logind.conf and
10105 /etc/systemd/systemd-journald.conf have been renamed to
10106 logind.conf and journald.conf. Package updates should rename
10107 the files to the new names on upgrade.
10109 * For almost all files the license is now LGPL2.1+, changed
10110 from the previous GPL2.0+. Exceptions are some minor stuff
10111 of udev (which will be changed to LGPL2.1 eventually, too),
10112 and the MIT licensed sd-daemon.[ch] library that is suitable
10113 to be used as drop-in files.
10115 * systemd and logind now handle system sleep states, in
10116 particular suspending and hibernating.
10118 * logind now implements a sleep/shutdown/idle inhibiting logic
10119 suitable for a variety of uses. Soonishly Lennart will blog
10120 about this in more detail.
10122 * var-run.mount and var-lock.mount are no longer provided
10123 (which previously bind mounted these directories to their new
10124 places). Distributions which have not converted these
10125 directories to symlinks should consider stealing these files
10126 from git history and add them downstream.
10128 * We introduced the Documentation= field for units and added
10129 this to all our shipped units. This is useful to make it
10130 easier to explore the boot and the purpose of the various
10133 * All smaller setup units (such as
10134 systemd-vconsole-setup.service) now detect properly if they
10135 are run in a container and are skipped when
10136 appropriate. This guarantees an entirely noise-free boot in
10137 Linux container environments such as systemd-nspawn.
10139 * A framework for implementing offline system updates is now
10140 integrated, for details see:
10141 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
10143 * A new service type Type=idle is available now which helps us
10144 avoiding ugly interleaving of getty output and boot status
10147 * There's now a system-wide CapabilityBoundingSet= option to
10148 globally reduce the set of capabilities for the
10149 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
10150 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
10151 even CAP_NET_ADMIN system-wide for secure systems.
10153 * There are now system-wide DefaultLimitXXX= options to
10154 globally change the defaults of the various resource limits
10155 for all units started by PID 1.
10157 * Harald Hoyer's systemd test suite has been integrated into
10158 systemd which allows easy testing of systemd builds in qemu
10159 and nspawn. (This is really awesome! Ask us for details!)
10161 * The fstab parser is now implemented as generator, not inside
10164 * systemctl will now warn you if .mount units generated from
10165 /etc/fstab are out of date due to changes in fstab that
10166 have not been read by systemd yet.
10168 * systemd is now suitable for usage in initrds. Dracut has
10169 already been updated to make use of this. With this in place
10170 initrds get a slight bit faster but primarily are much
10171 easier to introspect and debug since "systemctl status" in
10172 the host system can be used to introspect initrd services,
10173 and the journal from the initrd is kept around too.
10175 * systemd-delta has been added, a tool to explore differences
10176 between user/admin configuration and vendor defaults.
10178 * PrivateTmp= now affects both /tmp and /var/tmp.
10180 * Boot time status messages are now much prettier and feature
10181 proper english language. Booting up systemd has never been
10184 * Read-ahead pack files now include the inode number of all
10185 files to pre-cache. When the inode changes the pre-caching
10186 is not attempted. This should be nicer to deal with updated
10187 packages which might result in changes of read-ahead
10190 * We now temporaritly lower the kernel's read_ahead_kb variable
10191 when collecting read-ahead data to ensure the kernel's
10192 built-in read-ahead does not add noise to our measurements
10193 of necessary blocks to pre-cache.
10195 * There's now RequiresMountsFor= to add automatic dependencies
10196 for all mounts necessary for a specific file system path.
10198 * MountAuto= and SwapAuto= have been removed from
10199 system.conf. Mounting file systems at boot has to take place
10202 * nspawn now learned a new switch --uuid= to set the machine
10203 ID on the command line.
10205 * nspawn now learned the -b switch to automatically search
10206 for an init system.
10208 * vt102 is now the default TERM for serial TTYs, upgraded from
10211 * systemd-logind now works on VT-less systems.
10213 * The build tree has been reorganized. The individual
10214 components now have directories of their own.
10216 * A new condition type ConditionPathIsReadWrite= is now available.
10218 * nspawn learned the new -C switch to create cgroups for the
10219 container in other hierarchies.
