1 systemd System and Service Manager
3 CHANGES WITH 240 in spe:
5 * A new service type has been added: Type=exec. It's very similar to
6 Type=simple but ensures the service manager will wait for both fork()
7 and execve() of the main service binary to complete before proceeding
8 with follow-up units. This is primarily useful so that the manager
9 propagates any errors in the preparation phase of service execution
10 back to the job that requested the unit to be started. For example,
11 consider a service that has ExecStart= set to a file system binary
12 that doesn't exist. With Type=simple starting the unit would be
13 considered instantly successful, as only fork() has to complete
14 successfully and the manager does not wait for execve(), and hence
15 its failure is seen "too late". With the new Type=exec service type
16 starting the unit will fail, as the manager will wait for the
17 execve() and notice its failure, which is then propagated back to the
20 NOTE: with the next release 241 of systemd we intend to change the
21 systemd-run tool to default to Type=exec for transient services
22 started by it. This should be mostly safe, but in specific corner
23 cases might result in problems, as the systemd-run tool will then
24 block on NSS calls (such as user name look-ups due to User=) done
25 between the fork() and execve(), which under specific circumstances
26 might cause problems. It is recommended to specify "-p Type=simple"
27 explicitly in the few cases where this applies. For regular,
28 non-transient services (i.e. those defined with unit files on disk)
29 we will continue to default to Type=simple.
31 * The Linux kernel's current default RLIMIT_NOFILE resource limit for
32 userspace processes is set to 1024 (soft) and 4096
33 (hard). Previously, systemd passed this on unmodified to all
34 processes it forked off. With this systemd release the hard limit
35 systemd passes on is increased to 512K, overriding the kernel's
36 defaults and substantially increasing the number of simultaneous file
37 descriptors unprivileged userspace processes can allocate. Note that
38 the soft limit remains at 1024 for compatibility reasons: the
39 traditional UNIX select() call cannot deal with file descriptors >=
40 1024 and increasing the soft limit globally might thus result in
41 programs unexpectedly allocating a high file descriptor and thus
42 failing abnormally when attempting to use it with select() (of
43 course, programs shouldn't use select() anymore, and prefer
44 poll()/epoll, but the call unfortunately remains undeservedly popular
45 at this time). This change reflects the fact that file descriptor
46 handling in the Linux kernel has been optimized in more recent
47 kernels and allocating large numbers of them should be much cheaper
48 both in memory and in performance than it used to be. Programs that
49 want to take benefit of the increased limit have to "opt-in" into
50 high file descriptors explicitly by raising their soft limit. Of
51 course, when they do that they must acknowledge that they cannot use
52 select() anymore (and neither can any shared library they use — or
53 any shared library used by any shared library they use and so on).
54 Which default hard limit is most appropriate is of course hard to
55 decide. However, given reports that ~300K file descriptors are used
56 in real-life applications we believe 512K is sufficiently high as new
57 default for now. Note that there are also reports that using very
58 high hard limits (e.g. 1G) is problematic: some software allocates
59 large arrays with one element for each potential file descriptor
60 (Java, …) — a high hard limit thus triggers excessively large memory
61 allocations in these applications. Hopefully, the new default of 512K
62 is a good middle ground: higher than what real-life applications
63 currently need, and low enough for avoid triggering excessively large
64 allocations in problematic software. (And yes, somebody should fix
67 * The fs.nr_open and fs.file-max sysctls are now automatically bumped
68 to the highest possible values, as separate accounting of file
69 descriptors is no longer necessary, as memcg tracks them correctly as
70 part of the memory accounting anyway. Thus, from the four limits on
71 file descriptors currently enforced (fs.file-max, fs.nr_open,
72 RLIMIT_NOFILE hard, RLIMIT_NOFILE soft) we turn off the first two,
73 and keep only the latter two. A set of build-time options
74 (-Dbump-proc-sys-fs-file-max=no and -Dbump-proc-sys-fs-nr-open=no)
75 has been added to revert this change in behaviour, which might be
76 an option for systems that turn off memcg in the kernel.
78 * When no /etc/locale.conf file exists (and hence no locale settings
79 are in place), systemd will now use the "C.UTF-8" locale by default,
80 and set LANG= to it. This locale is supported by various
81 distributions including Fedora, with clear indications that upstream
82 glibc is going to make it available too. This locale enables UTF-8
83 mode by default, which appears appropriate for 2018.
85 * The "net.ipv4.conf.all.rp_filter" sysctl will now be set to 2 by
86 default. This effectively switches the RFC3704 Reverse Path filtering
87 from Strict mode to Loose mode. This is more appropriate for hosts
88 that have multiple links with routes to the same networks (e.g.
89 a client with a Wi-Fi and Ethernet both connected to the internet).
91 Consult the kernel documentation for details on this sysctl:
92 https://www.kernel.org/doc/Documentation/networking/ip-sysctl.txt
94 * CPUAccounting=yes no longer enables the CPU controller when using
95 kernel 4.15+ and the unified cgroup hierarchy, as required accounting
96 statistics are now provided independently from the CPU controller.
98 * Support for disabling a particular cgroup controller within a sub-tree
99 has been added through the DisableControllers= directive.
101 * The new "MemoryMin=" unit file property may now be used to set the
102 memory usage protection limit of processes invoked by the unit. This
103 controls the cgroupsv2 memory.min attribute. Similarly, the new
104 "IODeviceLatencyTargetSec=" property has been added, wrapping the new
105 cgroupsv2 io.latency cgroup property for configuring per-service I/O
108 * systemd now supports the cgroupsv2 devices BPF logic, as counterpart
109 to the cgroupsv1 "devices" cgroup controller.
111 * systemd-escape now is able to combine --unescape with --template. It
112 also learnt a new option --instance for extracting and unescaping the
113 instance part of a unit name.
115 * sd-bus now provides the sd_bus_message_readv() which is similar to
116 sd_bus_message_read() but takes a va_list object. The pair
117 sd_bus_set_method_call_timeout() and sd_bus_get_method_call_timeout()
118 has been added for configuring the default method call timeout to
119 use. sd_bus_error_move() may be used to efficiently move the contents
120 from one sd_bus_error structure to another, invalidating the
121 source. sd_bus_set_close_on_exit() and sd_bus_get_close_on_exit() may
122 be used to control whether a bus connection object is automatically
123 flushed when an sd-event loop is exited.
125 * When processing classic BSD syslog log messages, journald will now
126 save the original time-stamp string supplied in the new
127 SYSLOG_TIMESTAMP= journal field. This permits consumers to
128 reconstruct the original BSD syslog message more correctly.
130 * StandardOutput=/StandardError= in service files gained support for
131 new "append:…" parameters, for connecting STDOUT/STDERR of a service
132 to a file, and appending to it.
134 * The signal to use as last step of killing of unit processes is now
135 configurable. Previously it was hard-coded to SIGKILL, which may now
136 be overridden with the new KillSignal= setting. Note that this is the
137 signal used when regular termination (i.e. SIGTERM) does suffice.
138 Similarly, the signal used when aborting a program in case of a
139 watchdog timeout may now be configured too (WatchdogSignal=).
141 * The XDG_SESSION_DESKTOP environment variable may now be configured in
142 the pam_systemd argument line, using the new desktop= switch. This is
143 useful to initialize it properly from a display manager without
144 having to touch C code.
146 * Most configuration options that previously accepted percentage values
147 now also accept permille values with the '‰' suffix (instead of '%').
149 * systemd-logind will offer hibernation only if the currently used
150 kernel image is still available on disk.
152 * systemd-resolved may now optionally use OpenSSL instead of GnuTLS for
155 * systemd-resolved's configuration file resolved.conf gained a new
156 option ReadEtcHosts= which may be used to turn off processing and
157 honoring /etc/hosts entries.
159 * The "--wait" switch may now be passed to "systemctl
160 is-system-running", in which case the tool will synchronously wait
161 until the system finished start-up.
163 * hostnamed gained a new bus call to determine the DMI product UUID.
165 * On x86-64 systemd will now prefer using the RDRAND processor
166 instruction over /dev/urandom whenever it requires randomness that
167 neither has to be crypto-grade nor should be reproducible. This
168 should substantially reduce the amount of entropy systemd requests
169 from the kernel during initialization on such systems, though not
170 reduce it to zero. (Why not zero? systemd still needs to allocate
171 UUIDs and such uniquely, which require high-quality randomness.)
173 * networkd gained support for Foo-Over-UDP, ERSPAN and ISATAP
174 tunnels. It also gained a new option ForceDHCPv6PDOtherInformation=
175 for forcing the "Other Information" bit in IPv6 RA messages. The
176 bonding logic gained three new options AdActorSystemPriority=,
177 AdUserPortKey=, AdActorSystem= for configuring various 802.3ad
178 aspects. The tunnel logic gained a new IPv6RapidDeploymentPrefix=
179 option for configuring IPv6 Rapid Deployment. The policy rule logic
180 gained four new options IPProtocol=, SourcePort= and
181 DestinationPort=, InvertRule=. The bridge logic gained support for
182 the MulticastToUnicast= option. networkd also gained support for
183 configuring static IPv4 ARP or IPv6 neighbor entries.
185 * .preset files (as read by 'systemctl preset') may now be used to
186 instantiate services.
188 * /etc/crypttab now understands the sector-size= option to configure
189 the sector size for an encrypted partition.
191 * Key material for encrypted disks may now be placed on a formatted
192 medium, and referenced from /etc/crypttab by the UUID of the file
193 system, followed by "=" suffixed by the path to the key file.
195 * The "collect" udev component has been removed without replacement, as
196 it is neither used nor maintained.
198 * When the RuntimeDirectory=, StateDirectory=, CacheDirectory=,
199 LogsDirectory=, ConfigurationDirectory= settings are used in a
200 service the executed processes will now receive a set of environment
201 variables containing the full paths of these directories.
202 Specifically, RUNTIME_DIRECTORY=, STATE_DIRECTORY, CACHE_DIRECTORY,
203 LOGS_DIRECTORY, CONFIGURATION_DIRECTORY are now set if these options
204 are used. Note that these options may be used multiple times per
205 service in which case the resulting paths will be concatenated and
208 * Predictable interface naming has been extended to cover InfiniBand
209 NICs. They will be exposed with an "ib" prefix.
211 * tmpfiles.d/ line types may now be suffixed with a '-' character, in
212 which case the respective line failing is ignored.
214 * .link files may now be used to configure the equivalent to the
215 "ethtool advertise" commands.
217 * The sd-device.h and sd-hwdb.h APIs are now exported, as an
218 alternative to libudev.h. Previously, the latter was just an internal
219 wrapper around the former, but now these two APIs are exposed
222 * sd-id128.h gained a new function sd_id128_get_boot_app_specific()
223 which calculates an app-specific boot ID similar to how
224 sd_id128_get_machine_app_specific() generates an app-specific machine
227 * A new tool systemd-id128 has been added that can be used to determine
228 and generate various 128bit IDs.
230 * /etc/os-release gained two new standardized fields DOCUMENTATION_URL=
233 * systemd-hibernate-resume-generator will now honor the "noresume"
234 kernel command line option, in which case it will bypass resuming
235 from any hibernated image.
237 * The systemd-sleep.conf configuration file gained new options
238 AllowSuspend=, AllowHibernation=, AllowSuspendThenHibernate=,
239 AllowHybridSleep= for prohibiting specific sleep modes even if the
242 * portablectl is now officially supported and has thus moved to
245 * bootctl learnt the two new commands "set-default" and "set-oneshot"
246 for setting the default boot loader item to boot to (either
247 persistently or only for the next boot). This is currently only
248 compatible with sd-boot, but may be implemented on other boot loaders
249 too, that follow the boot loader interface. The updated interface is
252 https://systemd.io/BOOT_LOADER_INTERFACE
254 * A new kernel command line option systemd.early_core_pattern= is now
255 understood which may be used to influence the core_pattern PID 1
256 installs during early boot.
258 * busctl learnt two new options -j and --json= for outputting method
259 call replies, properties and monitoring output in JSON.
261 * journalctl's JSON output now supports simple ANSI coloring as well as
262 a new "json-seq" mode for generating RFC7464 output.
264 * Unit files now support the %g/%G specifiers that resolve to the UNIX
265 group/GID of the service manager runs as, similar to the existing
266 %u/%U specifiers that resolve to the UNIX user/UID.
268 * systemd-logind learnt a new global configuration option
269 UserStopDelaySec= that may be set in logind.conf. It specifies how
270 long the systemd --user instance shall remain started after a user
271 logs out. This is useful to speed up repetitive re-connections of the
272 same user, as it means the user's service manager doesn't have to be
273 stopped/restarted on each iteration, but can be reused between
274 subsequent options. This setting defaults to 10s. systemd-logind also
275 exports two new properties on its Manager D-Bus objects indicating
276 whether the system's lid is currently closed, and whether the system
279 * systemd gained support for a generic boot counting logic, which
280 generically permits automatic reverting to older boot loader entries
281 if newer updated ones don't work. The boot loader side is implemented
282 in sd-boot, but is kept open for other boot loaders too. For details
285 https://systemd.io/AUTOMATIC_BOOT_ASSESSMENT
287 * The SuccessAction=/FailureAction= unit file settings now learnt two
288 new parameters: "exit" and "exit-force", which result in immediate
289 exiting of the service manager, and are only useful in systemd --user
290 and container environments.
292 * Unit files gained support for a pair of options
293 FailureActionExitStatus=/SuccessActionExitStatus= for configuring the
294 exit status to use as service manager exit status when
295 SuccessAction=/FailureAction= is set to exit or exit-force.
297 * A pair of LogRateLimitIntervalSec=/LogRateLimitBurst= per-service
298 options may now be used to configure the log rate limiting applied by
299 journald per-service.
301 * systemd-analyze gained a new verb "timespan" for parsing and
302 normalizing time span values (i.e. strings like "5min 7s 8us").
304 * systemd-analyze also gained a new verb "security" for analyzing the
305 security and sand-boxing settings of services in order to determine an
306 "exposure level" for them, indicating whether a service would benefit
307 from more sand-boxing options turned on for them.
309 * "systemd-analyze syscall-filter" will now also show system calls
310 supported by the local kernel but not included in any of the defined
313 * .nspawn files now understand the Ephemeral= setting, matching the
314 --ephemeral command line switch.
316 * sd-event gained the new APIs sd_event_source_get_floating() and
317 sd_event_source_set_floating() for controlling whether a specific
318 event source is "floating", i.e. destroyed along with the even loop
321 * Unit objects on D-Bus gained a new "Refs" property that lists all
322 clients that currently have a reference on the unit (to ensure it is
325 * The JoinControllers= option in system.conf is no longer supported, as
326 it didn't work correctly, is hard to support properly, is legacy (as
327 the concept only exists on cgroupsv1) and apparently wasn't used.
329 * Journal messages that are generated whenever a unit enters the failed
330 state are now tagged with a unique MESSAGE_ID. Similarly, messages
331 generated whenever a service process exits are now made recognizable,
332 too. A taged message is also emitted whenever a unit enters the
333 "dead" state on success.
335 * systemd-run gained a new switch --working-directory= for configuring
336 the working directory of the service to start. A shortcut -d is
337 equivalent, setting the working directory of the service to the
338 current working directory of the invoking program. The new --shell
339 (or just -S) option has been added for invoking the $SHELL of the
340 caller as a service, and implies --pty --same-dir --wait --collect
341 --service-type=exec. Or in other words, "systemd-run -S" is now the
342 quickest way to quickly get an interactive in a fully clean and
343 well-defined system service context.
345 * machinectl gained a new verb "import-fs" for importing an OS tree
346 from a directory. Moreover, when a directory or tarball is imported
347 and single top-level directory found with the OS itself below the OS
348 tree is automatically mangled and moved one level up.
350 * systemd-importd will no longer set up an implicit btrfs loop-back
351 file system on /var/lib/machines. If one is already set up, it will
354 * A new generator "systemd-run-generator" has been added. It will
355 synthesize a unit from one or more program command lines included in
356 the kernel command line. This is very useful in container managers
359 # systemd-nspawn -i someimage.raw -b systemd.run='"some command line"'
361 This will run "systemd-nspawn" on an image, invoke the specified
362 command line and immediately shut down the container again, returning
363 the command line's exit code.
365 * The block device locking logic is now documented:
367 https://systemd.io/BLOCK_DEVICE_LOCKING
369 * loginctl and machinectl now optionally output the various tables in
370 JSON using the --output= switch. It is our intention to add similar
371 support to systemctl and all other commands.
373 * udevadm's query and trigger verb now optionally take a .device unit
376 * systemd-udevd's network naming logic now understands a new
377 net.naming-scheme= kernel command line switch, which may be used to
378 pick a specific version of the naming scheme. This helps stabilizing
379 interface names even as systemd/udev are updated and the naming logic
382 Contributions from: afg, Alan Jenkins, Aleksei Timofeyev, Alexander
383 Filippov, Alexander Kurtz, Alexey Bogdanenko, Andreas Henriksson,
384 Andrew Jorgensen, Anita Zhang, apnix-uk, Arkan49, Arseny Maslennikov,
385 asavah, Asbjørn Apeland, aszlig, Bastien Nocera, Ben Boeckel, Benedikt
386 Morbach, Benjamin Berg, Carlo Caione, Cedric Viou, Chen Qi, ChenQi1989,
387 Chris Chiu, Chris Down, Chris Morin, Christian Rebischke, Claudius
388 Ellsel, ColinGuthrie, dana, Daniel, Daniele Medri, Daniel Kahn Gillmor,
389 Daniel Rusek, Daniel van Vugt, Dariusz Gadomski, Dave Reisner, David
390 Anderson, Davide Cavalca, David Leeds, David Malcolm, David Strauss,
391 David Tardon, Dimitri John Ledkov, dj-kaktus, Dongsu Park, Elias
392 Probst, Emil Soleyman, Erik Kooistra, Ervin Peters, Evgeni Golov,
393 Evgeny Vereshchagin, Fabrice Fontaine, Faheel Ahmad, faizalluthfi,
394 Felix Yan, Filipe Brandenburger, Franck Bui, Frank Schaefer, Frantisek
395 Sumsal, Gianluca Boiano, Giuseppe Scrivano, glitsj16, Hans de Goede,
396 Harald Hoyer, Harry Mallon, Harshit Jain, hellcp, Helmut Grohne, Henry
397 Tung, Hui Yiqun, imayoda, Insun Pyo, INSUN PYO, Iwan Timmer,
398 jambonmcyeah, Jan Janssen, Jan Pokorný, Jan Synacek, Jason
399 A. Donenfeld, javitoom, Jérémy Nouhaud, Jiuyang liu, João Paulo Rechi
400 Vita, Joe Hershberger, Joe Rayhawk, Joerg Behrmann, Joerg Steffens,
401 Jonas DOREL, Jon Ringle, Josh Soref, Julian Andres Klode, Jürg
402 Billeter, Keith Busch, killermoehre, Kirill Marinushkin, Lennart
403 Poettering, LennartPoettering, Liberasys, Lion Yang, Li Song, Lorenz
404 Hübschle-Schneider, Lubomir Rintel, Lucas Werkmeister, Ludwin Janvier,
405 Lukáš Nykrýn, Luke Shumaker, mal, Marc-Antoine Perennou, Marcin
406 Skarbek, Marco Trevisan (Treviño), Marian Cepok, Mario Hros, Marko
407 Myllynen, Markus Grimm, Martin Pitt, Martin Sobotka, Martin Wilck,
408 Mathieu Trudel-Lapierre, Matthew Leeds, Michael Biebl, Michael Olbrich,
409 Michael 'pbone' Pobega, Michal Koutný, Michal Sekletar, Michal Soltys,
410 Mike Gilbert, Mike Palmer, Muhammet Kara, Neal Gompa, Network Silence,
411 nikolas, NOGISAKA Sadata, Oliver Smith, Patrik Flykt, Pavel Hrdina,
412 Paweł Szewczyk, Peter Hutterer, Piotr Drąg, Ray Strode, remueller,
413 Renaud Métrich, Roman Gushchin, Ronny Chevalier, Rubén Suárez Alvarez,
414 Ruixin Bao, RussianNeuroMancer, Ryutaroh Matsumoto, Saleem Rashid,
415 Samuel Morris, Sandy, scootergrisen, seb128, Sergey Ptashnick, Shawn
416 Landden, Shengyao Xue, Shih-Yuan Lee (FourDollars), Sjoerd Simons,
417 Stephen Gallagher, Steven Allen, Steve Ramage, Susant Sahani, Sven
418 Joachim, Sylvain Plantefève, TanuKaskinen, Tejun Heo, Thiago Macieira,
419 Thomas Blume, Thomas Haller, Thomas H. P. Andersen, Tim Ruffing, TJ,
420 Tobias Jungel, Todd Walton, Tommi Rantala, Tomsod M, Tony Novak,
421 Trevonn, Victor Laskurain, Victor Tapia, Violet Halo, Vojtech Trefny,
422 welaq, William A. Kennington III, William Douglas, Wyatt Ward, Xiang
423 Fan, Xi Ruoyao, Xuanwo, Yann E. MORIN, YmrDtnJu, Yu Watanabe, Zbigniew
424 Jędrzejewski-Szmek, Zhang Xianwei, Zsolt Dollenstein
426 — Somewhere, 2018-xx-yy
430 * NETWORK INTERFACE DEVICE NAMING CHANGES: systemd-udevd's "net_id"
431 builtin will name network interfaces differently than in previous
432 versions for virtual network interfaces created with SR-IOV and NPAR
433 and for devices where the PCI network controller device does not have
434 a slot number associated.
436 SR-IOV virtual devices are now named based on the name of the parent
437 interface, with a suffix of "v<N>", where <N> is the virtual device
438 number. Previously those virtual devices were named as if completely
441 The ninth and later NPAR virtual devices will be named following the
442 scheme used for the first eight NPAR partitions. Previously those
443 devices were not renamed and the kernel default (eth<n>) was used.
445 "net_id" will also generate names for PCI devices where the PCI
446 network controller device does not have an associated slot number
447 itself, but one of its parents does. Previously those devices were
448 not renamed and the kernel default (eth<n>) was used.
450 * AF_INET and AF_INET6 are dropped from RestrictAddressFamilies= in
451 systemd-logind.service. Since v235, IPAddressDeny=any has been set to
452 the unit. So, it is expected that the default behavior of
453 systemd-logind is not changed. However, if distribution packagers or
454 administrators disabled or modified IPAddressDeny= setting by a
455 drop-in config file, then it may be necessary to update the file to
456 re-enable AF_INET and AF_INET6 to support network user name services,
459 * When the RestrictNamespaces= unit property is specified multiple
460 times, then the specified types are merged now. Previously, only the
461 last assignment was used. So, if distribution packagers or
462 administrators modified the setting by a drop-in config file, then it
463 may be necessary to update the file.
465 * When OnFailure= is used in combination with Restart= on a service
466 unit, then the specified units will no longer be triggered on
467 failures that result in restarting. Previously, the specified units
468 would be activated each time the unit failed, even when the unit was
469 going to be restarted automatically. This behaviour contradicted the
470 documentation. With this release the code is adjusted to match the
473 * systemd-tmpfiles will now print a notice whenever it encounters
474 tmpfiles.d/ lines referencing the /var/run/ directory. It will
475 recommend reworking them to use the /run/ directory instead (for
476 which /var/run/ is simply a symlinked compatibility alias). This way
477 systemd-tmpfiles can properly detect line conflicts and merge lines
478 referencing the same file by two paths, without having to access
481 * systemctl disable/unmask/preset/preset-all cannot be used with
482 --runtime. Previously this was allowed, but resulted in unintuitive
483 behaviour that wasn't useful. systemctl disable/unmask will now undo
484 both runtime and persistent enablement/masking, i.e. it will remove
485 any relevant symlinks both in /run and /etc.
487 * Note that all long-running system services shipped with systemd will
488 now default to a system call whitelist (rather than a blacklist, as
489 before). In particular, systemd-udevd will now enforce one too. For
490 most cases this should be safe, however downstream distributions
491 which disabled sandboxing of systemd-udevd (specifically the
492 MountFlags= setting), might want to disable this security feature
493 too, as the default whitelisting will prohibit all mount, swap,
494 reboot and clock changing operations from udev rules.
496 * sd-boot acquired new loader configuration settings to optionally turn
497 off Windows and MacOS boot partition discovery as well as
498 reboot-into-firmware menu items. It is also able to pick a better
499 screen resolution for HiDPI systems, and now provides loader
500 configuration settings to change the resolution explicitly.
502 * systemd-resolved now supports DNS-over-TLS. It's still
503 turned off by default, use DNSOverTLS=opportunistic to turn it on in
504 resolved.conf. We intend to make this the default as soon as couple
505 of additional techniques for optimizing the initial latency caused by
506 establishing a TLS/TCP connection are implemented.
508 * systemd-resolved.service and systemd-networkd.service now set
509 DynamicUser=yes. The users systemd-resolve and systemd-network are
510 not created by systemd-sysusers anymore.
512 NOTE: This has a chance of breaking nss-ldap and similar NSS modules
513 that embedd a network facing module into any process using getpwuid()
514 or related call: the dynamic allocation of the user ID for
515 systemd-resolved.service means the service manager has to check NSS
516 if the user name is already taken when forking off the service. Since
517 the user in the common case won't be defined in /etc/passwd the
518 lookup is likely to trigger nss-ldap which in turn might use NSS to
519 ask systemd-resolved for hostname lookups. This will hence result in
520 a deadlock: a user name lookup in order to start
521 systemd-resolved.service will result in a host name lookup for which
522 systemd-resolved.service needs to be started already. There are
523 multiple ways to work around this problem: pre-allocate the
524 "systemd-resolve" user on such systems, so that nss-ldap won't be
525 triggered; or use a different NSS package that doesn't do networking
526 in-process but provides a local asynchronous name cache; or configure
527 the NSS package to avoid lookups for UIDs in the range `pkg-config
528 systemd --variable=dynamicuidmin` … `pkg-config systemd
529 --variable=dynamicuidmax`, so that it does not consider itself
530 authoritative for the same UID range systemd allocates dynamic users
533 * The systemd-resolve tool has been renamed to resolvectl (it also
534 remains available under the old name, for compatibility), and its
535 interface is now verb-based, similar in style to the other <xyz>ctl
536 tools, such as systemctl or loginctl.
538 * The resolvectl/systemd-resolve tool also provides 'resolvconf'
539 compatibility. It may be symlinked under the 'resolvconf' name, in
540 which case it will take arguments and input compatible with the
541 Debian and FreeBSD resolvconf tool.
543 * Support for suspend-then-hibernate has been added, i.e. a sleep mode
544 where the system initially suspends, and after a time-out resumes and
547 * networkd's ClientIdentifier= now accepts a new option "duid-only". If
548 set the client will only send a DUID as client identifier.
550 * The nss-systemd glibc NSS module will now enumerate dynamic users and
551 groups in effect. Previously, it could resolve UIDs/GIDs to user
552 names/groups and vice versa, but did not support enumeration.
554 * journald's Compress= configuration setting now optionally accepts a
555 byte threshold value. All journal objects larger than this threshold
556 will be compressed, smaller ones will not. Previously this threshold
557 was not configurable and set to 512.
559 * A new system.conf setting NoNewPrivileges= is now available which may
560 be used to turn off acquisition of new privileges system-wide
561 (i.e. set Linux' PR_SET_NO_NEW_PRIVS for PID 1 itself, and thus also
562 for all its children). Note that turning this option on means setuid
563 binaries and file system capabilities lose their special powers.
564 While turning on this option is a big step towards a more secure
565 system, doing so is likely to break numerous pre-existing UNIX tools,
566 in particular su and sudo.
568 * A new service systemd-time-sync-wait.service has been added. If
569 enabled it will delay the time-sync.target unit at boot until time
570 synchronization has been received from the network. This
571 functionality is useful on systems lacking a local RTC or where it is
572 acceptable that the boot process shall be delayed by external network
575 * When hibernating, systemd will now inform the kernel of the image
576 write offset, on kernels new enough to support this. This means swap
577 files should work for hibernation now.
579 * When loading unit files, systemd will now look for drop-in unit files
580 extensions in additional places. Previously, for a unit file name
581 "foo-bar-baz.service" it would look for dropin files in
582 "foo-bar-baz.service.d/*.conf". Now, it will also look in
583 "foo-bar-.service.d/*.conf" and "foo-.service.d/", i.e. at the
584 service name truncated after all inner dashes. This scheme allows
585 writing drop-ins easily that apply to a whole set of unit files at
586 once. It's particularly useful for mount and slice units (as their
587 naming is prefix based), but is also useful for service and other
588 units, for packages that install multiple unit files at once,
589 following a strict naming regime of beginning the unit file name with
590 the package's name. Two new specifiers are now supported in unit
591 files to match this: %j and %J are replaced by the part of the unit
592 name following the last dash.
594 * Unit files and other configuration files that support specifier
595 expansion now understand another three new specifiers: %T and %V will
596 resolve to /tmp and /var/tmp respectively, or whatever temporary
597 directory has been set for the calling user. %E will expand to either
598 /etc (for system units) or $XDG_CONFIG_HOME (for user units).
600 * The ExecStart= lines of unit files are no longer required to
601 reference absolute paths. If non-absolute paths are specified the
602 specified binary name is searched within the service manager's
603 built-in $PATH, which may be queried with 'systemd-path
604 search-binaries-default'. It's generally recommended to continue to
605 use absolute paths for all binaries specified in unit files.
607 * Units gained a new load state "bad-setting", which is used when a
608 unit file was loaded, but contained fatal errors which prevent it
609 from being started (for example, a service unit has been defined
610 lacking both ExecStart= and ExecStop= lines).
612 * coredumpctl's "gdb" verb has been renamed to "debug", in order to
613 support alternative debuggers, for example lldb. The old name
614 continues to be available however, for compatibility reasons. Use the
615 new --debugger= switch or the $SYSTEMD_DEBUGGER environment variable
616 to pick an alternative debugger instead of the default gdb.
618 * systemctl and the other tools will now output escape sequences that
619 generate proper clickable hyperlinks in various terminal emulators
620 where useful (for example, in the "systemctl status" output you can
621 now click on the unit file name to quickly open it in the
622 editor/viewer of your choice). Note that not all terminal emulators
623 support this functionality yet, but many do. Unfortunately, the
624 "less" pager doesn't support this yet, hence this functionality is
625 currently automatically turned off when a pager is started (which
626 happens quite often due to auto-paging). We hope to remove this
627 limitation as soon as "less" learns these escape sequences. This new
628 behaviour may also be turned off explicitly with the $SYSTEMD_URLIFY
629 environment variable. For details on these escape sequences see:
630 https://gist.github.com/egmontkob/eb114294efbcd5adb1944c9f3cb5feda
632 * networkd's .network files now support a new IPv6MTUBytes= option for
633 setting the MTU used by IPv6 explicitly as well as a new MTUBytes=
634 option in the [Route] section to configure the MTU to use for
635 specific routes. It also gained support for configuration of the DHCP
636 "UserClass" option through the new UserClass= setting. It gained
637 three new options in the new [CAN] section for configuring CAN
638 networks. The MULTICAST and ALLMULTI interface flags may now be
639 controlled explicitly with the new Multicast= and AllMulticast=
642 * networkd will now automatically make use of the kernel's route
643 expiration feature, if it is available.
645 * udevd's .link files now support setting the number of receive and
646 transmit channels, using the RxChannels=, TxChannels=,
647 OtherChannels=, CombinedChannels= settings.
649 * Support for UDPSegmentationOffload= has been removed, given its
650 limited support in hardware, and waning software support.
652 * networkd's .netdev files now support creating "netdevsim" interfaces.
654 * PID 1 learnt a new bus call GetUnitByControlGroup() which may be used
655 to query the unit belonging to a specific kernel control group.
657 * systemd-analyze gained a new verb "cat-config", which may be used to
658 dump the contents of any configuration file, with all its matching
659 drop-in files added in, and honouring the usual search and masking
660 logic applied to systemd configuration files. For example use
661 "systemd-analyze cat-config systemd/system.conf" to get the complete
662 system configuration file of systemd how it would be loaded by PID 1
663 itself. Similar to this, various tools such as systemd-tmpfiles or
664 systemd-sysusers, gained a new option "--cat-config", which does the
665 corresponding operation for their own configuration settings. For
666 example, "systemd-tmpfiles --cat-config" will now output the full
667 list of tmpfiles.d/ lines in place.
669 * timedatectl gained three new verbs: "show" shows bus properties of
670 systemd-timedated, "timesync-status" shows the current NTP
671 synchronization state of systemd-timesyncd, and "show-timesync"
672 shows bus properties of systemd-timesyncd.
674 * systemd-timesyncd gained a bus interface on which it exposes details
677 * A new environment variable $SYSTEMD_TIMEDATED_NTP_SERVICES is now
678 understood by systemd-timedated. It takes a colon-separated list of
679 unit names of NTP client services. The list is used by
680 "timedatectl set-ntp".
682 * systemd-nspawn gained a new --rlimit= switch for setting initial
683 resource limits for the container payload. There's a new switch
684 --hostname= to explicitly override the container's hostname. A new
685 --no-new-privileges= switch may be used to control the
686 PR_SET_NO_NEW_PRIVS flag for the container payload. A new
687 --oom-score-adjust= switch controls the OOM scoring adjustment value
688 for the payload. The new --cpu-affinity= switch controls the CPU
689 affinity of the container payload. The new --resolv-conf= switch
690 allows more detailed control of /etc/resolv.conf handling of the
691 container. Similarly, the new --timezone= switch allows more detailed
692 control of /etc/localtime handling of the container.
694 * systemd-detect-virt gained a new --list switch, which will print a
695 list of all currently known VM and container environments.
697 * Support for "Portable Services" has been added, see
698 doc/PORTABLE_SERVICES.md for details. Currently, the support is still
699 experimental, but this is expected to change soon. Reflecting this
700 experimental state, the "portablectl" binary is not installed into
701 /usr/bin yet. The binary has to be called with the full path
702 /usr/lib/systemd/portablectl instead.
704 * journalctl's and systemctl's -o switch now knows a new log output
705 mode "with-unit". The output it generates is very similar to the
706 regular "short" mode, but displays the unit name instead of the
707 syslog tag for each log line. Also, the date is shown with timezone
708 information. This mode is probably more useful than the classic
709 "short" output mode for most purposes, except where pixel-perfect
710 compatibility with classic /var/log/messages formatting is required.
712 * A new --dump-bus-properties switch has been added to the systemd
713 binary, which may be used to dump all supported D-Bus properties.
714 (Options which are still supported, but are deprecated, are *not*
717 * sd-bus gained a set of new calls:
718 sd_bus_slot_set_floating()/sd_bus_slot_get_floating() may be used to
719 enable/disable the "floating" state of a bus slot object,
720 i.e. whether the slot object pins the bus it is allocated for into
721 memory or if the bus slot object gets disconnected when the bus goes
722 away. sd_bus_open_with_description(),
723 sd_bus_open_user_with_description(),
724 sd_bus_open_system_with_description() may be used to allocate bus
725 objects and set their description string already during allocation.
727 * sd-event gained support for watching inotify events from the event
728 loop, in an efficient way, sharing inotify handles between multiple
729 users. For this a new function sd_event_add_inotify() has been added.
731 * sd-event and sd-bus gained support for calling special user-supplied
732 destructor functions for userdata pointers associated with
733 sd_event_source, sd_bus_slot, and sd_bus_track objects. For this new
734 functions sd_bus_slot_set_destroy_callback,
735 sd_bus_slot_get_destroy_callback, sd_bus_track_set_destroy_callback,
736 sd_bus_track_get_destroy_callback,
737 sd_event_source_set_destroy_callback,
738 sd_event_source_get_destroy_callback have been added.
740 * The "net.ipv4.tcp_ecn" sysctl will now be turned on by default.
742 * PID 1 will now automatically reschedule .timer units whenever the
743 local timezone changes. (They previously got rescheduled
744 automatically when the system clock changed.)
746 * New documentation has been added to document cgroups delegation,
747 portable services and the various code quality tools we have set up:
749 https://github.com/systemd/systemd/blob/master/docs/CGROUP_DELEGATION.md
750 https://github.com/systemd/systemd/blob/master/docs/PORTABLE_SERVICES.md
751 https://github.com/systemd/systemd/blob/master/docs/CODE_QUALITY.md
753 * The Boot Loader Specification has been added to the source tree.
755 https://github.com/systemd/systemd/blob/master/docs/BOOT_LOADER_SPECIFICATION.md
757 While moving it into our source tree we have updated it and further
758 changes are now accepted through the usual github PR workflow.
760 * pam_systemd will now look for PAM userdata fields systemd.memory_max,
761 systemd.tasks_max, systemd.cpu_weight, systemd.io_weight set by
762 earlier PAM modules. The data in these fields is used to initialize
763 the session scope's resource properties. Thus external PAM modules
764 may now configure per-session limits, for example sourced from
765 external user databases.
767 * socket units with Accept=yes will now maintain a "refused" counter in
768 addition to the existing "accepted" counter, counting connections
769 refused due to the enforced limits.
771 * The "systemd-path search-binaries-default" command may now be use to
772 query the default, built-in $PATH PID 1 will pass to the services it
775 * A new unit file setting PrivateMounts= has been added. It's a boolean
776 option. If enabled the unit's processes are invoked in their own file
777 system namespace. Note that this behaviour is also implied if any
778 other file system namespacing options (such as PrivateTmp=,
779 PrivateDevices=, ProtectSystem=, …) are used. This option is hence
780 primarily useful for services that do not use any of the other file
781 system namespacing options. One such service is systemd-udevd.service
782 wher this is now used by default.
784 * ConditionSecurity= gained a new value "uefi-secureboot" that is true
785 when the system is booted in UEFI "secure mode".
787 * A new unit "system-update-pre.target" is added, which defines an
788 optional synchronization point for offline system updates, as
789 implemented by the pre-existing "system-update.target" unit. It
790 allows ordering services before the service that executes the actual
791 update process in a generic way.
793 Contributions from: Adam Duskett, Alan Jenkins, Alessandro Casale,
794 Alexander Kurtz, Alex Gartrell, Anssi Hannula, Arnaud Rebillout, Brian
795 J. Murrell, Bruno Vernay, Chris Lamb, Chris Lesiak, Christian Brauner,
796 Christian Hesse, Christian Rebischke, Colin Guthrie, Daniel Dao, Daniel
797 Lin, Danylo Korostil, Davide Cavalca, David Tardon, Dimitri John
798 Ledkov, Dmitriy Geels, Douglas Christman, Elia Geretto, emelenas, Emil
799 Velikov, Evgeny Vereshchagin, Felipe Sateler, Feng Sun, Filipe
800 Brandenburger, Franck Bui, futpib, Giuseppe Scrivano, Guillem Jover,
801 guixxx, Hannes Reinecke, Hans de Goede, Harald Hoyer, Henrique Dante de
802 Almeida, Hiram van Paassen, Ian Miell, Igor Gnatenko, Ivan Shapovalov,
803 Iwan Timmer, James Cowgill, Jan Janssen, Jan Synacek, Jared Kazimir,
804 Jérémy Rosen, João Paulo Rechi Vita, Joost Heitbrink, Jui-Chi Ricky
805 Liang, Jürg Billeter, Kai-Heng Feng, Karol Augustin, Kay Sievers,
806 Krzysztof Nowicki, Lauri Tirkkonen, Lennart Poettering, Leonard König,
807 Long Li, Luca Boccassi, Lucas Werkmeister, Marcel Hoppe, Marc
808 Kleine-Budde, Mario Limonciello, Martin Jansa, Martin Wilck, Mathieu
809 Malaterre, Matteo F. Vescovi, Matthew McGinn, Matthias-Christian Ott,
810 Michael Biebl, Michael Olbrich, Michael Prokop, Michal Koutný, Michal
811 Sekletar, Mike Gilbert, Mikhail Kasimov, Milan Broz, Milan Pässler,
812 Mladen Pejaković, Muhammet Kara, Nicolas Boichat, Omer Katz, Paride
813 Legovini, Paul Menzel, Paul Milliken, Pavel Hrdina, Peter A. Bigot,
814 Peter D'Hoye, Peter Hutterer, Peter Jones, Philip Sequeira, Philip
815 Withnall, Piotr Drąg, Radostin Stoyanov, Ricardo Salveti de Araujo,
816 Ronny Chevalier, Rosen Penev, Rubén Suárez Alvarez, Ryan Gonzalez,
817 Salvo Tomaselli, Sebastian Reichel, Sergey Ptashnick, Sergio Lindo
818 Mansilla, Stefan Schweter, Stephen Hemminger, Stuart Hayes, Susant
819 Sahani, Sylvain Plantefève, Thomas H. P. Andersen, Tobias Jungel,
820 Tomasz Torcz, Vito Caputo, Will Dietz, Will Thompson, Wim van Mourik,
821 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
827 * The MemoryAccounting= unit property now defaults to on. After
828 discussions with the upstream control group maintainers we learnt
829 that the negative impact of cgroup memory accounting on current
830 kernels is finally relatively minimal, so that it should be safe to
831 enable this by default without affecting system performance. Besides
832 memory accounting only task accounting is turned on by default, all
833 other forms of resource accounting (CPU, IO, IP) remain off for now,
834 because it's not clear yet that their impact is small enough to move
835 from opt-in to opt-out. We recommend downstreams to leave memory
836 accounting on by default if kernel 4.14 or higher is primarily
837 used. On very resource constrained systems or when support for old
838 kernels is a necessity, -Dmemory-accounting-default=false can be used
839 to revert this change.
841 * rpm scriptlets to update the udev hwdb and rules (%udev_hwdb_update,
842 %udev_rules_update) and the journal catalog (%journal_catalog_update)
843 from the upgrade scriptlets of individual packages now do nothing.
844 Transfiletriggers have been added which will perform those updates
845 once at the end of the transaction.
847 Similar transfiletriggers have been added to execute any sysctl.d
848 and binfmt.d rules. Thus, it should be unnecessary to provide any
849 scriptlets to execute this configuration from package installation
852 * systemd-sysusers gained a mode where the configuration to execute is
853 specified on the command line, but this configuration is not executed
854 directly, but instead it is merged with the configuration on disk,
855 and the result is executed. This is useful for package installation
856 scripts which want to create the user before installing any files on
857 disk (in case some of those files are owned by that user), while
858 still allowing local admin overrides.
860 This functionality is exposed to rpm scriptlets through a new
861 %sysusers_create_package macro. Old %sysusers_create and
862 %sysusers_create_inline macros are deprecated.
864 A transfiletrigger for sysusers.d configuration is now installed,
865 which means that it should be unnecessary to call systemd-sysusers from
866 package installation scripts, unless the package installs any files
867 owned by those newly-created users, in which case
868 %sysusers_create_package should be used.
870 * Analogous change has been done for systemd-tmpfiles: it gained a mode
871 where the command-line configuration is merged with the configuration
872 on disk. This is exposed as the new %tmpfiles_create_package macro,
873 and %tmpfiles_create is deprecated. A transfiletrigger is installed
874 for tmpfiles.d, hence it should be unnecessary to call systemd-tmpfiles
875 from package installation scripts.
877 * sysusers.d configuration for a user may now also specify the group
878 number, in addition to the user number ("u username 123:456"), or
879 without the user number ("u username -:456").
881 * Configution items for systemd-sysusers can now be specified as
882 positional arguments when the new --inline switch is used.
884 * The login shell of users created through sysusers.d may now be
885 specified (previously, it was always /bin/sh for root and
886 /sbin/nologin for other users).
888 * systemd-analyze gained a new --global switch to look at global user
889 configuration. It also gained a unit-paths verb to list the unit load
890 paths that are compiled into systemd (which can be used with
891 --systemd, --user, or --global).
893 * udevadm trigger gained a new --settle/-w option to wait for any
894 triggered events to finish (but just those, and not any other events
895 which are triggered meanwhile).
897 * The action that systemd-logind takes when the lid is closed and the
898 machine is connected to external power can now be configured using
899 HandleLidSwitchExternalPower= in logind.conf. Previously, this action
900 was determined by HandleLidSwitch=, and, for backwards compatibility,
901 is still is, if HandleLidSwitchExternalPower= is not explicitly set.
903 * journalctl will periodically call sd_journal_process() to make it
904 resilient against inotify queue overruns when journal files are
905 rotated very quickly.
907 * Two new functions in libsystemd — sd_bus_get_n_queued_read and
908 sd_bus_get_n_queued_write — may be used to check the number of
909 pending bus messages.
911 * systemd gained a new
912 org.freedesktop.systemd1.Manager.AttachProcessesToUnit dbus call
913 which can be used to migrate foreign processes to scope and service
914 units. The primary user for this new API is systemd itself: the
915 systemd --user instance uses this call of the systemd --system
916 instance to migrate processes if it itself gets the request to
917 migrate processes and the kernel refuses this due to access
918 restrictions. Thanks to this "systemd-run --scope --user …" works
919 again in pure cgroups v2 environments when invoked from the user
922 * A new TemporaryFileSystem= setting can be used to mask out part of
923 the real file system tree with tmpfs mounts. This may be combined
924 with BindPaths= and BindReadOnlyPaths= to hide files or directories
925 not relevant to the unit, while still allowing some paths lower in
926 the tree to be accessed.
928 ProtectHome=tmpfs may now be used to hide user home and runtime
929 directories from units, in a way that is mostly equivalent to
930 "TemporaryFileSystem=/home /run/user /root".
932 * Non-service units are now started with KeyringMode=shared by default.
933 This means that mount and swapon and other mount tools have access
934 to keys in the main keyring.
936 * /sys/fs/bpf is now mounted automatically.
938 * QNX virtualization is now detected by systemd-detect-virt and may
939 be used in ConditionVirtualization=.
941 * IPAccounting= may now be enabled also for slice units.
943 * A new -Dsplit-bin= build configuration switch may be used to specify
944 whether bin and sbin directories are merged, or if they should be
945 included separately in $PATH and various listings of executable
946 directories. The build configuration scripts will try to autodetect
947 the proper values of -Dsplit-usr= and -Dsplit-bin= based on build
948 system, but distributions are encouraged to configure this
951 * A new -Dok-color= build configuration switch may be used to change
952 the colour of "OK" status messages.
954 * UPGRADE ISSUE: serialization of units using JoinsNamespaceOf= with
955 PrivateNetwork=yes was buggy in previous versions of systemd. This
956 means that after the upgrade and daemon-reexec, any such units must
959 * INCOMPATIBILITY: as announced in the NEWS for 237, systemd-tmpfiles
960 will not exclude read-only files owned by root from cleanup.
962 Contributions from: Alan Jenkins, Alexander F Rødseth, Alexis Jeandet,
963 Andika Triwidada, Andrei Gherzan, Ansgar Burchardt, antizealot1337,
964 Batuhan Osman Taşkaya, Beniamino Galvani, Bill Yodlowsky, Caio Marcelo
965 de Oliveira Filho, CuBiC, Daniele Medri, Daniel Mouritzen, Daniel
966 Rusek, Davide Cavalca, Dimitri John Ledkov, Douglas Christman, Evgeny
967 Vereshchagin, Faalagorn, Filipe Brandenburger, Franck Bui, futpib,
968 Giacomo Longo, Gunnar Hjalmarsson, Hans de Goede, Hermann Gausterer,
969 Iago López Galeiras, Jakub Filak, Jan Synacek, Jason A. Donenfeld,
970 Javier Martinez Canillas, Jérémy Rosen, Lennart Poettering, Lucas
971 Werkmeister, Mao Huang, Marco Gulino, Michael Biebl, Michael Vogt,
972 MilhouseVH, Neal Gompa (ニール・ゴンパ), Oleander Reis, Olof Mogren,
973 Patrick Uiterwijk, Peter Hutterer, Peter Portante, Piotr Drąg, Robert
974 Antoni Buj Gelonch, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
975 Fowler, SjonHortensius, snorreflorre, Susant Sahani, Sylvain
976 Plantefève, Thomas Blume, Thomas Haller, Vito Caputo, Yu Watanabe,
977 Zbigniew Jędrzejewski-Szmek, Марко М. Костић (Marko M. Kostić)
983 * Some keyboards come with a zoom see-saw or rocker which until now got
984 mapped to the Linux "zoomin/out" keys in hwdb. However, these
985 keycodes are not recognized by any major desktop. They now produce
986 Up/Down key events so that they can be used for scrolling.
988 * INCOMPATIBILITY: systemd-tmpfiles' "f" lines changed behaviour
989 slightly: previously, if an argument was specified for lines of this
990 type (i.e. the right-most column was set) this string was appended to
991 existing files each time systemd-tmpfiles was run. This behaviour was
992 different from what the documentation said, and not particularly
993 useful, as repeated systemd-tmpfiles invocations would not be
994 idempotent and grow such files without bounds. With this release
995 behaviour has been altered to match what the documentation says:
996 lines of this type only have an effect if the indicated files don't
997 exist yet, and only then the argument string is written to the file.
999 * FUTURE INCOMPATIBILITY: In systemd v238 we intend to slightly change
1000 systemd-tmpfiles behaviour: previously, read-only files owned by root
1001 were always excluded from the file "aging" algorithm (i.e. the
1002 automatic clean-up of directories like /tmp based on
1003 atime/mtime/ctime). We intend to drop this restriction, and age files
1004 by default even when owned by root and read-only. This behaviour was
1005 inherited from older tools, but there have been requests to remove
1006 it, and it's not obvious why this restriction was made in the first
1007 place. Please speak up now, if you are aware of software that reqires
1008 this behaviour, otherwise we'll remove the restriction in v238.
1010 * A new environment variable $SYSTEMD_OFFLINE is now understood by
1011 systemctl. It takes a boolean argument. If on, systemctl assumes it
1012 operates on an "offline" OS tree, and will not attempt to talk to the
1013 service manager. Previously, this mode was implicitly enabled if a
1014 chroot() environment was detected, and this new environment variable
1015 now provides explicit control.
1017 * .path and .socket units may now be created transiently, too.
1018 Previously only service, mount, automount and timer units were
1019 supported as transient units. The systemd-run tool has been updated
1020 to expose this new functionality, you may hence use it now to bind
1021 arbitrary commands to path or socket activation on-the-fly from the
1022 command line. Moreover, almost all properties are now exposed for the
1023 unit types that already supported transient operation.
1025 * The systemd-mount command gained support for a new --owner= parameter
1026 which takes a user name, which is then resolved and included in uid=
1027 and gid= mount options string of the file system to mount.
1029 * A new unit condition ConditionControlGroupController= has been added
1030 that checks whether a specific cgroup controller is available.
1032 * Unit files, udev's .link files, and systemd-networkd's .netdev and
1033 .network files all gained support for a new condition
1034 ConditionKernelVersion= for checking against specific kernel
1037 * In systemd-networkd, the [IPVLAN] section in .netdev files gained
1038 support for configuring device flags in the Flags= setting. In the
1039 same files, the [Tunnel] section gained support for configuring
1040 AllowLocalRemote=. The [Route] section in .network files gained
1041 support for configuring InitialCongestionWindow=,
1042 InitialAdvertisedReceiveWindow= and QuickAck=. The [DHCP] section now
1043 understands RapidCommit=.
1045 * systemd-networkd's DHCPv6 support gained support for Prefix
1048 * sd-bus gained support for a new "watch-bind" feature. When this
1049 feature is enabled, an sd_bus connection may be set up to connect to
1050 an AF_UNIX socket in the file system as soon as it is created. This
1051 functionality is useful for writing early-boot services that
1052 automatically connect to the system bus as soon as it is started,
1053 without ugly time-based polling. systemd-networkd and
1054 systemd-resolved have been updated to make use of this
1055 functionality. busctl exposes this functionality in a new
1056 --watch-bind= command line switch.
1058 * sd-bus will now optionally synthesize a local "Connected" signal as
1059 soon as a D-Bus connection is set up fully. This message mirrors the
1060 already existing "Disconnected" signal which is synthesized when the
1061 connection is terminated. This signal is generally useful but
1062 particularly handy in combination with the "watch-bind" feature
1063 described above. Synthesizing of this message has to be requested
1064 explicitly through the new API call sd_bus_set_connected_signal(). In
1065 addition a new call sd_bus_is_ready() has been added that checks
1066 whether a connection is fully set up (i.e. between the "Connected" and
1067 "Disconnected" signals).
1069 * sd-bus gained two new calls sd_bus_request_name_async() and
1070 sd_bus_release_name_async() for asynchronously registering bus
1071 names. Similar, there is now sd_bus_add_match_async() for installing
1072 a signal match asynchronously. All of systemd's own services have
1073 been updated to make use of these calls. Doing these operations
1074 asynchronously has two benefits: it reduces the risk of deadlocks in
1075 case of cyclic dependencies between bus services, and it speeds up
1076 service initialization since synchronization points for bus
1077 round-trips are removed.
1079 * sd-bus gained two new calls sd_bus_match_signal() and
1080 sd_bus_match_signal_async(), which are similar to sd_bus_add_match()
1081 and sd_bus_add_match_async() but instead of taking a D-Bus match
1082 string take match fields as normal function parameters.
1084 * sd-bus gained two new calls sd_bus_set_sender() and
1085 sd_bus_message_set_sender() for setting the sender name of outgoing
1086 messages (either for all outgoing messages or for just one specific
1087 one). These calls are only useful in direct connections as on
1088 brokered connections the broker fills in the sender anyway,
1089 overwriting whatever the client filled in.
1091 * sd-event gained a new pseudo-handle that may be specified on all API
1092 calls where an "sd_event*" object is expected: SD_EVENT_DEFAULT. When
1093 used this refers to the default event loop object of the calling
1094 thread. Note however that this does not implicitly allocate one —
1095 which has to be done prior by using sd_event_default(). Similarly
1096 sd-bus gained three new pseudo-handles SD_BUS_DEFAULT,
1097 SD_BUS_DEFAULT_USER, SD_BUS_DEFAULT_SYSTEM that may be used to refer
1098 to the default bus of the specified type of the calling thread. Here
1099 too this does not implicitly allocate bus connection objects, this
1100 has to be done prior with sd_bus_default() and friends.
1102 * sd-event gained a new call pair
1103 sd_event_source_{get|set}_io_fd_own(). This may be used to request
1104 automatic closure of the file descriptor an IO event source watches
1105 when the event source is destroyed.
1107 * systemd-networkd gained support for natively configuring WireGuard
1110 * In previous versions systemd synthesized user records both for the
1111 "nobody" (UID 65534) and "root" (UID 0) users in nss-systemd and
1112 internally. In order to simplify distribution-wide renames of the
1113 "nobody" user (like it is planned in Fedora: nfsnobody → nobody), a
1114 new transitional flag file has been added: if
1115 /etc/systemd/dont-synthesize-nobody exists synthesizing of the 65534
1116 user and group record within the systemd codebase is disabled.
1118 * systemd-notify gained a new --uid= option for selecting the source
1119 user/UID to use for notification messages sent to the service
1122 * journalctl gained a new --grep= option to list only entries in which
1123 the message matches a certain pattern. By default matching is case
1124 insensitive if the pattern is lowercase, and case sensitive
1125 otherwise. Option --case-sensitive=yes|no can be used to override
1126 this an specify case sensitivity or case insensitivity.
1128 * There's now a "systemd-analyze service-watchdogs" command for printing
1129 the current state of the service runtime watchdog, and optionally
1130 enabling or disabling the per-service watchdogs system-wide if given a
1131 boolean argument (i.e. the concept you configure in WatchdogSec=), for
1132 debugging purposes. There's also a kernel command line option
1133 systemd.service_watchdogs= for controlling the same.
1135 * Two new "log-level" and "log-target" options for systemd-analyze were
1136 added that merge the now deprecated get-log-level, set-log-level and
1137 get-log-target, set-log-target pairs. The deprecated options are still
1138 understood for backwards compatibility. The two new options print the
1139 current value when no arguments are given, and set them when a
1140 level/target is given as an argument.
1142 * sysusers.d's "u" lines now optionally accept both a UID and a GID
1143 specification, separated by a ":" character, in order to create users
1144 where UID and GID do not match.
1146 Contributions from: Adam Duskett, Alan Jenkins, Alexander Kuleshov,
1147 Alexis Deruelle, Andrew Jeddeloh, Armin Widegreen, Batuhan Osman
1148 Taşkaya, Björn Esser, bleep_blop, Bruce A. Johnson, Chris Down, Clinton
1149 Roy, Colin Walters, Daniel Rusek, Dimitri John Ledkov, Dmitry Rozhkov,
1150 Evgeny Vereshchagin, Ewout van Mansom, Felipe Sateler, Franck Bui,
1151 Frantisek Sumsal, George Gaydarov, Gianluca Boiano, Hans-Christian
1152 Noren Egtvedt, Hans de Goede, Henrik Grindal Bakken, Jan Alexander
1153 Steffens, Jan Klötzke, Jason A. Donenfeld, jdkbx, Jérémy Rosen,
1154 Jerónimo Borque, John Lin, John Paul Herold, Jonathan Rudenberg, Jörg
1155 Thalheim, Ken (Bitsko) MacLeod, Larry Bernstone, Lennart Poettering,
1156 Lucas Werkmeister, Maciej S. Szmigiero, Marek Čermák, Martin Pitt,
1157 Mathieu Malaterre, Matthew Thode, Matthias-Christian Ott, Max Harmathy,
1158 Michael Biebl, Michael Vogt, Michal Koutný, Michal Sekletar, Michał
1159 Szczepański, Mike Gilbert, Nathaniel McCallum, Nicolas Chauvet, Olaf
1160 Hering, Olivier Schwander, Patrik Flykt, Paul Cercueil, Peter Hutterer,
1161 Piotr Drąg, Raphael Vogelgsang, Reverend Homer, Robert Kolchmeyer,
1162 Samuel Dionne-Riel, Sergey Ptashnick, Shawn Landden, Susant Sahani,
1163 Sylvain Plantefève, Thomas H. P. Andersen, Thomas Huth, Tomasz
1164 Bachorski, Vladislav Vishnyakov, Wieland Hoffmann, Yu Watanabe, Zachary
1165 Winnerman, Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски, Дилян
1172 * The modprobe.d/ drop-in for the bonding.ko kernel module introduced
1173 in v235 has been extended to also set the dummy.ko module option
1174 numdummies=0, preventing the kernel from automatically creating
1175 dummy0. All dummy interfaces must now be explicitly created.
1177 * Unknown '%' specifiers in configuration files are now rejected. This
1178 applies to units and tmpfiles.d configuration. Any percent characters
1179 that are followed by a letter or digit that are not supposed to be
1180 interpreted as the beginning of a specifier should be escaped by
1181 doubling ("%%"). (So "size=5%" is still accepted, as well as
1182 "size=5%,foo=bar", but not "LABEL=x%y%z" since %y and %z are not
1183 valid specifiers today.)
1185 * systemd-resolved now maintains a new dynamic
1186 /run/systemd/resolve/stub-resolv.conf compatibility file. It is
1187 recommended to make /etc/resolv.conf a symlink to it. This file
1188 points at the systemd-resolved stub DNS 127.0.0.53 resolver and
1189 includes dynamically acquired search domains, achieving more correct
1190 DNS resolution by software that bypasses local DNS APIs such as NSS.
1192 * The "uaccess" udev tag has been dropped from /dev/kvm and
1193 /dev/dri/renderD*. These devices now have the 0666 permissions by
1194 default (but this may be changed at build-time). /dev/dri/renderD*
1195 will now be owned by the "render" group along with /dev/kfd.
1197 * "DynamicUser=yes" has been enabled for systemd-timesyncd.service,
1198 systemd-journal-gatewayd.service and
1199 systemd-journal-upload.service. This means "nss-systemd" must be
1200 enabled in /etc/nsswitch.conf to ensure the UIDs assigned to these
1201 services are resolved properly.
1203 * In /etc/fstab two new mount options are now understood:
1204 x-systemd.makefs and x-systemd.growfs. The former has the effect that
1205 the configured file system is formatted before it is mounted, the
1206 latter that the file system is resized to the full block device size
1207 after it is mounted (i.e. if the file system is smaller than the
1208 partition it resides on, it's grown). This is similar to the fsck
1209 logic in /etc/fstab, and pulls in systemd-makefs@.service and
1210 systemd-growfs@.service as necessary, similar to
1211 systemd-fsck@.service. Resizing is currently only supported on ext4
1214 * In systemd-networkd, the IPv6 RA logic now optionally may announce
1215 DNS server and domain information.
1217 * Support for the LUKS2 on-disk format for encrypted partitions has
1218 been added. This requires libcryptsetup2 during compilation and
1221 * The systemd --user instance will now signal "readiness" when its
1222 basic.target unit has been reached, instead of when the run queue ran
1223 empty for the first time.
1225 * Tmpfiles.d with user configuration are now also supported.
1226 systemd-tmpfiles gained a new --user switch, and snippets placed in
1227 ~/.config/user-tmpfiles.d/ and corresponding directories will be
1228 executed by systemd-tmpfiles --user running in the new
1229 systemd-tmpfiles-setup.service and systemd-tmpfiles-clean.service
1230 running in the user session.
1232 * Unit files and tmpfiles.d snippets learnt three new % specifiers:
1233 %S resolves to the top-level state directory (/var/lib for the system
1234 instance, $XDG_CONFIG_HOME for the user instance), %C resolves to the
1235 top-level cache directory (/var/cache for the system instance,
1236 $XDG_CACHE_HOME for the user instance), %L resolves to the top-level
1237 logs directory (/var/log for the system instance,
1238 $XDG_CONFIG_HOME/log/ for the user instance). This matches the
1239 existing %t specifier, that resolves to the top-level runtime
1240 directory (/run for the system instance, and $XDG_RUNTIME_DIR for the
1243 * journalctl learnt a new parameter --output-fields= for limiting the
1244 set of journal fields to output in verbose and JSON output modes.
1246 * systemd-timesyncd's configuration file gained a new option
1247 RootDistanceMaxSec= for setting the maximum root distance of servers
1248 it'll use, as well as the new options PollIntervalMinSec= and
1249 PollIntervalMaxSec= to tweak the minimum and maximum poll interval.
1251 * bootctl gained a new command "list" for listing all available boot
1252 menu items on systems that follow the boot loader specification.
1254 * systemctl gained a new --dry-run switch that shows what would be done
1255 instead of doing it, and is currently supported by the shutdown and
1258 * ConditionSecurity= can now detect the TOMOYO security module.
1260 * Unit file [Install] sections are now also respected in unit drop-in
1261 files. This is intended to be used by drop-ins under /usr/lib/.
1263 * systemd-firstboot may now also set the initial keyboard mapping.
1265 * Udev "changed" events for devices which are exposed as systemd
1266 .device units are now propagated to units specified in
1267 ReloadPropagatedFrom= as reload requests.
1269 * If a udev device has a SYSTEMD_WANTS= property containing a systemd
1270 unit template name (i.e. a name in the form of 'foobar@.service',
1271 without the instance component between the '@' and - the '.'), then
1272 the escaped sysfs path of the device is automatically used as the
1275 * SystemCallFilter= in unit files has been extended so that an "errno"
1276 can be specified individually for each system call. Example:
1277 SystemCallFilter=~uname:EILSEQ.
1279 * The cgroup delegation logic has been substantially updated. Delegate=
1280 now optionally takes a list of controllers (instead of a boolean, as
1281 before), which lists the controllers to delegate at least.
1283 * The networkd DHCPv6 client now implements the FQDN option (RFC 4704).
1285 * A new LogLevelMax= setting configures the maximum log level any
1286 process of the service may log at (i.e. anything with a lesser
1287 priority than what is specified is automatically dropped). A new
1288 LogExtraFields= setting allows configuration of additional journal
1289 fields to attach to all log records generated by any of the unit's
1292 * New StandardInputData= and StandardInputText= settings along with the
1293 new option StandardInput=data may be used to configure textual or
1294 binary data that shall be passed to the executed service process via
1295 standard input, encoded in-line in the unit file.
1297 * StandardInput=, StandardOutput= and StandardError= may now be used to
1298 connect stdin/stdout/stderr of executed processes directly with a
1299 file or AF_UNIX socket in the file system, using the new "file:" option.
1301 * A new unit file option CollectMode= has been added, that allows
1302 tweaking the garbage collection logic for units. It may be used to
1303 tell systemd to garbage collect units that have failed automatically
1304 (normally it only GCs units that exited successfully). systemd-run
1305 and systemd-mount expose this new functionality with a new -G option.
1307 * "machinectl bind" may now be used to bind mount non-directories
1308 (i.e. regularfiles, devices, fifos, sockets).
1310 * systemd-analyze gained a new verb "calendar" for validating and
1311 testing calendar time specifications to use for OnCalendar= in timer
1312 units. Besides validating the expression it will calculate the next
1313 time the specified expression would elapse.
1315 * In addition to the pre-existing FailureAction= unit file setting
1316 there's now SuccessAction=, for configuring a shutdown action to
1317 execute when a unit completes successfully. This is useful in
1318 particular inside containers that shall terminate after some workload
1319 has been completed. Also, both options are now supported for all unit
1320 types, not just services.
1322 * networkds's IP rule support gained two new options
1323 IncomingInterface= and OutgoingInterface= for configuring the incoming
1324 and outgoing interfaces of configured rules. systemd-networkd also
1325 gained support for "vxcan" network devices.
1327 * networkd gained a new setting RequiredForOnline=, taking a
1328 boolean. If set, systemd-wait-online will take it into consideration
1329 when determining that the system is up, otherwise it will ignore the
1330 interface for this purpose.
1332 * The sd_notify() protocol gained support for a new operation: with
1333 FDSTOREREMOVE=1 file descriptors may be removed from the per-service
1334 store again, ahead of POLLHUP or POLLERR when they are removed
1337 * A new document doc/UIDS-GIDS.md has been added to the source tree,
1338 that documents the UID/GID range and assignment assumptions and
1339 requirements of systemd.
1341 * The watchdog device PID 1 will ping may now be configured through the
1342 WatchdogDevice= configuration file setting, or by setting the
1343 systemd.watchdog_service= kernel commandline option.
1345 * systemd-resolved's gained support for registering DNS-SD services on
1346 the local network using MulticastDNS. Services may either be
1347 registered by dropping in a .dnssd file in /etc/systemd/dnssd/ (or
1348 the same dir below /run, /usr/lib), or through its D-Bus API.
1350 * The sd_notify() protocol can now with EXTEND_TIMEOUT_USEC=microsecond
1351 extend the effective start, runtime, and stop time. The service must
1352 continue to send EXTEND_TIMEOUT_USEC within the period specified to
1353 prevent the service manager from making the service as timedout.
1355 * systemd-resolved's DNSSEC support gained support for RFC 8080
1356 (Ed25519 keys and signatures).
1358 * The systemd-resolve command line tool gained a new set of options
1359 --set-dns=, --set-domain=, --set-llmnr=, --set-mdns=, --set-dnssec=,
1360 --set-nta= and --revert to configure per-interface DNS configuration
1361 dynamically during runtime. It's useful for pushing DNS information
1362 into systemd-resolved from DNS hook scripts that various interface
1363 managing software supports (such as pppd).
1365 * systemd-nspawn gained a new --network-namespace-path= command line
1366 option, which may be used to make a container join an existing
1367 network namespace, by specifying a path to a "netns" file.
1369 Contributions from: Alan Jenkins, Alan Robertson, Alessandro Ghedini,
1370 Andrew Jeddeloh, Antonio Rojas, Ari, asavah, bleep_blop, Carsten
1371 Strotmann, Christian Brauner, Christian Hesse, Clinton Roy, Collin
1372 Eggert, Cong Wang, Daniel Black, Daniel Lockyer, Daniel Rusek, Dimitri
1373 John Ledkov, Dmitry Rozhkov, Dongsu Park, Edward A. James, Evgeny
1374 Vereshchagin, Florian Klink, Franck Bui, Gwendal Grignou, Hans de
1375 Goede, Harald Hoyer, Hristo Venev, Iago López Galeiras, Ikey Doherty,
1376 Jakub Wilk, Jérémy Rosen, Jiahui Xie, John Lin, José Bollo, Josef
1377 Andersson, juga0, Krzysztof Nowicki, Kyle Walker, Lars Karlitski, Lars
1378 Kellogg-Stedman, Lauri Tirkkonen, Lennart Poettering, Lubomir Rintel,
1379 Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn, Lukáš Říha, Lukasz
1380 Rubaszewski, Maciej S. Szmigiero, Mantas Mikulėnas, Marcus Folkesson,
1381 Martin Steuer, Mathieu Trudel-Lapierre, Matija Skala,
1382 Matthias-Christian Ott, Max Resch, Michael Biebl, Michael Vogt, Michal
1383 Koutný, Michal Sekletar, Mike Gilbert, Muhammet Kara, Neil Brown, Olaf
1384 Hering, Ondrej Kozina, Patrik Flykt, Patryk Kocielnik, Peter Hutterer,
1385 Piotr Drąg, Razvan Cojocaru, Robin McCorkell, Roland Hieber, Saran
1386 Tunyasuvunakool, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
1387 Arlott, Simon Peeters, Stanislav Angelovič, Stefan Agner, Susant
1388 Sahani, Sylvain Plantefève, Thomas Blume, Thomas Haller, Tiago Salem
1389 Herrmann, Tinu Weber, Tom Stellard, Topi Miettinen, Torsten Hilbrich,
1390 Vito Caputo, Vladislav Vishnyakov, WaLyong Cho, Yu Watanabe, Zbigniew
1391 Jędrzejewski-Szmek, Zeal Jagannatha
1393 — Berlin, 2017-12-14
1397 * INCOMPATIBILITY: systemd-logind.service and other long-running
1398 services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP
1399 communication with the outside. This generally improves security of
1400 the system, and is in almost all cases a safe and good choice, as
1401 these services do not and should not provide any network-facing
1402 functionality. However, systemd-logind uses the glibc NSS API to
1403 query the user database. This creates problems on systems where NSS
1404 is set up to directly consult network services for user database
1405 lookups. In particular, this creates incompatibilities with the
1406 "nss-nis" module, which attempts to directly contact the NIS/YP
1407 network servers it is configured for, and will now consistently
1408 fail. In such cases, it is possible to turn off IP sandboxing for
1409 systemd-logind.service (set IPAddressDeny= in its [Service] section
1410 to the empty string, via a .d/ unit file drop-in). Downstream
1411 distributions might want to update their nss-nis packaging to include
1412 such a drop-in snippet, accordingly, to hide this incompatibility
1413 from the user. Another option is to make use of glibc's nscd service
1414 to proxy such network requests through a privilege-separated, minimal
1415 local caching daemon, or to switch to more modern technologies such
1416 sssd, whose NSS hook-ups generally do not involve direct network
1417 access. In general, we think it's definitely time to question the
1418 implementation choices of nss-nis, i.e. whether it's a good idea
1419 today to embed a network-facing loadable module into all local
1420 processes that need to query the user database, including the most
1421 trivial and benign ones, such as "ls". For more details about
1422 IPAddressDeny= see below.
1424 * A new modprobe.d drop-in is now shipped by default that sets the
1425 bonding module option max_bonds=0. This overrides the kernel default,
1426 to avoid conflicts and ambiguity as to whether or not bond0 should be
1427 managed by systemd-networkd or not. This resolves multiple issues
1428 with bond0 properties not being applied, when bond0 is configured
1429 with systemd-networkd. Distributors may choose to not package this,
1430 however in that case users will be prevented from correctly managing
1431 bond0 interface using systemd-networkd.
1433 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
1434 which print the logging level and target of the system manager. They
1435 complement the existing "set-log-level" and "set-log-target" verbs
1436 used to change those values.
1438 * journald.conf gained a new boolean setting ReadKMsg= which defaults
1439 to on. If turned off kernel log messages will not be read by
1440 systemd-journald or included in the logs. It also gained a new
1441 setting LineMax= for configuring the maximum line length in
1442 STDOUT/STDERR log streams. The new default for this value is 48K, up
1443 from the previous hardcoded 2048.
1445 * A new unit setting RuntimeDirectoryPreserve= has been added, which
1446 allows more detailed control of what to do with a runtime directory
1447 configured with RuntimeDirectory= (i.e. a directory below /run or
1448 $XDG_RUNTIME_DIR) after a unit is stopped.
1450 * The RuntimeDirectory= setting for units gained support for creating
1451 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
1452 one top-level directory.
1454 * Units gained new options StateDirectory=, CacheDirectory=,
1455 LogsDirectory= and ConfigurationDirectory= which are closely related
1456 to RuntimeDirectory= but manage per-service directories below
1457 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
1458 possible to write unit files which when activated automatically gain
1459 properly owned service specific directories in these locations, thus
1460 making unit files self-contained and increasing compatibility with
1461 stateless systems and factory reset where /etc or /var are
1462 unpopulated at boot. Matching these new settings there's also
1463 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
1464 ConfigurationDirectoryMode= for configuring the access mode of these
1465 directories. These settings are particularly useful in combination
1466 with DynamicUser=yes as they provide secure, properly-owned,
1467 writable, and stateful locations for storage, excluded from the
1468 sandbox that such services live in otherwise.
1470 * Automake support has been removed from this release. systemd is now
1473 * systemd-journald will now aggressively cache client metadata during
1474 runtime, speeding up log write performance under pressure. This comes
1475 at a small price though: as much of the metadata is read
1476 asynchronously from /proc/ (and isn't implicitly attached to log
1477 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
1478 metadata stored alongside a log entry might be slightly
1479 out-of-date. Previously it could only be slightly newer than the log
1480 message. The time window is small however, and given that the kernel
1481 is unlikely to be improved anytime soon in this regard, this appears
1484 * nss-myhostname/systemd-resolved will now by default synthesize an
1485 A/AAAA resource record for the "_gateway" hostname, pointing to the
1486 current default IP gateway. Previously it did that for the "gateway"
1487 name, hampering adoption, as some distributions wanted to leave that
1488 host name open for local use. The old behaviour may still be
1489 requested at build time.
1491 * systemd-networkd's [Address] section in .network files gained a new
1492 Scope= setting for configuring the IP address scope. The [Network]
1493 section gained a new boolean setting ConfigureWithoutCarrier= that
1494 tells systemd-networkd to ignore link sensing when configuring the
1495 device. The [DHCP] section gained a new Anonymize= boolean option for
1496 turning on a number of options suggested in RFC 7844. A new
1497 [RoutingPolicyRule] section has been added for configuring the IP
1498 routing policy. The [Route] section has gained support for a new
1499 Type= setting which permits configuring
1500 blackhole/unreachable/prohibit routes.
1502 * The [VRF] section in .netdev files gained a new Table= setting for
1503 configuring the routing table to use. The [Tunnel] section gained a
1504 new Independent= boolean field for configuring tunnels independent of
1505 an underlying network interface. The [Bridge] section gained a new
1506 GroupForwardMask= option for configuration of propagation of link
1507 local frames between bridge ports.
1509 * The WakeOnLan= setting in .link files gained support for a number of
1510 new modes. A new TCP6SegmentationOffload= setting has been added for
1511 configuring TCP/IPv6 hardware segmentation offload.
1513 * The IPv6 RA sender implementation may now optionally send out RDNSS
1514 and RDNSSL records to supply DNS configuration to peers.
1516 * systemd-nspawn gained support for a new --system-call-filter= command
1517 line option for adding and removing entries in the default system
1518 call filter it applies. Moreover systemd-nspawn has been changed to
1519 implement a system call whitelist instead of a blacklist.
1521 * systemd-run gained support for a new --pipe command line option. If
1522 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
1523 are directly passed on to the activated transient service
1524 executable. This allows invoking arbitrary processes as systemd
1525 services (for example to take benefit of dependency management,
1526 accounting management, resource management or log management that is
1527 done automatically for services) — while still allowing them to be
1528 integrated in a classic UNIX shell pipeline.
1530 * When a service sends RELOAD=1 via sd_notify() and reload propagation
1531 using ReloadPropagationTo= is configured, a reload is now propagated
1532 to configured units. (Previously this was only done on explicitly
1533 requested reloads, using "systemctl reload" or an equivalent
1536 * For each service unit a restart counter is now kept: it is increased
1537 each time the service is restarted due to Restart=, and may be
1538 queried using "systemctl show -p NRestarts …".
1540 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
1541 @signal and @timer have been added, for usage with SystemCallFilter=
1542 in unit files and the new --system-call-filter= command line option
1543 of systemd-nspawn (see above).
1545 * ExecStart= lines in unit files gained two new modifiers: when a
1546 command line is prefixed with "!" the command will be executed as
1547 configured, except for the credentials applied by
1548 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
1549 "+", but does still apply namespacing options unlike "+". There's
1550 also "!!" now, which is mostly identical, but becomes a NOP on
1551 systems that support ambient capabilities. This is useful to write
1552 unit files that work with ambient capabilities where possible but
1553 automatically fall back to traditional privilege dropping mechanisms
1554 on systems where this is not supported.
1556 * ListenNetlink= settings in socket units now support RDMA netlink
1559 * A new unit file setting LockPersonality= has been added which permits
1560 locking down the chosen execution domain ("personality") of a service
1563 * A new special target "getty-pre.target" has been added, which is
1564 ordered before all text logins, and may be used to order services
1565 before textual logins acquire access to the console.
1567 * systemd will now attempt to load the virtio-rng.ko kernel module very
1568 early on if a VM environment supporting this is detected. This should
1569 improve entropy during early boot in virtualized environments.
1571 * A _netdev option is now supported in /etc/crypttab that operates in a
1572 similar way as the same option in /etc/fstab: it permits configuring
1573 encrypted devices that need to be ordered after the network is up.
1574 Following this logic, two new special targets
1575 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
1576 added that are to cryptsetup.target what remote-fs.target and
1577 remote-fs-pre.target are to local-fs.target.
1579 * Service units gained a new UnsetEnvironment= setting which permits
1580 unsetting specific environment variables for services that are
1581 normally passed to it (for example in order to mask out locale
1582 settings for specific services that can't deal with it).
1584 * Units acquired a new boolean option IPAccounting=. When turned on, IP
1585 traffic accounting (packet count as well as byte count) is done for
1586 the service, and shown as part of "systemctl status" or "systemd-run
1589 * Service units acquired two new options IPAddressAllow= and
1590 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
1591 for configuring a simple IP access control list for all sockets of
1592 the unit. These options are available also on .slice and .socket
1593 units, permitting flexible access list configuration for individual
1594 services as well as groups of services (as defined by a slice unit),
1595 including system-wide. Note that IP ACLs configured this way are
1596 enforced on every single IPv4 and IPv6 socket created by any process
1597 of the service unit, and apply to ingress as well as egress traffic.
1599 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
1600 structured log message is generated each time the unit is stopped,
1601 containing information about the consumed resources of this
1604 * A new setting KeyringMode= has been added to unit files, which may be
1605 used to control how the kernel keyring is set up for executed
1608 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
1609 "systemctl kexec" and "systemctl exit" are now always asynchronous in
1610 behaviour (that is: these commands return immediately after the
1611 operation was enqueued instead of waiting for the operation to
1612 complete). Previously, "systemctl poweroff" and "systemctl reboot"
1613 were asynchronous on systems using systemd-logind (i.e. almost
1614 always, and like they were on sysvinit), and the other three commands
1615 were unconditionally synchronous. With this release this is cleaned
1616 up, and callers will see the same asynchronous behaviour on all
1617 systems for all five operations.
1619 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
1622 * .timer units now accept calendar specifications in other timezones
1623 than UTC or the local timezone.
1625 * The tmpfiles snippet var.conf has been changed to create
1626 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
1627 the "utmp" group already, and it appears to be generally understood
1628 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
1629 databases. Previously this was implemented correctly for all these
1630 databases excepts btmp, which has been opened up like this now
1631 too. Note that while the other databases are world-readable
1632 (i.e. 0644), btmp is not and remains more restrictive.
1634 * The systemd-resolve tool gained a new --reset-server-features
1635 switch. When invoked like this systemd-resolved will forget
1636 everything it learnt about the features supported by the configured
1637 upstream DNS servers, and restarts the feature probing logic on the
1638 next resolver look-up for them at the highest feature level
1641 * The status dump systemd-resolved sends to the logs upon receiving
1642 SIGUSR1 now also includes information about all DNS servers it is
1643 configured to use, and the features levels it probed for them.
1645 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
1646 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
1647 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
1648 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
1649 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
1650 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
1651 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
1652 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
1653 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
1654 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
1655 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
1656 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
1657 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
1658 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
1659 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
1660 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
1661 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
1662 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
1663 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
1664 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
1666 — Berlin, 2017-10-06
1670 * Meson is now supported as build system in addition to Automake. It is
1671 our plan to remove Automake in one of our next releases, so that
1672 Meson becomes our exclusive build system. Hence, please start using
1673 the Meson build system in your downstream packaging. There's plenty
1674 of documentation around how to use Meson, the extremely brief
1677 ./autogen.sh && ./configure && make && sudo make install
1681 meson build && ninja -C build && sudo ninja -C build install
1683 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
1684 which permits configuring a timeout on the time a job is
1685 running. This is particularly useful for setting timeouts on jobs for
1688 * Unit files gained two new options ConditionUser= and ConditionGroup=
1689 for conditionalizing units based on the identity of the user/group
1690 running a systemd user instance.
1692 * systemd-networkd now understands a new FlowLabel= setting in the
1693 [VXLAN] section of .network files, as well as a Priority= in
1694 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
1695 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
1696 gained support for configuration of GENEVE links, and IPv6 address
1697 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
1699 * .link files now understand a new Port= setting.
1701 * systemd-networkd's DHCP support gained support for DHCP option 119
1702 (domain search list).
1704 * systemd-networkd gained support for serving IPv6 address ranges using
1705 the Router Advertisement protocol. The new .network configuration
1706 section [IPv6Prefix] may be used to configure the ranges to
1707 serve. This is implemented based on a new, minimal, native server
1708 implementation of RA.
1710 * journalctl's --output= switch gained support for a new parameter
1711 "short-iso-precise" for a mode where timestamps are shown as precise
1714 * systemd-udevd's "net_id" builtin may now generate stable network
1715 interface names from IBM PowerVM VIO devices as well as ACPI platform
1718 * MulticastDNS support in systemd-resolved may now be explicitly
1719 enabled/disabled using the new MulticastDNS= configuration file
1722 * systemd-resolved may now optionally use libidn2 instead of the libidn
1723 for processing internationalized domain names. Support for libidn2
1724 should be considered experimental and should not be enabled by
1727 * "machinectl pull-tar" and related call may now do verification of
1728 downloaded images using SUSE-style .sha256 checksum files in addition
1729 to the already existing support for validating using Ubuntu-style
1732 * sd-bus gained support for a new sd_bus_message_appendv() call which
1733 is va_list equivalent of sd_bus_message_append().
1735 * sd-boot gained support for validating images using SHIM/MOK.
1737 * The SMACK code learnt support for "onlycap".
1739 * systemd-mount --umount is now much smarter in figuring out how to
1740 properly unmount a device given its mount or device path.
1742 * The code to call libnss_dns as a fallback from libnss_resolve when
1743 the communication with systemd-resolved fails was removed. This
1744 fallback was redundant and interfered with the [!UNAVAIL=return]
1745 suffix. See nss-resolve(8) for the recommended configuration.
1747 * systemd-logind may now be restarted without losing state. It stores
1748 the file descriptors for devices it manages in the system manager
1749 using the FDSTORE= mechanism. Please note that further changes in
1750 other components may be required to make use of this (for example
1751 Xorg has code to listen for stops of systemd-logind and terminate
1752 itself when logind is stopped or restarted, in order to avoid using
1753 stale file descriptors for graphical devices, which is now
1754 counterproductive and must be reverted in order for restarts of
1755 systemd-logind to be safe. See
1756 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
1758 * All kernel install plugins are called with the environment variable
1759 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
1760 /etc/machine-id. If the file is missing or empty, the variable is
1761 empty and BOOT_DIR_ABS is the path of a temporary directory which is
1762 removed after all the plugins exit. So, if KERNEL_INSTALL_MACHINE_ID
1763 is empty, all plugins should not put anything in BOOT_DIR_ABS.
1765 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
1766 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
1767 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
1768 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
1769 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
1770 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
1771 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
1772 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
1773 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
1774 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
1775 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
1776 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
1777 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
1778 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
1779 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
1780 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
1781 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
1782 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
1783 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
1784 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
1785 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
1786 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
1787 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
1788 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
1789 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
1790 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
1791 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
1792 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
1793 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
1796 — Berlin, 2017-07-12
1800 * The "hybrid" control group mode has been modified to improve
1801 compatibility with "legacy" cgroups-v1 setups. Specifically, the
1802 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
1803 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
1804 cgroups-v1 hierarchy), the only externally visible change being that
1805 the cgroups-v2 hierarchy is also mounted, to
1806 /sys/fs/cgroup/unified. This should provide a large degree of
1807 compatibility with "legacy" cgroups-v1, while taking benefit of the
1808 better management capabilities of cgroups-v2.
1810 * The default control group setup mode may be selected both a boot-time
1811 via a set of kernel command line parameters (specifically:
1812 systemd.unified_cgroup_hierarchy= and
1813 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
1814 default selected on the configure command line
1815 (--with-default-hierarchy=). The upstream default is "hybrid"
1816 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
1817 this will change in a future systemd version to be "unified" (pure
1818 cgroups-v2 mode). The third option for the compile time option is
1819 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
1820 distributions to default to "hybrid" mode for release distributions,
1821 starting with v233. We recommend "unified" for development
1822 distributions (specifically: distributions such as Fedora's rawhide)
1823 as that's where things are headed in the long run. Use "legacy" for
1824 greatest stability and compatibility only.
1826 * Note one current limitation of "unified" and "hybrid" control group
1827 setup modes: the kernel currently does not permit the systemd --user
1828 instance (i.e. unprivileged code) to migrate processes between two
1829 disconnected cgroup subtrees, even if both are managed and owned by
1830 the user. This effectively means "systemd-run --user --scope" doesn't
1831 work when invoked from outside of any "systemd --user" service or
1832 scope. Specifically, it is not supported from session scopes. We are
1833 working on fixing this in a future systemd version. (See #3388 for
1834 further details about this.)
1836 * DBus policy files are now installed into /usr rather than /etc. Make
1837 sure your system has dbus >= 1.9.18 running before upgrading to this
1838 version, or override the install path with --with-dbuspolicydir= .
1840 * All python scripts shipped with systemd (specifically: the various
1841 tests written in Python) now require Python 3.
1843 * systemd unit tests can now run standalone (without the source or
1844 build directories), and can be installed into /usr/lib/systemd/tests/
1845 with 'make install-tests'.
1847 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
1848 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
1851 * Support for the %c, %r, %R specifiers in unit files has been
1852 removed. Specifiers are not supposed to be dependent on configuration
1853 in the unit file itself (so that they resolve the same regardless
1854 where used in the unit files), but these specifiers were influenced
1855 by the Slice= option.
1857 * The shell invoked by debug-shell.service now defaults to /bin/sh in
1858 all cases. If distributions want to use a different shell for this
1859 purpose (for example Fedora's /sbin/sushell) they need to specify
1860 this explicitly at configure time using --with-debug-shell=.
1862 * The confirmation spawn prompt has been reworked to offer the
1865 (c)ontinue, proceed without asking anymore
1866 (D)ump, show the state of the unit
1867 (f)ail, don't execute the command and pretend it failed
1869 (i)nfo, show a short summary of the unit
1870 (j)obs, show jobs that are in progress
1871 (s)kip, don't execute the command and pretend it succeeded
1872 (y)es, execute the command
1874 The 'n' choice for the confirmation spawn prompt has been removed,
1875 because its meaning was confusing.
1877 The prompt may now also be redirected to an alternative console by
1878 specifying the console as parameter to systemd.confirm_spawn=.
1880 * Services of Type=notify require a READY=1 notification to be sent
1881 during startup. If no such message is sent, the service now fails,
1882 even if the main process exited with a successful exit code.
1884 * Services that fail to start up correctly now always have their
1885 ExecStopPost= commands executed. Previously, they'd enter "failed"
1886 state directly, without executing these commands.
1888 * The option MulticastDNS= of network configuration files has acquired
1889 an actual implementation. With MulticastDNS=yes a host can resolve
1890 names of remote hosts and reply to mDNS A and AAAA requests.
1892 * When units are about to be started an additional check is now done to
1893 ensure that all dependencies of type BindsTo= (when used in
1894 combination with After=) have been started.
1896 * systemd-analyze gained a new verb "syscall-filter" which shows which
1897 system call groups are defined for the SystemCallFilter= unit file
1898 setting, and which system calls they contain.
1900 * A new system call filter group "@filesystem" has been added,
1901 consisting of various file system related system calls. Group
1902 "@reboot" has been added, covering reboot, kexec and shutdown related
1903 calls. Finally, group "@swap" has been added covering swap
1904 configuration related calls.
1906 * A new unit file option RestrictNamespaces= has been added that may be
1907 used to restrict access to the various process namespace types the
1908 Linux kernel provides. Specifically, it may be used to take away the
1909 right for a service unit to create additional file system, network,
1910 user, and other namespaces. This sandboxing option is particularly
1911 relevant due to the high amount of recently discovered namespacing
1912 related vulnerabilities in the kernel.
1914 * systemd-udev's .link files gained support for a new AutoNegotiation=
1915 setting for configuring Ethernet auto-negotiation.
1917 * systemd-networkd's .network files gained support for a new
1918 ListenPort= setting in the [DHCP] section to explicitly configure the
1919 UDP client port the DHCP client shall listen on.
1921 * .network files gained a new Unmanaged= boolean setting for explicitly
1922 excluding one or more interfaces from management by systemd-networkd.
1924 * The systemd-networkd ProxyARP= option has been renamed to
1925 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
1926 renamed to ReduceARPProxy=. The old names continue to be available
1929 * systemd-networkd gained support for configuring IPv6 Proxy NDP
1930 addresses via the new IPv6ProxyNDPAddress= .network file setting.
1932 * systemd-networkd's bonding device support gained support for two new
1933 configuration options ActiveSlave= and PrimarySlave=.
1935 * The various options in the [Match] section of .network files gained
1936 support for negative matching.
1938 * New systemd-specific mount options are now understood in /etc/fstab:
1940 x-systemd.mount-timeout= may be used to configure the maximum
1941 permitted runtime of the mount command.
1943 x-systemd.device-bound may be set to bind a mount point to its
1944 backing device unit, in order to automatically remove a mount point
1945 if its backing device is unplugged. This option may also be
1946 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
1947 on the block device, which is now automatically set for all CDROM
1948 drives, so that mounted CDs are automatically unmounted when they are
1949 removed from the drive.
1951 x-systemd.after= and x-systemd.before= may be used to explicitly
1952 order a mount after or before another unit or mount point.
1954 * Enqueued start jobs for device units are now automatically garbage
1955 collected if there are no jobs waiting for them anymore.
1957 * systemctl list-jobs gained two new switches: with --after, for every
1958 queued job the jobs it's waiting for are shown; with --before the
1959 jobs which it's blocking are shown.
1961 * systemd-nspawn gained support for ephemeral boots from disk images
1962 (or in other words: --ephemeral and --image= may now be
1963 combined). Moreover, ephemeral boots are now supported for normal
1964 directories, even if the backing file system is not btrfs. Of course,
1965 if the file system does not support file system snapshots or
1966 reflinks, the initial copy operation will be relatively expensive, but
1967 this should still be suitable for many use cases.
1969 * Calendar time specifications in .timer units now support
1970 specifications relative to the end of a month by using "~" instead of
1971 "-" as separator between month and day. For example, "*-02~03" means
1972 "the third last day in February". In addition a new syntax for
1973 repeated events has been added using the "/" character. For example,
1974 "9..17/2:00" means "every two hours from 9am to 5pm".
1976 * systemd-socket-proxyd gained a new parameter --connections-max= for
1977 configuring the maximum number of concurrent connections.
1979 * sd-id128 gained a new API for generating unique IDs for the host in a
1980 way that does not leak the machine ID. Specifically,
1981 sd_id128_get_machine_app_specific() derives an ID based on the
1982 machine ID a in well-defined, non-reversible, stable way. This is
1983 useful whenever an identifier for the host is needed but where the
1984 identifier shall not be useful to identify the system beyond the
1985 scope of the application itself. (Internally this uses HMAC-SHA256 as
1986 keyed hash function using the machine ID as input.)
1988 * NotifyAccess= gained a new supported value "exec". When set
1989 notifications are accepted from all processes systemd itself invoked,
1990 including all control processes.
1992 * .nspawn files gained support for defining overlay mounts using the
1993 Overlay= and OverlayReadOnly= options. Previously this functionality
1994 was only available on the systemd-nspawn command line.
1996 * systemd-nspawn's --bind= and --overlay= options gained support for
1997 bind/overlay mounts whose source lies within the container tree by
1998 prefixing the source path with "+".
2000 * systemd-nspawn's --bind= and --overlay= options gained support for
2001 automatically allocating a temporary source directory in /var/tmp
2002 that is removed when the container dies. Specifically, if the source
2003 directory is specified as empty string this mechanism is selected. An
2004 example usage is --overlay=+/var::/var, which creates an overlay
2005 mount based on the original /var contained in the image, overlayed
2006 with a temporary directory in the host's /var/tmp. This way changes
2007 to /var are automatically flushed when the container shuts down.
2009 * systemd-nspawn --image= option does now permit raw file system block
2010 devices (in addition to images containing partition tables, as
2013 * The disk image dissection logic in systemd-nspawn gained support for
2014 automatically setting up LUKS encrypted as well as Verity protected
2015 partitions. When a container is booted from an encrypted image the
2016 passphrase is queried at start-up time. When a container with Verity
2017 data is started, the root hash is search in a ".roothash" file
2018 accompanying the disk image (alternatively, pass the root hash via
2019 the new --root-hash= command line option).
2021 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
2022 be used to dissect disk images the same way as systemd-nspawn does
2023 it, following the Bootable Partition Specification. It may even be
2024 used to mount disk images with complex partition setups (including
2025 LUKS and Verity partitions) to a local host directory, in order to
2026 inspect them. This tool is not considered public API (yet), and is
2027 thus not installed into /usr/bin. Please do not rely on its
2028 existence, since it might go away or be changed in later systemd
2031 * A new generator "systemd-verity-generator" has been added, similar in
2032 style to "systemd-cryptsetup-generator", permitting automatic setup of
2033 Verity root partitions when systemd boots up. In order to make use of
2034 this your partition setup should follow the Discoverable Partitions
2035 Specification, and the GPT partition ID of the root file system
2036 partition should be identical to the upper 128bit of the Verity root
2037 hash. The GPT partition ID of the Verity partition protecting it
2038 should be the lower 128bit of the Verity root hash. If the partition
2039 image follows this model it is sufficient to specify a single
2040 "roothash=" kernel command line argument to both configure which root
2041 image and verity partition to use as well as the root hash for
2042 it. Note that systemd-nspawn's Verity support follows the same
2043 semantics, meaning that disk images with proper Verity data in place
2044 may be booted in containers with systemd-nspawn as well as on
2045 physical systems via the verity generator. Also note that the "mkosi"
2046 tool available at https://github.com/systemd/mkosi has been updated
2047 to generate Verity protected disk images following this scheme. In
2048 fact, it has been updated to generate disk images that optionally
2049 implement a complete UEFI SecureBoot trust chain, involving a signed
2050 kernel and initrd image that incorporates such a root hash as well as
2051 a Verity-enabled root partition.
2053 * The hardware database (hwdb) udev supports has been updated to carry
2054 accelerometer quirks.
2056 * All system services are now run with a fresh kernel keyring set up
2057 for them. The invocation ID is stored by default in it, thus
2058 providing a safe, non-overridable way to determine the invocation
2061 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
2062 options for bind mounting arbitrary paths in a service-specific
2063 way. When these options are used, arbitrary host or service files and
2064 directories may be mounted to arbitrary locations in the service's
2067 * Documentation has been added that lists all of systemd's low-level
2068 environment variables:
2070 https://github.com/systemd/systemd/blob/master/docs/ENVIRONMENT.md
2072 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
2073 whether a specific socket file descriptor matches a specified socket
2076 * systemd-firstboot has been updated to check for the
2077 systemd.firstboot= kernel command line option. It accepts a boolean
2078 and when set to false the first boot questions are skipped.
2080 * systemd-fstab-generator has been updated to check for the
2081 systemd.volatile= kernel command line option, which either takes an
2082 optional boolean parameter or the special value "state". If used the
2083 system may be booted in a "volatile" boot mode. Specifically,
2084 "systemd.volatile" is used, the root directory will be mounted as
2085 tmpfs, and only /usr is mounted from the actual root file system. If
2086 "systemd.volatile=state" is used, the root directory will be mounted
2087 as usual, but /var is mounted as tmpfs. This concept provides similar
2088 functionality as systemd-nspawn's --volatile= option, but provides it
2089 on physical boots. Use this option for implementing stateless
2090 systems, or testing systems with all state and/or configuration reset
2091 to the defaults. (Note though that many distributions are not
2092 prepared to boot up without a populated /etc or /var, though.)
2094 * systemd-gpt-auto-generator gained support for LUKS encrypted root
2095 partitions. Previously it only supported LUKS encrypted partitions
2096 for all other uses, except for the root partition itself.
2098 * Socket units gained support for listening on AF_VSOCK sockets for
2099 communication in virtualized QEMU environments.
2101 * The "configure" script gained a new option --with-fallback-hostname=
2102 for specifying the fallback hostname to use if none is configured in
2103 /etc/hostname. For example, by specifying
2104 --with-fallback-hostname=fedora it is possible to default to a
2105 hostname of "fedora" on pristine installations.
2107 * systemd-cgls gained support for a new --unit= switch for listing only
2108 the control groups of a specific unit. Similar --user-unit= has been
2109 added for listing only the control groups of a specific user unit.
2111 * systemd-mount gained a new --umount switch for unmounting a mount or
2112 automount point (and all mount/automount points below it).
2114 * systemd will now refuse full configuration reloads (via systemctl
2115 daemon-reload and related calls) unless at least 16MiB of free space
2116 are available in /run. This is a safety precaution in order to ensure
2117 that generators can safely operate after the reload completed.
2119 * A new unit file option RootImage= has been added, which has a similar
2120 effect as RootDirectory= but mounts the service's root directory from
2121 a disk image instead of plain directory. This logic reuses the same
2122 image dissection and mount logic that systemd-nspawn already uses,
2123 and hence supports any disk images systemd-nspawn supports, including
2124 those following the Discoverable Partition Specification, as well as
2125 Verity enabled images. This option enables systemd to run system
2126 services directly off disk images acting as resource bundles,
2127 possibly even including full integrity data.
2129 * A new MountAPIVFS= unit file option has been added, taking a boolean
2130 argument. If enabled /proc, /sys and /dev (collectively called the
2131 "API VFS") will be mounted for the service. This is only relevant if
2132 RootDirectory= or RootImage= is used for the service, as these mounts
2133 are of course in place in the host mount namespace anyway.
2135 * systemd-nspawn gained support for a new --pivot-root= switch. If
2136 specified the root directory within the container image is pivoted to
2137 the specified mount point, while the original root disk is moved to a
2138 different place. This option enables booting of ostree images
2139 directly with systemd-nspawn.
2141 * The systemd build scripts will no longer complain if the NTP server
2142 addresses are not changed from the defaults. Google now supports
2143 these NTP servers officially. We still recommend downstreams to
2144 properly register an NTP pool with the NTP pool project though.
2146 * coredumpctl gained a new "--reverse" option for printing the list
2147 of coredumps in reverse order.
2149 * coredumpctl will now show additional information about truncated and
2150 inaccessible coredumps, as well as coredumps that are still being
2151 processed. It also gained a new --quiet switch for suppressing
2152 additional informational message in its output.
2154 * coredumpctl gained support for only showing coredumps newer and/or
2155 older than specific timestamps, using the new --since= and --until=
2156 options, reminiscent of journalctl's options by the same name.
2158 * The systemd-coredump logic has been improved so that it may be reused
2159 to collect backtraces in non-compiled languages, for example in
2160 scripting languages such as Python.
2162 * machinectl will now show the UID shift of local containers, if user
2163 namespacing is enabled for them.
2165 * systemd will now optionally run "environment generator" binaries at
2166 configuration load time. They may be used to add environment
2167 variables to the environment block passed to services invoked. One
2168 user environment generator is shipped by default that sets up
2169 environment variables based on files dropped into /etc/environment.d
2170 and ~/.config/environment.d/.
2172 * systemd-resolved now includes the new, recently published 2017 DNSSEC
2175 * hostnamed has been updated to report a new chassis type of
2176 "convertible" to cover "foldable" laptops that can both act as a
2177 tablet and as a laptop, such as various Lenovo Yoga devices.
2179 Contributions from: Adrián López, Alexander Galanin, Alexander
2180 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
2181 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
2182 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
2183 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
2184 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
2185 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
2186 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
2187 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
2188 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
2189 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
2190 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
2191 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
2192 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
2193 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
2194 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
2195 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
2196 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
2197 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
2198 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
2199 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
2200 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
2201 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
2202 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
2203 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
2204 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
2205 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
2206 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
2209 — Berlin, 2017-03-01
2213 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
2214 RestrictAddressFamilies= enabled. These sandboxing options should
2215 generally be compatible with the various external udev call-out
2216 binaries we are aware of, however there may be exceptions, in
2217 particular when exotic languages for these call-outs are used. In
2218 this case, consider turning off these settings locally.
2220 * The new RemoveIPC= option can be used to remove IPC objects owned by
2221 the user or group of a service when that service exits.
2223 * The new ProtectKernelModules= option can be used to disable explicit
2224 load and unload operations of kernel modules by a service. In
2225 addition access to /usr/lib/modules is removed if this option is set.
2227 * ProtectSystem= option gained a new value "strict", which causes the
2228 whole file system tree with the exception of /dev, /proc, and /sys,
2229 to be remounted read-only for a service.
2231 * The new ProtectKernelTunables= option can be used to disable
2232 modification of configuration files in /sys and /proc by a service.
2233 Various directories and files are remounted read-only, so access is
2234 restricted even if the file permissions would allow it.
2236 * The new ProtectControlGroups= option can be used to disable write
2237 access by a service to /sys/fs/cgroup.
2239 * Various systemd services have been hardened with
2240 ProtectKernelTunables=yes, ProtectControlGroups=yes,
2241 RestrictAddressFamilies=.
2243 * Support for dynamically creating users for the lifetime of a service
2244 has been added. If DynamicUser=yes is specified, user and group IDs
2245 will be allocated from the range 61184..65519 for the lifetime of the
2246 service. They can be resolved using the new nss-systemd.so NSS
2247 module. The module must be enabled in /etc/nsswitch.conf. Services
2248 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
2249 any resources allocated by the service will be cleaned up when the
2250 service exits. They also have ProtectHome=read-only and
2251 ProtectSystem=strict enabled, so they are not able to make any
2252 permanent modifications to the system.
2254 * The nss-systemd module also always resolves root and nobody, making
2255 it possible to have no /etc/passwd or /etc/group files in minimal
2256 container or chroot environments.
2258 * Services may be started with their own user namespace using the new
2259 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
2260 under which the service is running are mapped. All other users are
2263 * Support for the cgroup namespace has been added to systemd-nspawn. If
2264 supported by kernel, the container system started by systemd-nspawn
2265 will have its own view of the cgroup hierarchy. This new behaviour
2266 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
2268 * The new MemorySwapMax= option can be used to limit the maximum swap
2269 usage under the unified cgroup hierarchy.
2271 * Support for the CPU controller in the unified cgroup hierarchy has
2272 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
2273 options. This controller requires out-of-tree patches for the kernel
2274 and the support is provisional.
2276 * Mount and automount units may now be created transiently
2277 (i.e. dynamically at runtime via the bus API, instead of requiring
2278 unit files in the file system).
2280 * systemd-mount is a new tool which may mount file systems – much like
2281 mount(8), optionally pulling in additional dependencies through
2282 transient .mount and .automount units. For example, this tool
2283 automatically runs fsck on a backing block device before mounting,
2284 and allows the automount logic to be used dynamically from the
2285 command line for establishing mount points. This tool is particularly
2286 useful when dealing with removable media, as it will ensure fsck is
2287 run – if necessary – before the first access and that the file system
2288 is quickly unmounted after each access by utilizing the automount
2289 logic. This maximizes the chance that the file system on the
2290 removable media stays in a clean state, and if it isn't in a clean
2291 state is fixed automatically.
2293 * LazyUnmount=yes option for mount units has been added to expose the
2294 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
2297 * /efi will be used as the mount point of the EFI boot partition, if
2298 the directory is present, and the mount point was not configured
2299 through other means (e.g. fstab). If /efi directory does not exist,
2300 /boot will be used as before. This makes it easier to automatically
2301 mount the EFI partition on systems where /boot is used for something
2304 * When operating on GPT disk images for containers, systemd-nspawn will
2305 now mount the ESP to /boot or /efi according to the same rules as PID
2306 1 running on a host. This allows tools like "bootctl" to operate
2307 correctly within such containers, in order to make container images
2308 bootable on physical systems.
2310 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
2312 * Two new user session targets have been added to support running
2313 graphical sessions under the systemd --user instance:
2314 graphical-session.target and graphical-session-pre.target. See
2315 systemd.special(7) for a description of how those targets should be
2318 * The vconsole initialization code has been significantly reworked to
2319 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
2320 support unicode keymaps. Font and keymap configuration will now be
2321 copied to all allocated virtual consoles.
2323 * FreeBSD's bhyve virtualization is now detected.
2325 * Information recorded in the journal for core dumps now includes the
2326 contents of /proc/mountinfo and the command line of the process at
2327 the top of the process hierarchy (which is usually the init process
2330 * systemd-journal-gatewayd learned the --directory= option to serve
2331 files from the specified location.
2333 * journalctl --root=… can be used to peruse the journal in the
2334 /var/log/ directories inside of a container tree. This is similar to
2335 the existing --machine= option, but does not require the container to
2338 * The hardware database has been extended to support
2339 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
2342 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
2343 specify the click rate for mice which include a horizontal wheel with
2344 a click rate that is different than the one for the vertical wheel.
2346 * systemd-run gained a new --wait option that makes service execution
2347 synchronous. (Specifically, the command will not return until the
2348 specified service binary exited.)
2350 * systemctl gained a new --wait option that causes the start command to
2351 wait until the units being started have terminated again.
2353 * A new journal output mode "short-full" has been added which displays
2354 timestamps with abbreviated English day names and adds a timezone
2355 suffix. Those timestamps include more information than the default
2356 "short" output mode, and can be passed directly to journalctl's
2357 --since= and --until= options.
2359 * /etc/resolv.conf will be bind-mounted into containers started by
2360 systemd-nspawn, if possible, so any changes to resolv.conf contents
2361 are automatically propagated to the container.
2363 * The number of instances for socket-activated services originating
2364 from a single IP address can be limited with
2365 MaxConnectionsPerSource=, extending the existing setting of
2368 * systemd-networkd gained support for vcan ("Virtual CAN") interface
2371 * .netdev and .network configuration can now be extended through
2374 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
2375 Segmentation Offload, Generic Receive Offload, Large Receive Offload
2376 can be enabled and disabled using the new UDPSegmentationOffload=,
2377 TCPSegmentationOffload=, GenericSegmentationOffload=,
2378 GenericReceiveOffload=, LargeReceiveOffload= options in the
2379 [Link] section of .link files.
2381 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
2382 Port VLAN ID can be configured for bridge devices using the new STP=,
2383 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
2384 section of .netdev files.
2386 * The route table to which routes received over DHCP or RA should be
2387 added can be configured with the new RouteTable= option in the [DHCP]
2388 and [IPv6AcceptRA] sections of .network files.
2390 * The Address Resolution Protocol can be disabled on links managed by
2391 systemd-networkd using the ARP=no setting in the [Link] section of
2394 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
2395 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
2396 encode information about the result and exit codes of the current
2397 service runtime cycle.
2399 * systemd-sysctl will now configure kernel parameters in the order
2400 they occur in the configuration files. This matches what sysctl
2401 has been traditionally doing.
2403 * kernel-install "plugins" that are executed to perform various
2404 tasks after a new kernel is added and before an old one is removed
2405 can now return a special value to terminate the procedure and
2406 prevent any later plugins from running.
2408 * Journald's SplitMode=login setting has been deprecated. It has been
2409 removed from documentation, and its use is discouraged. In a future
2410 release it will be completely removed, and made equivalent to current
2411 default of SplitMode=uid.
2413 * Storage=both option setting in /etc/systemd/coredump.conf has been
2414 removed. With fast LZ4 compression storing the core dump twice is not
2417 * The --share-system systemd-nspawn option has been replaced with an
2418 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
2419 this functionality is discouraged. In addition the variables
2420 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
2421 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
2422 individual namespaces.
2424 * "machinectl list" now shows the IP address of running containers in
2425 the output, as well as OS release information.
2427 * "loginctl list" now shows the TTY of each session in the output.
2429 * sd-bus gained new API calls sd_bus_track_set_recursive(),
2430 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
2431 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
2432 tracking objects in a "recursive" mode, where a single client can be
2433 counted multiple times, if it takes multiple references.
2435 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
2436 sd_bus_get_exit_on_disconnect(). They may be used to make a
2437 process using sd-bus automatically exit if the bus connection is
2440 * Bus clients of the service manager may now "pin" loaded units into
2441 memory, by taking an explicit reference on them. This is useful to
2442 ensure the client can retrieve runtime data about the service even
2443 after the service completed execution. Taking such a reference is
2444 available only for privileged clients and should be helpful to watch
2445 running services in a race-free manner, and in particular collect
2446 information about exit statuses and results.
2448 * The nss-resolve module has been changed to strictly return UNAVAIL
2449 when communication via D-Bus with resolved failed, and NOTFOUND when
2450 a lookup completed but was negative. This means it is now possible to
2451 neatly configure fallbacks using nsswitch.conf result checking
2452 expressions. Taking benefit of this, the new recommended
2453 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
2455 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
2457 * A new setting CtrlAltDelBurstAction= has been added to
2458 /etc/systemd/system.conf which may be used to configure the precise
2459 behaviour if the user on the console presses Ctrl-Alt-Del more often
2460 than 7 times in 2s. Previously this would unconditionally result in
2461 an expedited, immediate reboot. With this new setting the precise
2462 operation may be configured in more detail, and also turned off
2465 * In .netdev files two new settings RemoteChecksumTx= and
2466 RemoteChecksumRx= are now understood that permit configuring the
2467 remote checksumming logic for VXLAN networks.
2469 * The service manager learnt a new "invocation ID" concept for invoked
2470 services. Each runtime cycle of a service will get a new invocation
2471 ID (a 128bit random UUID) assigned that identifies the current
2472 run of the service uniquely and globally. A new invocation ID
2473 is generated each time a service starts up. The journal will store
2474 the invocation ID of a service along with any logged messages, thus
2475 making the invocation ID useful for matching the online runtime of a
2476 service with the offline log data it generated in a safe way without
2477 relying on synchronized timestamps. In many ways this new service
2478 invocation ID concept is similar to the kernel's boot ID concept that
2479 uniquely and globally identifies the runtime of each boot. The
2480 invocation ID of a service is passed to the service itself via an
2481 environment variable ($INVOCATION_ID). A new bus call
2482 GetUnitByInvocationID() has been added that is similar to GetUnit()
2483 but instead of retrieving the bus path for a unit by its name
2484 retrieves it by its invocation ID. The returned path is valid only as
2485 long as the passed invocation ID is current.
2487 * systemd-resolved gained a new "DNSStubListener" setting in
2488 resolved.conf. It either takes a boolean value or the special values
2489 "udp" and "tcp", and configures whether to enable the stub DNS
2490 listener on 127.0.0.53:53.
2492 * IP addresses configured via networkd may now carry additional
2493 configuration settings supported by the kernel. New options include:
2494 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
2495 PrefixRoute=, AutoJoin=.
2497 * The PAM configuration fragment file for "user@.service" shipped with
2498 systemd (i.e. the --user instance of systemd) has been stripped to
2499 the minimum necessary to make the system boot. Previously, it
2500 contained Fedora-specific stanzas that did not apply to other
2501 distributions. It is expected that downstream distributions add
2502 additional configuration lines, matching their needs to this file,
2503 using it only as rough template of what systemd itself needs. Note
2504 that this reduced fragment does not even include an invocation of
2505 pam_limits which most distributions probably want to add, even though
2506 systemd itself does not need it. (There's also the new build time
2507 option --with-pamconfdir=no to disable installation of the PAM
2510 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
2511 capability is now also dropped from its set (in addition to
2512 CAP_SYS_MKNOD as before).
2514 * In service unit files it is now possible to connect a specific named
2515 file descriptor with stdin/stdout/stdout of an executed service. The
2516 name may be specified in matching .socket units using the
2517 FileDescriptorName= setting.
2519 * A number of journal settings may now be configured on the kernel
2520 command line. Specifically, the following options are now understood:
2521 systemd.journald.max_level_console=,
2522 systemd.journald.max_level_store=,
2523 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
2524 systemd.journald.max_level_wall=.
2526 * "systemctl is-enabled --full" will now show by which symlinks a unit
2527 file is enabled in the unit dependency tree.
2529 * Support for VeraCrypt encrypted partitions has been added to the
2530 "cryptsetup" logic and /etc/crypttab.
2532 * systemd-detect-virt gained support for a new --private-users switch
2533 that checks whether the invoking processes are running inside a user
2534 namespace. Similar, a new special value "private-users" for the
2535 existing ConditionVirtualization= setting has been added, permitting
2536 skipping of specific units in user namespace environments.
2538 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
2539 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
2540 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
2541 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
2542 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
2543 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
2544 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
2545 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
2546 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
2547 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
2548 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
2549 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
2550 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
2551 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
2552 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
2553 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
2554 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
2555 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
2556 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
2557 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
2558 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
2559 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
2560 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
2561 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
2562 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
2563 Jędrzejewski-Szmek, Zeal Jagannatha
2565 — Santa Fe, 2016-11-03
2569 * In service units the various ExecXYZ= settings have been extended
2570 with an additional special character as first argument of the
2571 assigned value: if the character '+' is used the specified command
2572 line it will be run with full privileges, regardless of User=,
2573 Group=, CapabilityBoundingSet= and similar options. The effect is
2574 similar to the existing PermissionsStartOnly= option, but allows
2575 configuration of this concept for each executed command line
2578 * Services may now alter the service watchdog timeout at runtime by
2579 sending a WATCHDOG_USEC= message via sd_notify().
2581 * MemoryLimit= and related unit settings now optionally take percentage
2582 specifications. The percentage is taken relative to the amount of
2583 physical memory in the system (or in case of containers, the assigned
2584 amount of memory). This allows scaling service resources neatly with
2585 the amount of RAM available on the system. Similarly, systemd-logind's
2586 RuntimeDirectorySize= option now also optionally takes percentage
2589 * In similar fashion TasksMax= takes percentage values now, too. The
2590 value is taken relative to the configured maximum number of processes
2591 on the system. The per-service task maximum has been changed to 15%
2592 using this functionality. (Effectively this is an increase of 512 →
2593 4915 for service units, given the kernel's default pid_max setting.)
2595 * Calendar time specifications in .timer units now understand a ".."
2596 syntax for time ranges. Example: "4..7:10" may now be used for
2597 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
2600 * The InaccessableDirectories=, ReadOnlyDirectories= and
2601 ReadWriteDirectories= unit file settings have been renamed to
2602 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
2603 applied to all kinds of file nodes, and not just directories, with
2604 the exception of symlinks. Specifically these settings may now be
2605 used on block and character device nodes, UNIX sockets and FIFOS as
2606 well as regular files. The old names of these settings remain
2607 available for compatibility.
2609 * systemd will now log about all service processes it kills forcibly
2610 (using SIGKILL) because they remained after the clean shutdown phase
2611 of the service completed. This should help identifying services that
2612 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
2613 systemd-logind's configuration a similar log message is generated for
2614 processes killed at the end of each session due to this setting.
2616 * systemd will now set the $JOURNAL_STREAM environment variable for all
2617 services whose stdout/stderr are connected to the Journal (which
2618 effectively means by default: all services). The variable contains
2619 the device and inode number of the file descriptor used for
2620 stdout/stderr. This may be used by invoked programs to detect whether
2621 their stdout/stderr is connected to the Journal, in which case they
2622 can switch over to direct Journal communication, thus being able to
2623 pass extended, structured metadata along with their log messages. As
2624 one example, this is now used by glib's logging primitives.
2626 * When using systemd's default tmp.mount unit for /tmp, the mount point
2627 will now be established with the "nosuid" and "nodev" options. This
2628 avoids privilege escalation attacks that put traps and exploits into
2629 /tmp. However, this might cause problems if you e. g. put container
2630 images or overlays into /tmp; if you need this, override tmp.mount's
2631 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
2634 * systemd now supports the "memory" cgroup controller also on
2637 * The systemd-cgtop tool now optionally takes a control group path as
2638 command line argument. If specified, the control group list shown is
2639 limited to subgroups of that group.
2641 * The SystemCallFilter= unit file setting gained support for
2642 pre-defined, named system call filter sets. For example
2643 SystemCallFilter=@clock is now an effective way to make all clock
2644 changing-related system calls unavailable to a service. A number of
2645 similar pre-defined groups are defined. Writing system call filters
2646 for system services is simplified substantially with this new
2647 concept. Accordingly, all of systemd's own, long-running services now
2648 enable system call filtering based on this, by default.
2650 * A new service setting MemoryDenyWriteExecute= has been added, taking
2651 a boolean value. If turned on, a service may no longer create memory
2652 mappings that are writable and executable at the same time. This
2653 enhances security for services where this is enabled as it becomes
2654 harder to dynamically write and then execute memory in exploited
2655 service processes. This option has been enabled for all of systemd's
2656 own long-running services.
2658 * A new RestrictRealtime= service setting has been added, taking a
2659 boolean argument. If set the service's processes may no longer
2660 acquire realtime scheduling. This improves security as realtime
2661 scheduling may otherwise be used to easily freeze the system.
2663 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
2664 value. This may be used for requesting that the system manager inside
2665 of the container reports start-up completion to nspawn which then
2666 propagates this notification further to the service manager
2667 supervising nspawn itself. A related option NotifyReady= in .nspawn
2668 files has been added too. This functionality allows ordering of the
2669 start-up of multiple containers using the usual systemd ordering
2672 * machinectl gained a new command "stop" that is an alias for
2675 * systemd-resolved gained support for contacting DNS servers on
2676 link-local IPv6 addresses.
2678 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
2679 its caches. A method call for requesting the same operation has been
2680 added to the bus API too, and is made available via "systemd-resolve
2683 * systemd-resolve gained a new --status switch. If passed a brief
2684 summary of the used DNS configuration with per-interface information
2687 * resolved.conf gained a new Cache= boolean option, defaulting to
2688 on. If turned off local DNS caching is disabled. This comes with a
2689 performance penalty in particular when DNSSEC is enabled. Note that
2690 resolved disables its internal caching implicitly anyway, when the
2691 configured DNS server is on a host-local IP address such as ::1 or
2692 127.0.0.1, thus automatically avoiding double local caching.
2694 * systemd-resolved now listens on the local IP address 127.0.0.53:53
2695 for DNS requests. This improves compatibility with local programs
2696 that do not use the libc NSS or systemd-resolved's bus APIs for name
2697 resolution. This minimal DNS service is only available to local
2698 programs and does not implement the full DNS protocol, but enough to
2699 cover local DNS clients. A new, static resolv.conf file, listing just
2700 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
2701 now recommended to make /etc/resolv.conf a symlink to this file in
2702 order to route all DNS lookups to systemd-resolved, regardless if
2703 done via NSS, the bus API or raw DNS packets. Note that this local
2704 DNS service is not as fully featured as the libc NSS or
2705 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
2706 used to deliver link-local address information (as this implies
2707 sending a local interface index along), LLMNR/mDNS support via this
2708 interface is severely restricted. It is thus strongly recommended for
2709 all applications to use the libc NSS API or native systemd-resolved
2712 * systemd-networkd's bridge support learned a new setting
2713 VLANFiltering= for controlling VLAN filtering. Moreover a new section
2714 in .network files has been added for configuring VLAN bridging in
2715 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
2717 * systemd-networkd's IPv6 Router Advertisement code now makes use of
2718 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
2719 now be acquired without relying on DHCPv6. Two new options
2720 UseDomains= and UseDNS= have been added to configure this behaviour.
2722 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
2723 renamed IPv6AcceptRA=, without altering its behaviour. The old
2724 setting name remains available for compatibility reasons.
2726 * The systemd-networkd VTI/VTI6 tunneling support gained new options
2727 Key=, InputKey= and OutputKey=.
2729 * systemd-networkd gained support for VRF ("Virtual Routing Function")
2730 interface configuration.
2732 * "systemctl edit" may now be used to create new unit files by
2733 specifying the --force switch.
2735 * sd-event gained a new function sd_event_get_iteration() for
2736 requesting the current iteration counter of the event loop. It starts
2737 at zero and is increased by one with each event loop iteration.
2739 * A new rpm macro %systemd_ordering is provided by the macros.systemd
2740 file. It can be used in lieu of %systemd_requires in packages which
2741 don't use any systemd functionality and are intended to be installed
2742 in minimal containers without systemd present. This macro provides
2743 ordering dependencies to ensure that if the package is installed in
2744 the same rpm transaction as systemd, systemd will be installed before
2745 the scriptlets for the package are executed, allowing unit presets
2748 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
2749 been added to simplify packaging of generators.
2751 * The os-release file gained VERSION_CODENAME field for the
2752 distribution nickname (e.g. VERSION_CODENAME=woody).
2754 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
2755 can be set to disable parsing of metadata and the creation
2756 of persistent symlinks for that device.
2758 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
2759 to make them available to logged-in users has been reverted.
2761 * Much of the common code of the various systemd components is now
2762 built into an internal shared library libsystemd-shared-231.so
2763 (incorporating the systemd version number in the name, to be updated
2764 with future releases) that the components link to. This should
2765 decrease systemd footprint both in memory during runtime and on
2766 disk. Note that the shared library is not for public use, and is
2767 neither API nor ABI stable, but is likely to change with every new
2768 released update. Packagers need to make sure that binaries
2769 linking to libsystemd-shared.so are updated in step with the
2772 * Configuration for "mkosi" is now part of the systemd
2773 repository. mkosi is a tool to easily build legacy-free OS images,
2774 and is available on github: https://github.com/systemd/mkosi. If
2775 "mkosi" is invoked in the build tree a new raw OS image is generated
2776 incorporating the systemd sources currently being worked on and a
2777 clean, fresh distribution installation. The generated OS image may be
2778 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
2779 UEFI PC. This functionality is particularly useful to easily test
2780 local changes made to systemd in a pristine, defined environment. See
2781 doc/HACKING for details.
2783 * configure learned the --with-support-url= option to specify the
2784 distribution's bugtracker.
2786 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
2787 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
2788 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
2789 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
2790 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
2791 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
2792 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
2793 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
2794 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
2795 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
2796 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
2797 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
2798 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
2799 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
2800 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
2801 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
2802 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
2803 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
2804 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
2806 — Berlin, 2016-07-25
2810 * DNSSEC is now turned on by default in systemd-resolved (in
2811 "allow-downgrade" mode), but may be turned off during compile time by
2812 passing "--with-default-dnssec=no" to "configure" (and of course,
2813 during runtime with DNSSEC= in resolved.conf). We recommend
2814 downstreams to leave this on at least during development cycles and
2815 report any issues with the DNSSEC logic upstream. We are very
2816 interested in collecting feedback about the DNSSEC validator and its
2817 limitations in the wild. Note however, that DNSSEC support is
2818 probably nothing downstreams should turn on in stable distros just
2819 yet, as it might create incompatibilities with a few DNS servers and
2820 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
2821 automatically whenever we detect such incompatible setups, but there
2822 might be systems we do not cover yet. Hence: please help us testing
2823 the DNSSEC code, leave this on where you can, report back, but then
2824 again don't consider turning this on in your stable, LTS or
2825 production release just yet. (Note that you have to enable
2826 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
2827 and its DNSSEC mode for host name resolution from local
2830 * systemd-resolve conveniently resolves DANE records with the --tlsa
2831 option and OPENPGPKEY records with the --openpgp option. It also
2832 supports dumping raw DNS record data via the new --raw= switch.
2834 * systemd-logind will now by default terminate user processes that are
2835 part of the user session scope unit (session-XX.scope) when the user
2836 logs out. This behavior is controlled by the KillUserProcesses=
2837 setting in logind.conf, and the previous default of "no" is now
2838 changed to "yes". This means that user sessions will be properly
2839 cleaned up after, but additional steps are necessary to allow
2840 intentionally long-running processes to survive logout.
2842 While the user is logged in at least once, user@.service is running,
2843 and any service that should survive the end of any individual login
2844 session can be started at a user service or scope using systemd-run.
2845 systemd-run(1) man page has been extended with an example which shows
2846 how to run screen in a scope unit underneath user@.service. The same
2847 command works for tmux.
2849 After the user logs out of all sessions, user@.service will be
2850 terminated too, by default, unless the user has "lingering" enabled.
2851 To effectively allow users to run long-term tasks even if they are
2852 logged out, lingering must be enabled for them. See loginctl(1) for
2853 details. The default polkit policy was modified to allow users to
2854 set lingering for themselves without authentication.
2856 Previous defaults can be restored at compile time by the
2857 --without-kill-user-processes option to "configure".
2859 * systemd-logind gained new configuration settings SessionsMax= and
2860 InhibitorsMax=, both with a default of 8192. It will not register new
2861 user sessions or inhibitors above this limit.
2863 * systemd-logind will now reload configuration on SIGHUP.
2865 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
2866 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
2867 enable. Also, support for the "io" cgroup controller in the unified
2868 hierarchy has been added, so that the "memory", "pids" and "io" are
2869 now the controllers that are supported on the unified hierarchy.
2871 WARNING: it is not possible to use previous systemd versions with
2872 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
2873 is necessary to also update systemd in the initramfs if using the
2874 unified hierarchy. An updated SELinux policy is also required.
2876 * LLDP support has been extended, and both passive (receive-only) and
2877 active (sender) modes are supported. Passive mode ("routers-only") is
2878 enabled by default in systemd-networkd. Active LLDP mode is enabled
2879 by default for containers on the internal network. The "networkctl
2880 lldp" command may be used to list information gathered. "networkctl
2881 status" will also show basic LLDP information on connected peers now.
2883 * The IAID and DUID unique identifier sent in DHCP requests may now be
2884 configured for the system and each .network file managed by
2885 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
2887 * systemd-networkd gained support for configuring proxy ARP support for
2888 each interface, via the ProxyArp= setting in .network files. It also
2889 gained support for configuring the multicast querier feature of
2890 bridge devices, via the new MulticastQuerier= setting in .netdev
2891 files. Similarly, snooping on the IGMP traffic can be controlled
2892 via the new setting MulticastSnooping=.
2894 A new setting PreferredLifetime= has been added for addresses
2895 configured in .network file to configure the lifetime intended for an
2898 The systemd-networkd DHCP server gained the option EmitRouter=, which
2899 defaults to yes, to configure whether the DHCP Option 3 (Router)
2902 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
2903 systemd-socket-activate and installed into /usr/bin. It is now fully
2906 * systemd-journald now uses separate threads to flush changes to disk
2907 when closing journal files, thus reducing impact of slow disk I/O on
2908 logging performance.
2910 * The sd-journal API gained two new calls
2911 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
2912 can be used to open journal files using file descriptors instead of
2913 file or directory paths. sd_journal_open_container() has been
2914 deprecated, sd_journal_open_directory_fd() should be used instead
2915 with the flag SD_JOURNAL_OS_ROOT.
2917 * journalctl learned a new output mode "-o short-unix" that outputs log
2918 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
2919 UTC). It also gained support for a new --no-hostname setting to
2920 suppress the hostname column in the family of "short" output modes.
2922 * systemd-ask-password now optionally skips printing of the password to
2923 stdout with --no-output which can be useful in scripts.
2925 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
2926 (devices tagged with ID_MAKER_TOOL) are now tagged with
2927 "uaccess" and are available to logged in users.
2929 * The DeviceAllow= unit setting now supports specifiers (with "%").
2931 * "systemctl show" gained a new --value switch, which allows print a
2932 only the contents of a specific unit property, without also printing
2933 the property's name. Similar support was added to "show*" verbs
2934 of loginctl and machinectl that output "key=value" lists.
2936 * A new unit type "generated" was added for files dynamically generated
2937 by generator tools. Similarly, a new unit type "transient" is used
2938 for unit files created using the runtime API. "systemctl enable" will
2939 refuse to operate on such files.
2941 * A new command "systemctl revert" has been added that may be used to
2942 revert to the vendor version of a unit file, in case local changes
2943 have been made by adding drop-ins or overriding the unit file.
2945 * "machinectl clean" gained a new verb to automatically remove all or
2946 just hidden container images.
2948 * systemd-tmpfiles gained support for a new line type "e" for emptying
2949 directories, if they exist, without creating them if they don't.
2951 * systemd-nspawn gained support for automatically patching the UID/GIDs
2952 of the owners and the ACLs of all files and directories in a
2953 container tree to match the UID/GID user namespacing range selected
2954 for the container invocation. This mode is enabled via the new
2955 --private-users-chown switch. It also gained support for
2956 automatically choosing a free, previously unused UID/GID range when
2957 starting a container, via the new --private-users=pick setting (which
2958 implies --private-users-chown). Together, these options for the first
2959 time make user namespacing for nspawn containers fully automatic and
2960 thus deployable. The systemd-nspawn@.service template unit file has
2961 been changed to use this functionality by default.
2963 * systemd-nspawn gained a new --network-zone= switch, that allows
2964 creating ad-hoc virtual Ethernet links between multiple containers,
2965 that only exist as long as at least one container referencing them is
2966 running. This allows easy connecting of multiple containers with a
2967 common link that implements an Ethernet broadcast domain. Each of
2968 these network "zones" may be named relatively freely by the user, and
2969 may be referenced by any number of containers, but each container may
2970 only reference one of these "zones". On the lower level, this is
2971 implemented by an automatically managed bridge network interface for
2972 each zone, that is created when the first container referencing its
2973 zone is created and removed when the last one referencing its zone
2976 * The default start timeout may now be configured on the kernel command
2977 line via systemd.default_timeout_start_sec=. It was already
2978 configurable via the DefaultTimeoutStartSec= option in
2979 /etc/systemd/system.conf.
2981 * Socket units gained a new TriggerLimitIntervalSec= and
2982 TriggerLimitBurst= setting to configure a limit on the activation
2983 rate of the socket unit.
2985 * The LimitNICE= setting now optionally takes normal UNIX nice values
2986 in addition to the raw integer limit value. If the specified
2987 parameter is prefixed with "+" or "-" and is in the range -20..19 the
2988 value is understood as UNIX nice value. If not prefixed like this it
2989 is understood as raw RLIMIT_NICE limit.
2991 * Note that the effect of the PrivateDevices= unit file setting changed
2992 slightly with this release: the per-device /dev file system will be
2993 mounted read-only from this version on, and will have "noexec"
2994 set. This (minor) change of behavior might cause some (exceptional)
2995 legacy software to break, when PrivateDevices=yes is set for its
2996 service. Please leave PrivateDevices= off if you run into problems
2999 * systemd-bootchart has been split out to a separate repository:
3000 https://github.com/systemd/systemd-bootchart
3002 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
3003 merged into the kernel in its current form.
3005 * The compatibility libraries libsystemd-daemon.so,
3006 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
3007 which have been deprecated since systemd-209 have been removed along
3008 with the corresponding pkg-config files. All symbols provided by
3009 those libraries are provided by libsystemd.so.
3011 * The Capabilities= unit file setting has been removed (it is ignored
3012 for backwards compatibility). AmbientCapabilities= and
3013 CapabilityBoundingSet= should be used instead.
3015 * A new special target has been added, initrd-root-device.target,
3016 which creates a synchronization point for dependencies of the root
3017 device in early userspace. Initramfs builders must ensure that this
3018 target is now included in early userspace.
3020 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
3021 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
3022 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
3023 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
3024 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
3025 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
3026 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
3027 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
3028 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
3029 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
3030 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
3031 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
3032 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
3033 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
3034 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
3035 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
3036 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
3037 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
3038 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
3039 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
3040 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
3041 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
3042 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
3043 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
3044 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
3047 — Fairfax, 2016-05-21
3051 * The systemd-resolved DNS resolver service has gained a substantial
3052 set of new features, most prominently it may now act as a DNSSEC
3053 validating stub resolver. DNSSEC mode is currently turned off by
3054 default, but is expected to be turned on by default in one of the
3055 next releases. For now, we invite everybody to test the DNSSEC logic
3056 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
3057 service also gained a full set of D-Bus interfaces, including calls
3058 to configure DNS and DNSSEC settings per link (for use by external
3059 network management software). systemd-resolved and systemd-networkd
3060 now distinguish between "search" and "routing" domains. The former
3061 are used to qualify single-label names, the latter are used purely
3062 for routing lookups within certain domains to specific links.
3063 resolved now also synthesizes RRs for all entries from /etc/hosts.
3065 * The systemd-resolve tool (which is a client utility for
3066 systemd-resolved) has been improved considerably and is now fully
3067 supported and documented. Hence it has moved from /usr/lib/systemd to
3070 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
3073 * The coredump collection logic has been reworked: when a coredump is
3074 collected it is now written to disk, compressed and processed
3075 (including stacktrace extraction) from a new instantiated service
3076 systemd-coredump@.service, instead of directly from the
3077 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
3078 processing large coredumps can take up a substantial amount of
3079 resources and time, and this previously happened entirely outside of
3080 systemd's service supervision. With the new logic the core_pattern
3081 hook only does minimal metadata collection before passing off control
3082 to the new instantiated service, which is configured with a time
3083 limit, a nice level and other settings to minimize negative impact on
3084 the rest of the system. Also note that the new logic will honour the
3085 RLIMIT_CORE setting of the crashed process, which now allows users
3086 and processes to turn off coredumping for their processes by setting
3089 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
3090 and all forked processes by default. Previously, PID 1 would leave
3091 the setting at "0" for all processes, as set by the kernel. Note that
3092 the resource limit traditionally has no effect on the generated
3093 coredumps on the system if the /proc/sys/kernel/core_pattern hook
3094 logic is used. Since the limit is now honoured (see above) its
3095 default has been changed so that the coredumping logic is enabled by
3096 default for all processes, while allowing specific opt-out.
3098 * When the stacktrace is extracted from processes of system users, this
3099 is now done as "systemd-coredump" user, in order to sandbox this
3100 potentially security sensitive parsing operation. (Note that when
3101 processing coredumps of normal users this is done under the user ID
3102 of process that crashed, as before.) Packagers should take notice
3103 that it is now necessary to create the "systemd-coredump" system user
3104 and group at package installation time.
3106 * The systemd-activate socket activation testing tool gained support
3107 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
3108 and --seqpacket switches. It also has been extended to support both
3109 new-style and inetd-style file descriptor passing. Use the new
3110 --inetd switch to request inetd-style file descriptor passing.
3112 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
3113 variable, which takes a boolean value. If set to false, ANSI color
3114 output is disabled in the tools even when run on a terminal that
3117 * The VXLAN support in networkd now supports two new settings
3118 DestinationPort= and PortRange=.
3120 * A new systemd.machine_id= kernel command line switch has been added,
3121 that may be used to set the machine ID in /etc/machine-id if it is
3122 not initialized yet. This command line option has no effect if the
3123 file is already initialized.
3125 * systemd-nspawn gained a new --as-pid2 switch that invokes any
3126 specified command line as PID 2 rather than PID 1 in the
3127 container. In this mode PID 1 is a minimal stub init process that
3128 implements the special POSIX and Linux semantics of PID 1 regarding
3129 signal and child process management. Note that this stub init process
3130 is implemented in nspawn itself and requires no support from the
3131 container image. This new logic is useful to support running
3132 arbitrary commands in the container, as normal processes are
3133 generally not prepared to run as PID 1.
3135 * systemd-nspawn gained a new --chdir= switch for setting the current
3136 working directory for the process started in the container.
3138 * "journalctl /dev/sda" will now output all kernel log messages for
3139 specified device from the current boot, in addition to all devices
3140 that are parents of it. This should make log output about devices
3141 pretty useful, as long as kernel drivers attach enough metadata to
3142 the log messages. (The usual SATA drivers do.)
3144 * The sd-journal API gained two new calls
3145 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
3146 that report whether log data from /run or /var has been found.
3148 * journalctl gained a new switch "--fields" that prints all journal
3149 record field names currently in use in the journal. This is backed
3150 by two new sd-journal API calls sd_journal_enumerate_fields() and
3151 sd_journal_restart_fields().
3153 * Most configurable timeouts in systemd now expect an argument of
3154 "infinity" to turn them off, instead of "0" as before. The semantics
3155 from now on is that a timeout of "0" means "now", and "infinity"
3156 means "never". To maintain backwards compatibility, "0" continues to
3157 turn off previously existing timeout settings.
3159 * "systemctl reload-or-try-restart" has been renamed to "systemctl
3160 try-reload-or-restart" to clarify what it actually does: the "try"
3161 logic applies to both reloading and restarting, not just restarting.
3162 The old name continues to be accepted for compatibility.
3164 * On boot-up, when PID 1 detects that the system clock is behind the
3165 release date of the systemd version in use, the clock is now set
3166 to the latter. Previously, this was already done in timesyncd, in order
3167 to avoid running with clocks set to the various clock epochs such as
3168 1902, 1938 or 1970. With this change the logic is now done in PID 1
3169 in addition to timesyncd during early boot-up, so that it is enforced
3170 before the first process is spawned by systemd. Note that the logic
3171 in timesyncd remains, as it is more comprehensive and ensures
3172 clock monotonicity by maintaining a persistent timestamp file in
3173 /var. Since /var is generally not available in earliest boot or the
3174 initrd, this part of the logic remains in timesyncd, and is not done
3177 * Support for tweaking details in net_cls.class_id through the
3178 NetClass= configuration directive has been removed, as the kernel
3179 people have decided to deprecate that controller in cgroup v2.
3180 Userspace tools such as nftables are moving over to setting rules
3181 that are specific to the full cgroup path of a task, which obsoletes
3182 these controllers anyway. The NetClass= directive is kept around for
3183 legacy compatibility reasons. For a more in-depth description of the
3184 kernel change, please refer to the respective upstream commit:
3186 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
3188 * A new service setting RuntimeMaxSec= has been added that may be used
3189 to specify a maximum runtime for a service. If the timeout is hit, the
3190 service is terminated and put into a failure state.
3192 * A new service setting AmbientCapabilities= has been added. It allows
3193 configuration of additional Linux process capabilities that are
3194 passed to the activated processes. This is only available on very
3197 * The process resource limit settings in service units may now be used
3198 to configure hard and soft limits individually.
3200 * The various libsystemd APIs such as sd-bus or sd-event now publicly
3201 expose support for gcc's __attribute__((cleanup())) C extension.
3202 Specifically, for many object destructor functions alternative
3203 versions have been added that have names suffixed with "p" and take a
3204 pointer to a pointer to the object to destroy, instead of just a
3205 pointer to the object itself. This is useful because these destructor
3206 functions may be used directly as parameters to the cleanup
3207 construct. Internally, systemd has been a heavy user of this GCC
3208 extension for a long time, and with this change similar support is
3209 now available to consumers of the library outside of systemd. Note
3210 that by using this extension in your sources compatibility with old
3211 and strictly ANSI compatible C compilers is lost. However, all gcc or
3212 LLVM versions of recent years support this extension.
3214 * Timer units gained support for a new setting RandomizedDelaySec= that
3215 allows configuring some additional randomized delay to the configured
3216 time. This is useful to spread out timer events to avoid load peaks in
3217 clusters or larger setups.
3219 * Calendar time specifications now support sub-second accuracy.
3221 * Socket units now support listening on SCTP and UDP-lite protocol
3224 * The sd-event API now comes with a full set of man pages.
3226 * Older versions of systemd contained experimental support for
3227 compressing journal files and coredumps with the LZ4 compressor that
3228 was not compatible with the lz4 binary (due to API limitations of the
3229 lz4 library). This support has been removed; only support for files
3230 compatible with the lz4 binary remains. This LZ4 logic is now
3231 officially supported and no longer considered experimental.
3233 * The dkr image import logic has been removed again from importd. dkr's
3234 micro-services focus doesn't fit into the machine image focus of
3235 importd, and quickly got out of date with the upstream dkr API.
3237 * Creation of the /run/lock/lockdev/ directory was dropped from
3238 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
3239 been available for many years. If you still need this, you need to
3240 create your own tmpfiles.d config file with:
3242 d /run/lock/lockdev 0775 root lock -
3244 * The settings StartLimitBurst=, StartLimitInterval=, StartLimitAction=
3245 and RebootArgument= have been moved from the [Service] section of
3246 unit files to [Unit], and they are now supported on all unit types,
3247 not just service units. Of course, systemd will continue to
3248 understand these settings also at the old location, in order to
3249 maintain compatibility.
3251 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
3252 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
3253 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
3254 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
3255 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
3256 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
3257 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
3258 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
3259 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
3260 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
3261 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
3262 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
3263 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
3264 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
3265 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
3266 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
3267 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
3268 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
3269 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
3271 — Berlin, 2016-02-11
3275 * A number of properties previously only settable in unit
3276 files are now also available as properties to set when
3277 creating transient units programmatically via the bus, as it
3278 is exposed with systemd-run's --property=
3279 setting. Specifically, these are: SyslogIdentifier=,
3280 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
3281 EnvironmentFile=, ReadWriteDirectories=,
3282 ReadOnlyDirectories=, InaccessibleDirectories=,
3283 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
3285 * When creating transient services via the bus API it is now
3286 possible to pass in a set of file descriptors to use as
3287 STDIN/STDOUT/STDERR for the invoked process.
3289 * Slice units may now be created transiently via the bus APIs,
3290 similar to the way service and scope units may already be
3291 created transiently.
3293 * Wherever systemd expects a calendar timestamp specification
3294 (like in journalctl's --since= and --until= switches) UTC
3295 timestamps are now supported. Timestamps suffixed with "UTC"
3296 are now considered to be in Universal Time Coordinated
3297 instead of the local timezone. Also, timestamps may now
3298 optionally be specified with sub-second accuracy. Both of
3299 these additions also apply to recurring calendar event
3300 specification, such as OnCalendar= in timer units.
3302 * journalctl gained a new "--sync" switch that asks the
3303 journal daemon to write all so far unwritten log messages to
3304 disk and sync the files, before returning.
3306 * systemd-tmpfiles learned two new line types "q" and "Q" that
3307 operate like "v", but also set up a basic btrfs quota
3308 hierarchy when used on a btrfs file system with quota
3311 * tmpfiles' "v", "q" and "Q" will now create a plain directory
3312 instead of a subvolume (even on a btrfs file system) if the
3313 root directory is a plain directory, and not a
3314 subvolume. This should simplify things with certain chroot()
3315 environments which are not aware of the concept of btrfs
3318 * systemd-detect-virt gained a new --chroot switch to detect
3319 whether execution takes place in a chroot() environment.
3321 * CPUAffinity= now takes CPU index ranges in addition to
3324 * The various memory-related resource limit settings (such as
3325 LimitAS=) now understand the usual K, M, G, ... suffixes to
3326 the base of 1024 (IEC). Similar, the time-related resource
3327 limit settings understand the usual min, h, day, ...
3330 * There's a new system.conf setting DefaultTasksMax= to
3331 control the default TasksMax= setting for services and
3332 scopes running on the system. (TasksMax= is the primary
3333 setting that exposes the "pids" cgroup controller on systemd
3334 and was introduced in the previous systemd release.) The
3335 setting now defaults to 512, which means services that are
3336 not explicitly configured otherwise will only be able to
3337 create 512 processes or threads at maximum, from this
3338 version on. Note that this means that thread- or
3339 process-heavy services might need to be reconfigured to set
3340 TasksMax= to a higher value. It is sufficient to set
3341 TasksMax= in these specific unit files to a higher value, or
3342 even "infinity". Similar, there's now a logind.conf setting
3343 UserTasksMax= that defaults to 4096 and limits the total
3344 number of processes or tasks each user may own
3345 concurrently. nspawn containers also have the TasksMax=
3346 value set by default now, to 8192. Note that all of this
3347 only has an effect if the "pids" cgroup controller is
3348 enabled in the kernel. The general benefit of these changes
3349 should be a more robust and safer system, that provides a
3350 certain amount of per-service fork() bomb protection.
3352 * systemd-nspawn gained the new --network-veth-extra= switch
3353 to define additional and arbitrarily-named virtual Ethernet
3354 links between the host and the container.
3356 * A new service execution setting PassEnvironment= has been
3357 added that allows importing select environment variables
3358 from PID1's environment block into the environment block of
3361 * Timer units gained support for a new RemainAfterElapse=
3362 setting which takes a boolean argument. It defaults to on,
3363 exposing behaviour unchanged to previous releases. If set to
3364 off, timer units are unloaded after they elapsed if they
3365 cannot elapse again. This is particularly useful for
3366 transient timer units, which shall not stay around longer
3367 than until they first elapse.
3369 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
3370 default now (the kernel default is 16). This is beneficial
3371 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
3372 allows substantially larger numbers of queued
3373 datagrams. This should increase the capability of systemd to
3374 parallelize boot-up, as logging and sd_notify() are unlikely
3375 to stall execution anymore. If you need to change the value
3376 from the new defaults, use the usual sysctl.d/ snippets.
3378 * The compression framing format used by the journal or
3379 coredump processing has changed to be in line with what the
3380 official LZ4 tools generate. LZ4 compression support in
3381 systemd was considered unsupported previously, as the format
3382 was not compatible with the normal tools. With this release
3383 this has changed now, and it is hence safe for downstream
3384 distributions to turn it on. While not compressing as well
3385 as the XZ, LZ4 is substantially faster, which makes
3386 it a good default choice for the compression logic in the
3387 journal and in coredump handling.
3389 * Any reference to /etc/mtab has been dropped from
3390 systemd. The file has been obsolete since a while, but
3391 systemd refused to work on systems where it was incorrectly
3392 set up (it should be a symlink or non-existent). Please make
3393 sure to update to util-linux 2.27.1 or newer in conjunction
3394 with this systemd release, which also drops any reference to
3395 /etc/mtab. If you maintain a distribution make sure that no
3396 software you package still references it, as this is a
3397 likely source of bugs. There's also a glibc bug pending,
3398 asking for removal of any reference to this obsolete file:
3400 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
3402 Note that only util-linux versions built with
3403 --enable-libmount-force-mountinfo are supported.
3405 * Support for the ".snapshot" unit type has been removed. This
3406 feature turned out to be little useful and little used, and
3407 has now been removed from the core and from systemctl.
3409 * The dependency types RequiresOverridable= and
3410 RequisiteOverridable= have been removed from systemd. They
3411 have been used only very sparingly to our knowledge and
3412 other options that provide a similar effect (such as
3413 systemctl --mode=ignore-dependencies) are much more useful
3414 and commonly used. Moreover, they were only half-way
3415 implemented as the option to control behaviour regarding
3416 these dependencies was never added to systemctl. By removing
3417 these dependency types the execution engine becomes a bit
3418 simpler. Unit files that use these dependencies should be
3419 changed to use the non-Overridable dependency types
3420 instead. In fact, when parsing unit files with these
3421 options, that's what systemd will automatically convert them
3422 too, but it will also warn, asking users to fix the unit
3423 files accordingly. Removal of these dependency types should
3424 only affect a negligible number of unit files in the wild.
3426 * Behaviour of networkd's IPForward= option changed
3427 (again). It will no longer maintain a per-interface setting,
3428 but propagate one way from interfaces where this is enabled
3429 to the global kernel setting. The global setting will be
3430 enabled when requested by a network that is set up, but
3431 never be disabled again. This change was made to make sure
3432 IPv4 and IPv6 behaviour regarding packet forwarding is
3433 similar (as the Linux IPv6 stack does not support
3434 per-interface control of this setting) and to minimize
3437 * In unit files the behaviour of %u, %U, %h, %s has
3438 changed. These specifiers will now unconditionally resolve
3439 to the various user database fields of the user that the
3440 systemd instance is running as, instead of the user
3441 configured in the specific unit via User=. Note that this
3442 effectively doesn't change much, as resolving of these
3443 specifiers was already turned off in the --system instance
3444 of systemd, as we cannot do NSS lookups from PID 1. In the
3445 --user instance of systemd these specifiers where correctly
3446 resolved, but hardly made any sense, since the user instance
3447 lacks privileges to do user switches anyway, and User= is
3448 hence useless. Moreover, even in the --user instance of
3449 systemd behaviour was awkward as it would only take settings
3450 from User= assignment placed before the specifier into
3451 account. In order to unify and simplify the logic around
3452 this the specifiers will now always resolve to the
3453 credentials of the user invoking the manager (which in case
3454 of PID 1 is the root user).
3456 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
3457 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
3458 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
3459 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
3460 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
3461 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
3462 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
3463 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
3464 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
3465 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
3466 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
3467 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
3468 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
3469 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
3472 — Berlin, 2015-11-18
3476 * systemd now depends on util-linux v2.27. More specifically,
3477 the newly added mount monitor feature in libmount now
3478 replaces systemd's former own implementation.
3480 * libmount mandates /etc/mtab not to be regular file, and
3481 systemd now enforces this condition at early boot.
3482 /etc/mtab has been deprecated and warned about for a very
3483 long time, so systems running systemd should already have
3484 stopped having this file around as anything else than a
3485 symlink to /proc/self/mounts.
3487 * Support for the "pids" cgroup controller has been added. It
3488 allows accounting the number of tasks in a cgroup and
3489 enforcing limits on it. This adds two new setting
3490 TasksAccounting= and TasksMax= to each unit, as well as a
3491 global option DefaultTasksAccounting=.
3493 * Support for the "net_cls" cgroup controller has been added.
3494 It allows assigning a net class ID to each task in the
3495 cgroup, which can then be used in firewall rules and traffic
3496 shaping configurations. Note that the kernel netfilter net
3497 class code does not currently work reliably for ingress
3498 packets on unestablished sockets.
3500 This adds a new config directive called NetClass= to CGroup
3501 enabled units. Allowed values are positive numbers for fixed
3502 assignments and "auto" for picking a free value
3505 * 'systemctl is-system-running' now returns 'offline' if the
3506 system is not booted with systemd. This command can now be
3507 used as a substitute for 'systemd-notify --booted'.
3509 * Watchdog timeouts have been increased to 3 minutes for all
3510 in-tree service files. Apparently, disk IO issues are more
3511 frequent than we hoped, and user reported >1 minute waiting
3514 * 'machine-id-commit' functionality has been merged into
3515 'machine-id-setup --commit'. The separate binary has been
3518 * The WorkingDirectory= directive in unit files may now be set
3519 to the special value '~'. In this case, the working
3520 directory is set to the home directory of the user
3521 configured in User=.
3523 * "machinectl shell" will now open the shell in the home
3524 directory of the selected user by default.
3526 * The CrashChVT= configuration file setting is renamed to
3527 CrashChangeVT=, following our usual logic of not
3528 abbreviating unnecessarily. The old directive is still
3529 supported for compat reasons. Also, this directive now takes
3530 an integer value between 1 and 63, or a boolean value. The
3531 formerly supported '-1' value for disabling stays around for
3534 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
3535 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
3536 RootDirectory= properties can now be set for transient
3539 * The systemd-analyze tool gained a new "set-log-target" verb
3540 to change the logging target the system manager logs to
3541 dynamically during runtime. This is similar to how
3542 "systemd-analyze set-log-level" already changes the log
3545 * In nspawn /sys is now mounted as tmpfs, with only a selected
3546 set of subdirectories mounted in from the real sysfs. This
3547 enhances security slightly, and is useful for ensuring user
3548 namespaces work correctly.
3550 * Support for USB FunctionFS activation has been added. This
3551 allows implementation of USB gadget services that are
3552 activated as soon as they are requested, so that they don't
3553 have to run continuously, similar to classic socket
3556 * The "systemctl exit" command now optionally takes an
3557 additional parameter that sets the exit code to return from
3558 the systemd manager when exiting. This is only relevant when
3559 running the systemd user instance, or when running the
3560 system instance in a container.
3562 * sd-bus gained the new API calls sd_bus_path_encode_many()
3563 and sd_bus_path_decode_many() that allow easy encoding and
3564 decoding of multiple identifier strings inside a D-Bus
3565 object path. Another new call sd_bus_default_flush_close()
3566 has been added to flush and close per-thread default
3569 * systemd-cgtop gained support for a -M/--machine= switch to
3570 show the control groups within a certain container only.
3572 * "systemctl kill" gained support for an optional --fail
3573 switch. If specified the requested operation will fail of no
3574 processes have been killed, because the unit had no
3575 processes attached, or similar.
3577 * A new systemd.crash_reboot=1 kernel command line option has
3578 been added that triggers a reboot after crashing. This can
3579 also be set through CrashReboot= in systemd.conf.
3581 * The RuntimeDirectory= setting now understands unit
3582 specifiers like %i or %f.
3584 * A new (still internal) library API sd-ipv4acd has been added,
3585 that implements address conflict detection for IPv4. It's
3586 based on code from sd-ipv4ll, and will be useful for
3587 detecting DHCP address conflicts.
3589 * File descriptors passed during socket activation may now be
3590 named. A new API sd_listen_fds_with_names() is added to
3591 access the names. The default names may be overridden,
3592 either in the .socket file using the FileDescriptorName=
3593 parameter, or by passing FDNAME= when storing the file
3594 descriptors using sd_notify().
3596 * systemd-networkd gained support for:
3598 - Setting the IPv6 Router Advertisement settings via
3599 IPv6AcceptRouterAdvertisements= in .network files.
3601 - Configuring the HelloTimeSec=, MaxAgeSec= and
3602 ForwardDelaySec= bridge parameters in .netdev files.
3604 - Configuring PreferredSource= for static routes in
3607 * The "ask-password" framework used to query for LUKS harddisk
3608 passwords or SSL passwords during boot gained support for
3609 caching passwords in the kernel keyring, if it is
3610 available. This makes sure that the user only has to type in
3611 a passphrase once if there are multiple objects to unlock
3612 with the same one. Previously, such password caching was
3613 available only when Plymouth was used; this moves the
3614 caching logic into the systemd codebase itself. The
3615 "systemd-ask-password" utility gained a new --keyname=
3616 switch to control which kernel keyring key to use for
3617 caching a password in. This functionality is also useful for
3618 enabling display managers such as gdm to automatically
3619 unlock the user's GNOME keyring if its passphrase, the
3620 user's password and the harddisk password are the same, if
3621 gdm-autologin is used.
3623 * When downloading tar or raw images using "machinectl
3624 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
3625 file is now also downloaded, if it is available and stored
3626 next to the image file.
3628 * Units of type ".socket" gained a new boolean setting
3629 Writable= which is only useful in conjunction with
3630 ListenSpecial=. If true, enables opening the specified
3631 special file in O_RDWR mode rather than O_RDONLY mode.
3633 * systemd-rfkill has been reworked to become a singleton
3634 service that is activated through /dev/rfkill on each rfkill
3635 state change and saves the settings to disk. This way,
3636 systemd-rfkill is now compatible with devices that exist
3637 only intermittendly, and even restores state if the previous
3638 system shutdown was abrupt rather than clean.
3640 * The journal daemon gained support for vacuuming old journal
3641 files controlled by the number of files that shall remain,
3642 in addition to the already existing control by size and by
3643 date. This is useful as journal interleaving performance
3644 degrades with too many separate journal files, and allows
3645 putting an effective limit on them. The new setting defaults
3646 to 100, but this may be changed by setting SystemMaxFiles=
3647 and RuntimeMaxFiles= in journald.conf. Also, the
3648 "journalctl" tool gained the new --vacuum-files= switch to
3649 manually vacuum journal files to leave only the specified
3650 number of files in place.
3652 * udev will now create /dev/disk/by-path links for ATA devices
3653 on kernels where that is supported.
3655 * Galician, Serbian, Turkish and Korean translations were added.
3657 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
3658 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
3659 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
3660 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
3661 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
3662 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
3663 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
3664 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
3665 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
3666 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
3667 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
3668 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
3669 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
3670 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
3671 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
3672 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
3673 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
3674 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
3676 — Berlin, 2015-10-07
3680 * The DHCP implementation of systemd-networkd gained a set of
3683 - The DHCP server now supports emitting DNS and NTP
3684 information. It may be enabled and configured via
3685 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
3686 and NTP information is enabled, but no servers are
3687 configured, the corresponding uplink information (if there
3688 is any) is propagated.
3690 - Server and client now support transmission and reception
3691 of timezone information. It can be configured via the
3692 newly introduced network options UseTimezone=,
3693 EmitTimezone=, and Timezone=. Transmission of timezone
3694 information is enabled between host and containers by
3695 default now: the container will change its local timezone
3696 to what the host has set.
3698 - Lease timeouts can now be configured via
3699 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
3701 - The DHCP server improved on the stability of
3702 leases. Clients are more likely to get the same lease
3703 information back, even if the server loses state.
3705 - The DHCP server supports two new configuration options to
3706 control the lease address pool metrics, PoolOffset= and
3709 * The encapsulation limit of tunnels in systemd-networkd may
3710 now be configured via 'EncapsulationLimit='. It allows
3711 modifying the maximum additional levels of encapsulation
3712 that are permitted to be prepended to a packet.
3714 * systemd now supports the concept of user buses replacing
3715 session buses, if used with dbus-1.10 (and enabled via dbus
3716 --enable-user-session). It previously only supported this on
3717 kdbus-enabled systems, and this release expands this to
3718 'dbus-daemon' systems.
3720 * systemd-networkd now supports predictable interface names
3723 * systemd now optionally supports the new Linux kernel
3724 "unified" control group hierarchy. If enabled via the kernel
3725 command-line option 'systemd.unified_cgroup_hierarchy=1',
3726 systemd will try to mount the unified cgroup hierarchy
3727 directly on /sys/fs/cgroup. If not enabled, or not
3728 available, systemd will fall back to the legacy cgroup
3729 hierarchy setup, as before. Host system and containers can
3730 mix and match legacy and unified hierarchies as they
3731 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
3732 environment variable to individually select the hierarchy to
3733 use for executed containers. By default, nspawn will use the
3734 unified hierarchy for the containers if the host uses the
3735 unified hierarchy, and the legacy hierarchy otherwise.
3736 Please note that at this point the unified hierarchy is an
3737 experimental kernel feature and is likely to change in one
3738 of the next kernel releases. Therefore, it should not be
3739 enabled by default in downstream distributions yet. The
3740 minimum required kernel version for the unified hierarchy to
3741 work is 4.2. Note that when the unified hierarchy is used
3742 for the first time delegated access to controllers is
3743 safe. Because of this systemd-nspawn containers will get
3744 access to controllers now, as will systemd user
3745 sessions. This means containers and user sessions may now
3746 manage their own resources, partitioning up what the system
3749 * A new special scope unit "init.scope" has been introduced
3750 that encapsulates PID 1 of the system. It may be used to
3751 determine resource usage and enforce resource limits on PID
3752 1 itself. PID 1 hence moved out of the root of the control
3755 * The cgtop tool gained support for filtering out kernel
3756 threads when counting tasks in a control group. Also, the
3757 count of processes is now recursively summed up by
3758 default. Two options -k and --recursive= have been added to
3759 revert to old behaviour. The tool has also been updated to
3760 work correctly in containers now.
3762 * systemd-nspawn's --bind= and --bind-ro= options have been
3763 extended to allow creation of non-recursive bind mounts.
3765 * libsystemd gained two new calls sd_pid_get_cgroup() and
3766 sd_peer_get_cgroup() which return the control group path of
3767 a process or peer of a connected AF_UNIX socket. This
3768 function call is particularly useful when implementing
3769 delegated subtrees support in the control group hierarchy.
3771 * The "sd-event" event loop API of libsystemd now supports
3772 correct dequeuing of real-time signals, without losing
3775 * When systemd requests a polkit decision when managing units it
3776 will now add additional fields to the request, including unit
3777 name and desired operation. This enables more powerful polkit
3778 policies, that make decisions depending on these parameters.
3780 * nspawn learnt support for .nspawn settings files, that may
3781 accompany the image files or directories of containers, and
3782 may contain additional settings for the container. This is
3783 an alternative to configuring container parameters via the
3784 nspawn command line.
3786 Contributions from: Cristian Rodríguez, Daniel Mack, David
3787 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
3788 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
3789 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
3790 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
3791 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
3792 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
3793 Andersen, Tom Gundersen, Torstein Husebø
3795 — Berlin, 2015-09-08
3799 * machinectl gained a new verb 'shell' which opens a fresh
3800 shell on the target container or the host. It is similar to
3801 the existing 'login' command of machinectl, but spawns the
3802 shell directly without prompting for username or
3803 password. The pseudo machine '.host' now refers to the local
3804 host and is used by default. Hence, 'machinectl shell' can
3805 be used as replacement for 'su -' which spawns a session as
3806 a fresh systemd unit in a way that is fully isolated from
3807 the originating session.
3809 * systemd-networkd learned to cope with private-zone DHCP
3810 options and allows other programs to query the values.
3812 * SELinux access control when enabling/disabling units is no
3813 longer enforced with this release. The previous implementation
3814 was incorrect, and a new corrected implementation is not yet
3815 available. As unit file operations are still protected via
3816 polkit and D-Bus policy this is not a security problem. Yet,
3817 distributions which care about optimal SELinux support should
3818 probably not stabilize on this release.
3820 * sd-bus gained support for matches of type "arg0has=", that
3821 test for membership of strings in string arrays sent in bus
3824 * systemd-resolved now dumps the contents of its DNS and LLMNR
3825 caches to the logs on reception of the SIGUSR1 signal. This
3826 is useful to debug DNS behaviour.
3828 * The coredumpctl tool gained a new --directory= option to
3829 operate on journal files in a specific directory.
3831 * "systemctl reboot" and related commands gained a new
3832 "--message=" option which may be used to set a free-text
3833 wall message when shutting down or rebooting the
3834 system. This message is also logged, which is useful for
3835 figuring out the reason for a reboot or shutdown a
3838 * The "systemd-resolve-host" tool's -i switch now takes
3839 network interface numbers as alternative to interface names.
3841 * A new unit file setting for services has been introduced:
3842 UtmpMode= allows configuration of how precisely systemd
3843 handles utmp and wtmp entries for the service if this is
3844 enabled. This allows writing services that appear similar to
3845 user sessions in the output of the "w", "who", "last" and
3848 * systemd-resolved will now locally synthesize DNS resource
3849 records for the "localhost" and "gateway" domains as well as
3850 the local hostname. This should ensure that clients querying
3851 RRs via resolved will get similar results as those going via
3852 NSS, if nss-myhostname is enabled.
3854 Contributions from: Alastair Hughes, Alex Crawford, Daniel
3855 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
3856 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
3857 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
3858 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
3859 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
3860 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
3861 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
3862 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
3863 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
3864 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
3865 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
3867 — Berlin, 2015-08-27
3871 * The systemd-efi-boot-generator functionality was merged into
3872 systemd-gpt-auto-generator.
3874 * systemd-networkd now supports Group Policy for vxlan
3875 devices. It can be enabled via the new boolean configuration
3876 option called 'GroupPolicyExtension='.
3878 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
3879 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
3880 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
3882 — Berlin, 2015-07-31
3886 * The python-systemd code has been removed from the systemd repository.
3887 A new repository has been created which accommodates the code from
3888 now on, and we kindly ask distributions to create a separate package
3889 for this: https://github.com/systemd/python-systemd
3891 * The systemd daemon will now reload its main configuration
3892 (/etc/systemd/system.conf) on daemon-reload.
3894 * sd-dhcp now exposes vendor specific extensions via
3895 sd_dhcp_lease_get_vendor_specific().
3897 * systemd-networkd gained a number of new configuration options.
3899 - A new boolean configuration option for TAP devices called
3900 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
3901 device, thus allowing to send and receive GSO packets.
3903 - A new tunnel configuration option called 'CopyDSCP='.
3904 If enabled, the DSCP field of ip6 tunnels is copied into the
3905 decapsulated packet.
3907 - A set of boolean bridge configuration options were added.
3908 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
3909 and 'UnicastFlood=' are now parsed by networkd and applied to the
3910 respective bridge link device via the respective IFLA_BRPORT_*
3913 - A new string configuration option to override the hostname sent
3914 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
3915 is true, networkd will use the configured hostname instead of the
3916 system hostname when sending DHCP requests.
3918 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
3919 networkd will configure the IPv6 flow-label of the tunnel device
3920 according to RFC2460.
3922 - The 'macvtap' virtual network devices are now supported, similar to
3923 the already supported 'macvlan' devices.
3925 * systemd-resolved now implements RFC5452 to improve resilience against
3926 cache poisoning. Additionally, source port randomization is enabled
3927 by default to further protect against DNS spoofing attacks.
3929 * nss-mymachines now supports translating UIDs and GIDs of running
3930 containers with user-namespaces enabled. If a container 'foo'
3931 translates a host uid 'UID' to the container uid 'TUID', then
3932 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
3933 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
3934 mapped as 'vg-foo-TGID'.
3936 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
3937 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
3938 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
3939 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
3940 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
3941 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
3942 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
3943 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
3944 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
3945 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
3947 — Berlin, 2015-07-29
3951 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
3952 There are no known issues with current sysfs, and udev does not need
3953 or should be used to work around such bugs.
3955 * udev does no longer enable USB HID power management. Several reports
3956 indicate, that some devices cannot handle that setting.
3958 * The udev accelerometer helper was removed. The functionality
3959 is now fully included in iio-sensor-proxy. But this means,
3960 older iio-sensor-proxy versions will no longer provide
3961 accelerometer/orientation data with this systemd version.
3962 Please upgrade iio-sensor-proxy to version 1.0.
3964 * networkd gained a new configuration option IPv6PrivacyExtensions=
3965 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
3966 for Stateless Address") on selected networks.
3968 * For the sake of fewer build-time dependencies and less code in the
3969 main repository, the python bindings are about to be removed in the
3970 next release. A new repository has been created which accommodates
3971 the code from now on, and we kindly ask distributions to create a
3972 separate package for this. The removal will take place in v223.
3974 https://github.com/systemd/python-systemd
3976 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
3977 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
3978 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
3979 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
3980 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
3981 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
3982 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
3983 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
3984 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
3985 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
3987 — Berlin, 2015-07-07
3991 * The sd-bus.h and sd-event.h APIs have now been declared
3992 stable and have been added to the official interface of
3993 libsystemd.so. sd-bus implements an alternative D-Bus client
3994 library, that is relatively easy to use, very efficient and
3995 supports both classic D-Bus as well as kdbus as transport
3996 backend. sd-event is a generic event loop abstraction that
3997 is built around Linux epoll, but adds features such as event
3998 prioritization or efficient timer handling. Both APIs are good
3999 choices for C programs looking for a bus and/or event loop
4000 implementation that is minimal and does not have to be
4001 portable to other kernels.
4003 * kdbus support is no longer compile-time optional. It is now
4004 always built-in. However, it can still be disabled at
4005 runtime using the kdbus=0 kernel command line setting, and
4006 that setting may be changed to default to off, by specifying
4007 --disable-kdbus at build-time. Note though that the kernel
4008 command line setting has no effect if the kdbus.ko kernel
4009 module is not installed, in which case kdbus is (obviously)
4010 also disabled. We encourage all downstream distributions to
4011 begin testing kdbus by adding it to the kernel images in the
4012 development distributions, and leaving kdbus support in
4015 * The minimal required util-linux version has been bumped to
4018 * Support for chkconfig (--enable-chkconfig) was removed in
4019 favor of calling an abstraction tool
4020 /lib/systemd/systemd-sysv-install. This needs to be
4021 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
4022 in README for details.
4024 * If there's a systemd unit and a SysV init script for the
4025 same service name, and the user executes "systemctl enable"
4026 for it (or a related call), then this will now enable both
4027 (or execute the related operation on both), not just the
4030 * The libudev API documentation has been converted from gtkdoc
4033 * gudev has been removed from the systemd tree, it is now an
4036 * The systemd-cgtop tool learnt a new --raw switch to generate
4037 "raw" (machine parsable) output.
4039 * networkd's IPForwarding= .network file setting learnt the
4040 new setting "kernel", which ensures that networkd does not
4041 change the IP forwarding sysctl from the default kernel
4044 * The systemd-logind bus API now exposes a new boolean
4045 property "Docked" that reports whether logind considers the
4046 system "docked", i.e. connected to a docking station or not.
4048 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
4049 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
4050 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
4051 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
4052 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
4053 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
4054 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
4055 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
4056 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
4057 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
4058 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
4059 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
4060 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
4061 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
4062 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
4063 Fink, Zbigniew Jędrzejewski-Szmek
4065 — Berlin, 2015-06-19
4069 * The gudev library has been extracted into a separate repository
4070 available at: https://git.gnome.org/browse/libgudev/
4071 It is now managed as part of the Gnome project. Distributions
4072 are recommended to pass --disable-gudev to systemd and use
4073 gudev from the Gnome project instead. gudev is still included
4074 in systemd, for now. It will be removed soon, though. Please
4075 also see the announcement-thread on systemd-devel:
4076 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
4078 * systemd now exposes a CPUUsageNSec= property for each
4079 service unit on the bus, that contains the overall consumed
4080 CPU time of a service (the sum of what each process of the
4081 service consumed). This value is only available if
4082 CPUAccounting= is turned on for a service, and is then shown
4083 in the "systemctl status" output.
4085 * Support for configuring alternative mappings of the old SysV
4086 runlevels to systemd targets has been removed. They are now
4087 hardcoded in a way that runlevels 2, 3, 4 all map to
4088 multi-user.target and 5 to graphical.target (which
4089 previously was already the default behaviour).
4091 * The auto-mounter logic gained support for mount point
4092 expiry, using a new TimeoutIdleSec= setting in .automount
4093 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
4095 * The EFI System Partition (ESP) as mounted to /boot by
4096 systemd-efi-boot-generator will now be unmounted
4097 automatically after 2 minutes of not being used. This should
4098 minimize the risk of ESP corruptions.
4100 * New /etc/fstab options x-systemd.requires= and
4101 x-systemd.requires-mounts-for= are now supported to express
4102 additional dependencies for mounts. This is useful for
4103 journalling file systems that support external journal
4104 devices or overlay file systems that require underlying file
4105 systems to be mounted.
4107 * systemd does not support direct live-upgrades (via systemctl
4108 daemon-reexec) from versions older than v44 anymore. As no
4109 distribution we are aware of shipped such old versions in a
4110 stable release this should not be problematic.
4112 * When systemd forks off a new per-connection service instance
4113 it will now set the $REMOTE_ADDR environment variable to the
4114 remote IP address, and $REMOTE_PORT environment variable to
4115 the remote IP port. This behaviour is similar to the
4116 corresponding environment variables defined by CGI.
4118 * systemd-networkd gained support for uplink failure
4119 detection. The BindCarrier= option allows binding interface
4120 configuration dynamically to the link sense of other
4121 interfaces. This is useful to achieve behaviour like in
4124 * systemd-networkd gained support for configuring the DHCP
4125 client identifier to use when requesting leases.
4127 * systemd-networkd now has a per-network UseNTP= option to
4128 configure whether NTP server information acquired via DHCP
4129 is passed on to services like systemd-timesyncd.
4131 * systemd-networkd gained support for vti6 tunnels.
4133 * Note that systemd-networkd manages the sysctl variable
4134 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
4135 it is configured for since v219. The variable controls IP
4136 forwarding, and is a per-interface alternative to the global
4137 /proc/sys/net/ipv[46]/ip_forward. This setting is
4138 configurable in the IPForward= option, which defaults to
4139 "no". This means if networkd is used for an interface it is
4140 no longer sufficient to set the global sysctl option to turn
4141 on IP forwarding! Instead, the .network file option
4142 IPForward= needs to be turned on! Note that the
4143 implementation of this behaviour was broken in v219 and has
4146 * Many bonding and vxlan options are now configurable in
4149 * systemd-nspawn gained a new --property= setting to set unit
4150 properties for the container scope. This is useful for
4151 setting resource parameters (e.g. "CPUShares=500") on
4152 containers started from the command line.
4154 * systemd-nspawn gained a new --private-users= switch to make
4155 use of user namespacing available on recent Linux kernels.
4157 * systemd-nspawn may now be called as part of a shell pipeline
4158 in which case the pipes used for stdin and stdout are passed
4159 directly to the process invoked in the container, without
4160 indirection via a pseudo tty.
4162 * systemd-nspawn gained a new switch to control the UNIX
4163 signal to use when killing the init process of the container
4166 * systemd-nspawn gained a new --overlay= switch for mounting
4167 overlay file systems into the container using the new kernel
4170 * When a container image is imported via systemd-importd and
4171 the host file system is not btrfs, a loopback block device
4172 file is created in /var/lib/machines.raw with a btrfs file
4173 system inside. It is then mounted to /var/lib/machines to
4174 enable btrfs features for container management. The loopback
4175 file and btrfs file system is grown as needed when container
4176 images are imported via systemd-importd.
4178 * systemd-machined/systemd-importd gained support for btrfs
4179 quota, to enforce container disk space limits on disk. This
4180 is exposed in "machinectl set-limit".
4182 * systemd-importd now can import containers from local .tar,
4183 .raw and .qcow2 images, and export them to .tar and .raw. It
4184 can also import dkr v2 images now from the network (on top
4187 * systemd-importd gained support for verifying downloaded
4188 images with gpg2 (previously only gpg1 was supported).
4190 * systemd-machined, systemd-logind, systemd: most bus calls are
4191 now accessible to unprivileged processes via polkit. Also,
4192 systemd-logind will now allow users to kill their own sessions
4193 without further privileges or authorization.
4195 * systemd-shutdownd has been removed. This service was
4196 previously responsible for implementing scheduled shutdowns
4197 as exposed in /usr/bin/shutdown's time parameter. This
4198 functionality has now been moved into systemd-logind and is
4199 accessible via a bus interface.
4201 * "systemctl reboot" gained a new switch --firmware-setup that
4202 can be used to reboot into the EFI firmware setup, if that
4203 is available. systemd-logind now exposes an API on the bus
4204 to trigger such reboots, in case graphical desktop UIs want
4205 to cover this functionality.
4207 * "systemctl enable", "systemctl disable" and "systemctl mask"
4208 now support a new "--now" switch. If specified the units
4209 that are enabled will also be started, and the ones
4210 disabled/masked also stopped.
4212 * The Gummiboot EFI boot loader tool has been merged into
4213 systemd, and renamed to "systemd-boot". The bootctl tool has been
4214 updated to support systemd-boot.
4216 * An EFI kernel stub has been added that may be used to create
4217 kernel EFI binaries that contain not only the actual kernel,
4218 but also an initrd, boot splash, command line and OS release
4219 information. This combined binary can then be signed as a
4220 single image, so that the firmware can verify it all in one
4221 step. systemd-boot has special support for EFI binaries created
4222 like this and can extract OS release information from them
4223 and show them in the boot menu. This functionality is useful
4224 to implement cryptographically verified boot schemes.
4226 * Optional support has been added to systemd-fsck to pass
4227 fsck's progress report to an AF_UNIX socket in the file
4230 * udev will no longer create device symlinks for all block
4231 devices by default. A blacklist for excluding special block
4232 devices from this logic has been turned into a whitelist
4233 that requires picking block devices explicitly that require
4236 * A new (currently still internal) API sd-device.h has been
4237 added to libsystemd. This modernized API is supposed to
4238 replace libudev eventually. In fact, already much of libudev
4239 is now just a wrapper around sd-device.h.
4241 * A new hwdb database for storing metadata about pointing
4242 stick devices has been added.
4244 * systemd-tmpfiles gained support for setting file attributes
4245 similar to the "chattr" tool with new 'h' and 'H' lines.
4247 * systemd-journald will no longer unconditionally set the
4248 btrfs NOCOW flag on new journal files. This is instead done
4249 with tmpfiles snippet using the new 'h' line type. This
4250 allows easy disabling of this logic, by masking the
4251 journal-nocow.conf tmpfiles file.
4253 * systemd-journald will now translate audit message types to
4254 human readable identifiers when writing them to the
4255 journal. This should improve readability of audit messages.
4257 * The LUKS logic gained support for the offset= and skip=
4258 options in /etc/crypttab, as previously implemented by
4261 * /usr/lib/os-release gained a new optional field VARIANT= for
4262 distributions that support multiple variants (such as a
4263 desktop edition, a server edition, ...)
4265 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
4266 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
4267 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
4268 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
4269 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
4270 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
4271 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
4272 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
4273 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
4274 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
4275 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
4276 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
4277 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
4278 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
4279 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
4280 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
4281 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
4282 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
4283 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
4284 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
4285 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
4286 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
4287 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
4288 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
4289 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
4290 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
4291 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
4293 — Berlin, 2015-05-22
4297 * Introduce a new API "sd-hwdb.h" for querying the hardware
4298 metadata database. With this minimal interface one can query
4299 and enumerate the udev hwdb, decoupled from the old libudev
4300 library. libudev's interface for this is now only a wrapper
4301 around sd-hwdb. A new tool systemd-hwdb has been added to
4302 interface with and update the database.
4304 * When any of systemd's tools copies files (for example due to
4305 tmpfiles' C lines) a btrfs reflink will attempted first,
4306 before bytewise copying is done.
4308 * systemd-nspawn gained a new --ephemeral switch. When
4309 specified a btrfs snapshot is taken of the container's root
4310 directory, and immediately removed when the container
4311 terminates again. Thus, a container can be started whose
4312 changes never alter the container's root directory, and are
4313 lost on container termination. This switch can also be used
4314 for starting a container off the root file system of the
4315 host without affecting the host OS. This switch is only
4316 available on btrfs file systems.
4318 * systemd-nspawn gained a new --template= switch. It takes the
4319 path to a container tree to use as template for the tree
4320 specified via --directory=, should that directory be
4321 missing. This allows instantiating containers dynamically,
4322 on first run. This switch is only available on btrfs file
4325 * When a .mount unit refers to a mount point on which multiple
4326 mounts are stacked, and the .mount unit is stopped all of
4327 the stacked mount points will now be unmounted until no
4328 mount point remains.
4330 * systemd now has an explicit notion of supported and
4331 unsupported unit types. Jobs enqueued for unsupported unit
4332 types will now fail with an "unsupported" error code. More
4333 specifically .swap, .automount and .device units are not
4334 supported in containers, .busname units are not supported on
4335 non-kdbus systems. .swap and .automount are also not
4336 supported if their respective kernel compile time options
4339 * machinectl gained support for two new "copy-from" and
4340 "copy-to" commands for copying files from a running
4341 container to the host or vice versa.
4343 * machinectl gained support for a new "bind" command to bind
4344 mount host directories into local containers. This is
4345 currently only supported for nspawn containers.
4347 * networkd gained support for configuring bridge forwarding
4348 database entries (fdb) from .network files.
4350 * A new tiny daemon "systemd-importd" has been added that can
4351 download container images in tar, raw, qcow2 or dkr formats,
4352 and make them available locally in /var/lib/machines, so
4353 that they can run as nspawn containers. The daemon can GPG
4354 verify the downloads (not supported for dkr, since it has no
4355 provisions for verifying downloads). It will transparently
4356 decompress bz2, xz, gzip compressed downloads if necessary,
4357 and restore sparse files on disk. The daemon uses privilege
4358 separation to ensure the actual download logic runs with
4359 fewer privileges than the daemon itself. machinectl has
4360 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
4361 make the functionality of importd available to the
4362 user. With this in place the Fedora and Ubuntu "Cloud"
4363 images can be downloaded and booted as containers unmodified
4364 (the Fedora images lack the appropriate GPG signature files
4365 currently, so they cannot be verified, but this will change
4366 soon, hopefully). Note that downloading images is currently
4367 only fully supported on btrfs.
4369 * machinectl is now able to list container images found in
4370 /var/lib/machines, along with some metadata about sizes of
4371 disk and similar. If the directory is located on btrfs and
4372 quota is enabled, this includes quota display. A new command
4373 "image-status" has been added that shows additional
4374 information about images.
4376 * machinectl is now able to clone container images
4377 efficiently, if the underlying file system (btrfs) supports
4378 it, with the new "machinectl clone" command. It also
4379 gained commands for renaming and removing images, as well as
4380 marking them read-only or read-write (supported also on
4381 legacy file systems).
4383 * networkd gained support for collecting LLDP network
4384 announcements, from hardware that supports this. This is
4385 shown in networkctl output.
4387 * systemd-run gained support for a new -t (--pty) switch for
4388 invoking a binary on a pty whose input and output is
4389 connected to the invoking terminal. This allows executing
4390 processes as system services while interactively
4391 communicating with them via the terminal. Most interestingly
4392 this is supported across container boundaries. Invoking
4393 "systemd-run -t /bin/bash" is an alternative to running a
4394 full login session, the difference being that the former
4395 will not register a session, nor go through the PAM session
4398 * tmpfiles gained support for a new "v" line type for creating
4399 btrfs subvolumes. If the underlying file system is a legacy
4400 file system, this automatically degrades to creating a
4401 normal directory. Among others /var/lib/machines is now
4402 created like this at boot, should it be missing.
4404 * The directory /var/lib/containers/ has been deprecated and
4405 been replaced by /var/lib/machines. The term "machines" has
4406 been used in the systemd context as generic term for both
4407 VMs and containers, and hence appears more appropriate for
4408 this, as the directory can also contain raw images bootable
4411 * systemd-nspawn when invoked with -M but without --directory=
4412 or --image= is now capable of searching for the container
4413 root directory, subvolume or disk image automatically, in
4414 /var/lib/machines. systemd-nspawn@.service has been updated
4415 to make use of this, thus allowing it to be used for raw
4418 * A new machines.target unit has been introduced that is
4419 supposed to group all containers/VMs invoked as services on
4420 the system. systemd-nspawn@.service has been updated to
4421 integrate with that.
4423 * machinectl gained a new "start" command, for invoking a
4424 container as a service. "machinectl start foo" is mostly
4425 equivalent to "systemctl start systemd-nspawn@foo.service",
4426 but handles escaping in a nicer way.
4428 * systemd-nspawn will now mount most of the cgroupfs tree
4429 read-only into each container, with the exception of the
4430 container's own subtree in the name=systemd hierarchy.
4432 * journald now sets the special FS_NOCOW file flag for its
4433 journal files. This should improve performance on btrfs, by
4434 avoiding heavy fragmentation when journald's write-pattern
4435 is used on COW file systems. It degrades btrfs' data
4436 integrity guarantees for the files to the same levels as for
4437 ext3/ext4 however. This should be OK though as journald does
4438 its own data integrity checks and all its objects are
4439 checksummed on disk. Also, journald should handle btrfs disk
4440 full events a lot more gracefully now, by processing SIGBUS
4441 errors, and not relying on fallocate() anymore.
4443 * When journald detects that journal files it is writing to
4444 have been deleted it will immediately start new journal
4447 * systemd now provides a way to store file descriptors
4448 per-service in PID 1. This is useful for daemons to ensure
4449 that fds they require are not lost during a daemon
4450 restart. The fds are passed to the daemon on the next
4451 invocation in the same way socket activation fds are
4452 passed. This is now used by journald to ensure that the
4453 various sockets connected to all the system's stdout/stderr
4454 are not lost when journald is restarted. File descriptors
4455 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
4456 an extension to sd_notify(). Note that a limit is enforced
4457 on the number of fds a service can store in PID 1, and it
4458 defaults to 0, so that no fds may be stored, unless this is
4459 explicitly turned on.
4461 * The default TERM variable to use for units connected to a
4462 terminal, when no other value is explicitly is set is now
4463 vt220 rather than vt102. This should be fairly safe still,
4464 but allows PgUp/PgDn work.
4466 * The /etc/crypttab option header= as known from Debian is now
4469 * "loginctl user-status" and "loginctl session-status" will
4470 now show the last 10 lines of log messages of the
4471 user/session following the status output. Similar,
4472 "machinectl status" will show the last 10 log lines
4473 associated with a virtual machine or container
4474 service. (Note that this is usually not the log messages
4475 done in the VM/container itself, but simply what the
4476 container manager logs. For nspawn this includes all console
4479 * "loginctl session-status" without further argument will now
4480 show the status of the session of the caller. Similar,
4481 "lock-session", "unlock-session", "activate",
4482 "enable-linger", "disable-linger" may now be called without
4483 session/user parameter in which case they apply to the
4484 caller's session/user.
4486 * An X11 session scriptlet is now shipped that uploads
4487 $DISPLAY and $XAUTHORITY into the environment of the systemd
4488 --user daemon if a session begins. This should improve
4489 compatibility with X11 enabled applications run as systemd
4492 * Generators are now subject to masking via /etc and /run, the
4493 same way as unit files.
4495 * networkd .network files gained support for configuring
4496 per-link IPv4/IPv6 packet forwarding as well as IPv4
4497 masquerading. This is by default turned on for veth links to
4498 containers, as registered by systemd-nspawn. This means that
4499 nspawn containers run with --network-veth will now get
4500 automatic routed access to the host's networks without any
4501 further configuration or setup, as long as networkd runs on
4504 * systemd-nspawn gained the --port= (-p) switch to expose TCP
4505 or UDP posts of a container on the host. With this in place
4506 it is possible to run containers with private veth links
4507 (--network-veth), and have their functionality exposed on
4508 the host as if their services were running directly on the
4511 * systemd-nspawn's --network-veth switch now gained a short
4512 version "-n", since with the changes above it is now truly
4513 useful out-of-the-box. The systemd-nspawn@.service has been
4514 updated to make use of it too by default.
4516 * systemd-nspawn will now maintain a per-image R/W lock, to
4517 ensure that the same image is not started more than once
4518 writable. (It's OK to run an image multiple times
4519 simultaneously in read-only mode.)
4521 * systemd-nspawn's --image= option is now capable of
4522 dissecting and booting MBR and GPT disk images that contain
4523 only a single active Linux partition. Previously it
4524 supported only GPT disk images with proper GPT type
4525 IDs. This allows running cloud images from major
4526 distributions directly with systemd-nspawn, without
4529 * In addition to collecting mouse dpi data in the udev
4530 hardware database, there's now support for collecting angle
4531 information for mouse scroll wheels. The database is
4532 supposed to guarantee similar scrolling behavior on mice
4533 that it knows about. There's also support for collecting
4534 information about Touchpad types.
4536 * udev's input_id built-in will now also collect touch screen
4537 dimension data and attach it to probed devices.
4539 * /etc/os-release gained support for a Distribution Privacy
4542 * networkd gained support for creating "ipvlan", "gretap",
4543 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
4545 * systemd-tmpfiles gained support for "a" lines for setting
4548 * systemd-nspawn will now mount /tmp in the container to
4549 tmpfs, automatically.
4551 * systemd now exposes the memory.usage_in_bytes cgroup
4552 attribute and shows it for each service in the "systemctl
4553 status" output, if available.
4555 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
4556 immediate reboot is triggered. This useful if shutdown is
4557 hung and is unable to complete, to expedite the
4558 operation. Note that this kind of reboot will still unmount
4559 all file systems, and hence should not result in fsck being
4562 * A .device unit for an optical block device will now be
4563 considered active only when a medium is in the drive. Also,
4564 mount units are now bound to their backing devices thus
4565 triggering automatic unmounting when devices become
4566 unavailable. With this in place systemd will now
4567 automatically unmount left-over mounts when a CD-ROM is
4568 ejected or an USB stick is yanked from the system.
4570 * networkd-wait-online now has support for waiting for
4571 specific interfaces only (with globbing), and for giving up
4572 after a configurable timeout.
4574 * networkd now exits when idle. It will be automatically
4575 restarted as soon as interfaces show up, are removed or
4576 change state. networkd will stay around as long as there is
4577 at least one DHCP state machine or similar around, that keep
4580 * networkd may now configure IPv6 link-local addressing in
4581 addition to IPv4 link-local addressing.
4583 * The IPv6 "token" for use in SLAAC may now be configured for
4584 each .network interface in networkd.
4586 * Routes configured with networkd may now be assigned a scope
4589 * networkd's [Match] sections now support globbing and lists
4590 of multiple space-separated matches per item.
4592 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
4593 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
4594 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
4595 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
4596 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
4597 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
4598 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
4599 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
4600 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
4601 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
4602 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
4603 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
4604 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
4605 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
4606 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
4607 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
4608 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
4609 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
4610 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
4611 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
4612 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
4613 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
4614 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
4615 Hoffmann, Zbigniew Jędrzejewski-Szmek
4617 — Berlin, 2015-02-16
4621 * When querying unit file enablement status (for example via
4622 "systemctl is-enabled"), a new state "indirect" is now known
4623 which indicates that a unit might not be enabled itself, but
4624 another unit listed in its Also= setting might be.
4626 * Similar to the various existing ConditionXYZ= settings for
4627 units, there are now matching AssertXYZ= settings. While
4628 failing conditions cause a unit to be skipped, but its job
4629 to succeed, failing assertions declared like this will cause
4630 a unit start operation and its job to fail.
4632 * hostnamed now knows a new chassis type "embedded".
4634 * systemctl gained a new "edit" command. When used on a unit
4635 file, this allows extending unit files with .d/ drop-in
4636 configuration snippets or editing the full file (after
4637 copying it from /usr/lib to /etc). This will invoke the
4638 user's editor (as configured with $EDITOR), and reload the
4639 modified configuration after editing.
4641 * "systemctl status" now shows the suggested enablement state
4642 for a unit, as declared in the (usually vendor-supplied)
4643 system preset files.
4645 * nss-myhostname will now resolve the single-label host name
4646 "gateway" to the locally configured default IP routing
4647 gateways, ordered by their metrics. This assigns a stable
4648 name to the used gateways, regardless which ones are
4649 currently configured. Note that the name will only be
4650 resolved after all other name sources (if nss-myhostname is
4651 configured properly) and should hence not negatively impact
4652 systems that use the single-label host name "gateway" in
4655 * systemd-inhibit now allows filtering by mode when listing
4658 * Scope and service units gained a new "Delegate" boolean
4659 property, which, when set, allows processes running inside the
4660 unit to further partition resources. This is primarily
4661 useful for systemd user instances as well as container
4664 * journald will now pick up audit messages directly from
4665 the kernel, and log them like any other log message. The
4666 audit fields are split up and fully indexed. This means that
4667 journalctl in many ways is now a (nicer!) alternative to
4668 ausearch, the traditional audit client. Note that this
4669 implements only a minimal audit client. If you want the
4670 special audit modes like reboot-on-log-overflow, please use
4671 the traditional auditd instead, which can be used in
4672 parallel to journald.
4674 * The ConditionSecurity= unit file option now understands the
4675 special string "audit" to check whether auditing is
4678 * journalctl gained two new commands --vacuum-size= and
4679 --vacuum-time= to delete old journal files until the
4680 remaining ones take up no more than the specified size on disk,
4681 or are not older than the specified time.
4683 * A new, native PPPoE library has been added to sd-network,
4684 systemd's library of light-weight networking protocols. This
4685 library will be used in a future version of networkd to
4686 enable PPPoE communication without an external pppd daemon.
4688 * The busctl tool now understands a new "capture" verb that
4689 works similar to "monitor", but writes a packet capture
4690 trace to STDOUT that can be redirected to a file which is
4691 compatible with libcap's capture file format. This can then
4692 be loaded in Wireshark and similar tools to inspect bus
4695 * The busctl tool now understands a new "tree" verb that shows
4696 the object trees of a specific service on the bus, or of all
4699 * The busctl tool now understands a new "introspect" verb that
4700 shows all interfaces and members of objects on the bus,
4701 including their signature and values. This is particularly
4702 useful to get more information about bus objects shown by
4703 the new "busctl tree" command.
4705 * The busctl tool now understands new verbs "call",
4706 "set-property" and "get-property" for invoking bus method
4707 calls, setting and getting bus object properties in a
4710 * busctl gained a new --augment-creds= argument that controls
4711 whether the tool shall augment credential information it
4712 gets from the bus with data from /proc, in a possibly
4715 * nspawn's --link-journal= switch gained two new values
4716 "try-guest" and "try-host" that work like "guest" and
4717 "host", but do not fail if the host has no persistent
4718 journalling enabled. -j is now equivalent to
4719 --link-journal=try-guest.
4721 * macvlan network devices created by nspawn will now have
4722 stable MAC addresses.
4724 * A new SmackProcessLabel= unit setting has been added, which
4725 controls the SMACK security label processes forked off by
4726 the respective unit shall use.
4728 * If compiled with --enable-xkbcommon, systemd-localed will
4729 verify x11 keymap settings by compiling the given keymap. It
4730 will spew out warnings if the compilation fails. This
4731 requires libxkbcommon to be installed.
4733 * When a coredump is collected, a larger number of metadata
4734 fields is now collected and included in the journal records
4735 created for it. More specifically, control group membership,
4736 environment variables, memory maps, working directory,
4737 chroot directory, /proc/$PID/status, and a list of open file
4738 descriptors is now stored in the log entry.
4740 * The udev hwdb now contains DPI information for mice. For
4743 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
4745 * All systemd programs that read standalone configuration
4746 files in /etc now also support a corresponding series of
4747 .conf.d configuration directories in /etc/, /run/,
4748 /usr/local/lib/, /usr/lib/, and (if configured with
4749 --enable-split-usr) /lib/. In particular, the following
4750 configuration files now have corresponding configuration
4751 directories: system.conf user.conf, logind.conf,
4752 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
4753 resolved.conf, timesyncd.conf, journal-remote.conf, and
4754 journal-upload.conf. Note that distributions should use the
4755 configuration directories in /usr/lib/; the directories in
4756 /etc/ are reserved for the system administrator.
4758 * systemd-rfkill will no longer take the rfkill device name
4759 into account when storing rfkill state on disk, as the name
4760 might be dynamically assigned and not stable. Instead, the
4761 ID_PATH udev variable combined with the rfkill type (wlan,
4762 bluetooth, ...) is used.
4764 * A new service systemd-machine-id-commit.service has been
4765 added. When used on systems where /etc is read-only during
4766 boot, and /etc/machine-id is not initialized (but an empty
4767 file), this service will copy the temporary machine ID
4768 created as replacement into /etc after the system is fully
4769 booted up. This is useful for systems that are freshly
4770 installed with a non-initialized machine ID, but should get
4771 a fixed machine ID for subsequent boots.
4773 * networkd's .netdev files now provide a large set of
4774 configuration parameters for VXLAN devices. Similarly, the
4775 bridge port cost parameter is now configurable in .network
4776 files. There's also new support for configuring IP source
4777 routing. networkd .link files gained support for a new
4778 OriginalName= match that is useful to match against the
4779 original interface name the kernel assigned. .network files
4780 may include MTU= and MACAddress= fields for altering the MTU
4781 and MAC address while being connected to a specific network
4784 * The LUKS logic gained supported for configuring
4785 UUID-specific key files. There's also new support for naming
4786 LUKS device from the kernel command line, using the new
4787 luks.name= argument.
4789 * Timer units may now be transiently created via the bus API
4790 (this was previously already available for scope and service
4791 units). In addition it is now possible to create multiple
4792 transient units at the same time with a single bus call. The
4793 "systemd-run" tool has been updated to make use of this for
4794 running commands on a specified time, in at(1)-style.
4796 * tmpfiles gained support for "t" lines, for assigning
4797 extended attributes to files. Among other uses this may be
4798 used to assign SMACK labels to files.
4800 Contributions from: Alin Rauta, Alison Chaiken, Andrej
4801 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
4802 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
4803 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
4804 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
4805 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
4806 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
4807 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
4808 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
4809 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
4810 Michael Biebl, Michael Chapman, Michael Marineau, Michal
4811 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
4812 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
4813 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
4814 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
4815 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
4816 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
4817 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
4819 — Berlin, 2014-12-10
4823 * journalctl gained the new options -t/--identifier= to match
4824 on the syslog identifier (aka "tag"), as well as --utc to
4825 show log timestamps in the UTC timezone. journalctl now also
4826 accepts -n/--lines=all to disable line capping in a pager.
4828 * journalctl gained a new switch, --flush, that synchronously
4829 flushes logs from /run/log/journal to /var/log/journal if
4830 persistent storage is enabled. systemd-journal-flush.service
4831 now waits until the operation is complete.
4833 * Services can notify the manager before they start a reload
4834 (by sending RELOADING=1) or shutdown (by sending
4835 STOPPING=1). This allows the manager to track and show the
4836 internal state of daemons and closes a race condition when
4837 the process is still running but has closed its D-Bus
4840 * Services with Type=oneshot do not have to have any ExecStart
4843 * User units are now loaded also from
4844 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
4845 /run/systemd/user directory that was already previously
4846 supported, but is under the control of the user.
4848 * Job timeouts (i.e. time-outs on the time a job that is
4849 queued stays in the run queue) can now optionally result in
4850 immediate reboot or power-off actions (JobTimeoutAction= and
4851 JobTimeoutRebootArgument=). This is useful on ".target"
4852 units, to limit the maximum time a target remains
4853 undispatched in the run queue, and to trigger an emergency
4854 operation in such a case. This is now used by default to
4855 turn off the system if boot-up (as defined by everything in
4856 basic.target) hangs and does not complete for at least
4857 15min. Also, if power-off or reboot hang for at least 30min
4858 an immediate power-off/reboot operation is triggered. This
4859 functionality is particularly useful to increase reliability
4860 on embedded devices, but also on laptops which might
4861 accidentally get powered on when carried in a backpack and
4862 whose boot stays stuck in a hard disk encryption passphrase
4865 * systemd-logind can be configured to also handle lid switch
4866 events even when the machine is docked or multiple displays
4867 are attached (HandleLidSwitchDocked= option).
4869 * A helper binary and a service have been added which can be
4870 used to resume from hibernation in the initramfs. A
4871 generator will parse the resume= option on the kernel
4872 command line to trigger resume.
4874 * A user console daemon systemd-consoled has been
4875 added. Currently, it is a preview, and will so far open a
4876 single terminal on each session of the user marked as
4877 Desktop=systemd-console.
4879 * Route metrics can be specified for DHCP routes added by
4882 * The SELinux context of socket-activated services can be set
4883 from the information provided by the networking stack
4884 (SELinuxContextFromNet= option).
4886 * Userspace firmware loading support has been removed and
4887 the minimum supported kernel version is thus bumped to 3.7.
4889 * Timeout for udev workers has been increased from 1 to 3
4890 minutes, but a warning will be printed after 1 minute to
4891 help diagnose kernel modules that take a long time to load.
4893 * Udev rules can now remove tags on devices with TAG-="foobar".
4895 * systemd's readahead implementation has been removed. In many
4896 circumstances it didn't give expected benefits even for
4897 rotational disk drives and was becoming less relevant in the
4898 age of SSDs. As none of the developers has been using
4899 rotating media anymore, and nobody stepped up to actively
4900 maintain this component of systemd it has now been removed.
4902 * Swap units can use Options= to specify discard options.
4903 Discard options specified for swaps in /etc/fstab are now
4906 * Docker containers are now detected as a separate type of
4909 * The Password Agent protocol gained support for queries where
4910 the user input is shown, useful e.g. for user names.
4911 systemd-ask-password gained a new --echo option to turn that
4914 * The default sysctl.d/ snippets will now set:
4916 net.core.default_qdisc = fq_codel
4918 This selects Fair Queuing Controlled Delay as the default
4919 queuing discipline for network interfaces. fq_codel helps
4920 fight the network bufferbloat problem. It is believed to be
4921 a good default with no tuning required for most workloads.
4922 Downstream distributions may override this choice. On 10Gbit
4923 servers that do not do forwarding, "fq" may perform better.
4924 Systems without a good clocksource should use "pfifo_fast".
4926 * If kdbus is enabled during build a new option BusPolicy= is
4927 available for service units, that allows locking all service
4928 processes into a stricter bus policy, in order to limit
4929 access to various bus services, or even hide most of them
4930 from the service's view entirely.
4932 * networkctl will now show the .network and .link file
4933 networkd has applied to a specific interface.
4935 * sd-login gained a new API call sd_session_get_desktop() to
4936 query which desktop environment has been selected for a
4939 * UNIX utmp support is now compile-time optional to support
4940 legacy-free systems.
4942 * systemctl gained two new commands "add-wants" and
4943 "add-requires" for pulling in units from specific targets
4946 * If the word "rescue" is specified on the kernel command line
4947 the system will now boot into rescue mode (aka
4948 rescue.target), which was previously available only by
4949 specifying "1" or "systemd.unit=rescue.target" on the kernel
4950 command line. This new kernel command line option nicely
4951 mirrors the already existing "emergency" kernel command line
4954 * New kernel command line options mount.usr=, mount.usrflags=,
4955 mount.usrfstype= have been added that match root=, rootflags=,
4956 rootfstype= but allow mounting a specific file system to
4959 * The $NOTIFY_SOCKET is now also passed to control processes of
4960 services, not only the main process.
4962 * This version reenables support for fsck's -l switch. This
4963 means at least version v2.25 of util-linux is required for
4964 operation, otherwise dead-locks on device nodes may
4965 occur. Again: you need to update util-linux to at least
4966 v2.25 when updating systemd to v217.
4968 * The "multi-seat-x" tool has been removed from systemd, as
4969 its functionality has been integrated into X servers 1.16,
4970 and the tool is hence redundant. It is recommended to update
4971 display managers invoking this tool to simply invoke X
4972 directly from now on, again.
4974 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
4975 message flag has been added for all of systemd's polkit
4976 authenticated method calls has been added. In particular this
4977 