10221 * We now have support for hardware watchdogs, configurable in
10224 * The scheduled shutdown logic now has a public API.
10226 * We now mount /tmp as tmpfs by default, but this can be
10227 masked and /etc/fstab can override it.
10229 * Since udisks does not make use of /media anymore we are not
10230 mounting a tmpfs on it anymore.
10232 * journalctl gained a new --local switch to only interleave
10233 locally generated journal files.
10235 * We can now load the IMA policy at boot automatically.
10237 * The GTK tools have been split off into a systemd-ui.
10239 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
10240 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
10241 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
10242 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
10243 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
10244 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
10245 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
10246 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
10247 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
10252 * This is mostly a bugfix release
10254 * Support optional initialization of the machine ID from the
10255 KVM or container configured UUID.
10257 * Support immediate reboots with "systemctl reboot -ff"
10259 * Show /etc/os-release data in systemd-analyze output
10261 * Many bugfixes for the journal, including endianness fixes and
10262 ensuring that disk space enforcement works
10264 * sd-login.h is C++ compatible again
10266 * Extend the /etc/os-release format on request of the Debian
10269 * We now refuse non-UTF8 strings used in various configuration
10270 and unit files. This is done to ensure we do not pass invalid
10271 data over D-Bus or expose it elsewhere.
10273 * Register Mimo USB Screens as suitable for automatic seat
10276 * Read SELinux client context from journal clients in a race
10279 * Reorder configuration file lookup order. /etc now always
10280 overrides /run in order to allow the administrator to always
10281 and unconditionally override vendor-supplied or
10282 automatically generated data.
10284 * The various user visible bits of the journal now have man
10285 pages. We still lack man pages for the journal API calls
10288 * We now ship all man pages in HTML format again in the
10291 Contributions from: Dave Reisner, Dirk Eibach, Frederic
10292 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
10293 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
10298 * This is mostly a bugfix release
10300 * systems lacking /etc/os-release are no longer supported.
10302 * Various functionality updates to libsystemd-login.so
10304 * Track class of PAM logins to distinguish greeters from
10305 normal user logins.
10307 Contributions from: Kay Sievers, Lennart Poettering, Michael
10312 * This is an important bugfix release for v41.
10314 * Building man pages is now optional which should be useful
10315 for those building systemd from git but unwilling to install
10318 * Watchdog support for supervising services is now usable. In
10319 a future release support for hardware watchdogs
10320 (i.e. /dev/watchdog) will be added building on this.
10322 * Service start rate limiting is now configurable and can be
10323 turned off per service. When a start rate limit is hit a
10324 reboot can automatically be triggered.
10326 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
10328 Contributions from: Benjamin Franzke, Bill Nottingham,
10329 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
10330 Schmidt, Michał Górny, Piotr Drąg
10334 * The systemd binary is installed /usr/lib/systemd/systemd now;
10335 An existing /sbin/init symlink needs to be adapted with the
10338 * The code that loads kernel modules has been ported to invoke
10339 libkmod directly, instead of modprobe. This means we do not
10340 support systems with module-init-tools anymore.
10342 * Watchdog support is now already useful, but still not
10345 * A new kernel command line option systemd.setenv= is
10346 understood to set system wide environment variables
10347 dynamically at boot.
10349 * We now limit the set of capabilities of systemd-journald.
10351 * We now set SIGPIPE to ignore by default, since it only is
10352 useful in shell pipelines, and has little use in general
10353 code. This can be disabled with IgnoreSIPIPE=no in unit
10356 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
10357 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
10362 * This is mostly a bugfix release
10364 * We now expose the reason why a service failed in the
10365 "Result" D-Bus property.
10367 * Rudimentary service watchdog support (will be completed over
10368 the next few releases.)
10370 * When systemd forks off in order execute some service we will
10371 now immediately changes its argv[0] to reflect which process
10372 it will execute. This is useful to minimize the time window
10373 with a generic argv[0], which makes bootcharts more useful
10375 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
10376 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
10377 Mike Kazantsev, Ray Strode
10381 * This is mostly a test release, but incorporates many
10384 * New systemd-cgtop tool to show control groups by their
10387 * Linking against libacl for ACLs is optional again. If
10388 disabled, support tracking device access for active logins
10389 goes becomes unavailable, and so does access to the user
10390 journals by the respective users.
10392 * If a group "adm" exists, journal files are automatically
10393 owned by them, thus allow members of this group full access
10394 to the system journal as well as all user journals.
10396 * The journal now stores the SELinux context of the logging
10397 client for all entries.
10399 * Add C++ inclusion guards to all public headers
10401 * New output mode "cat" in the journal to print only text
10402 messages, without any meta data like date or time.
10404 * Include tiny X server wrapper as a temporary stop-gap to
10405 teach XOrg udev display enumeration. This is used by display
10406 managers such as gdm, and will go away as soon as XOrg
10407 learned native udev hotplugging for display devices.
10409 * Add new systemd-cat tool for executing arbitrary programs
10410 with STDERR/STDOUT connected to the journal. Can also act as
10411 BSD logger replacement, and does so by default.
10413 * Optionally store all locally generated coredumps in the
10414 journal along with meta data.
10416 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
10417 writing short strings to files (for usage for /sys), and for
10418 creating symlinks, character and block device nodes.
10420 * New unit file option ControlGroupPersistent= to make cgroups
10421 persistent, following the mechanisms outlined in
10422 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
10424 * Support multiple local RTCs in a sane way
10426 * No longer monopolize IO when replaying readahead data on
10427 rotating disks, since we might starve non-file-system IO to
10428 death, since fanotify() will not see accesses done by blkid,
10431 * Do not show kernel threads in systemd-cgls anymore, unless
10432 requested with new -k switch.
10434 Contributions from: Dan Horák, Kay Sievers, Lennart
10435 Poettering, Michal Schmidt
10439 * This is mostly a test release, but incorporates many
10442 * The git repository moved to:
10443 git://anongit.freedesktop.org/systemd/systemd
10444 ssh://git.freedesktop.org/git/systemd/systemd
10446 * First release with the journal
10447 http://0pointer.de/blog/projects/the-journal.html
10449 * The journal replaces both systemd-kmsg-syslogd and
10450 systemd-stdout-bridge.
10452 * New sd_pid_get_unit() API call in libsystemd-logind
10454 * Many systemadm clean-ups
10456 * Introduce remote-fs-pre.target which is ordered before all
10457 remote mounts and may be used to start services before all
10460 * Added Mageia support
10462 * Add bash completion for systemd-loginctl
10464 * Actively monitor PID file creation for daemons which exit in
10465 the parent process before having finished writing the PID
10466 file in the daemon process. Daemons which do this need to be
10467 fixed (i.e. PID file creation must have finished before the
10468 parent exits), but we now react a bit more gracefully to them.
10470 * Add colourful boot output, mimicking the well-known output
10471 of existing distributions.
10473 * New option PassCredentials= for socket units, for
10474 compatibility with a recent kernel ABI breakage.
10476 * /etc/rc.local is now hooked in via a generator binary, and
10477 thus will no longer act as synchronization point during
10480 * systemctl list-unit-files now supports --root=.
10482 * systemd-tmpfiles now understands two new commands: z, Z for
10483 relabelling files according to the SELinux database. This is
10484 useful to apply SELinux labels to specific files in /sys,
10485 among other things.
10487 * Output of SysV services is now forwarded to both the console
10488 and the journal by default, not only just the console.
10490 * New man pages for all APIs from libsystemd-login.
10492 * The build tree got reorganized and the build system is a
10493 lot more modular allowing embedded setups to specifically
10494 select the components of systemd they are interested in.
10496 * Support for Linux systems lacking the kernel VT subsystem is
10499 * configure's --with-rootdir= got renamed to
10500 --with-rootprefix= to follow the naming used by udev and
10503 * Unless specified otherwise we will now install to /usr instead
10504 of /usr/local by default.
10506 * Processes with '@' in argv[0][0] are now excluded from the
10507 final shut-down killing spree, following the logic explained
10509 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
10511 * All processes remaining in a service cgroup when we enter
10512 the START or START_PRE states are now killed with
10513 SIGKILL. That means it is no longer possible to spawn
10514 background processes from ExecStart= lines (which was never
10515 supported anyway, and bad style).
10517 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
10518 reloading of units together.
10520 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
10521 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
10522 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
10523 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
10524 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek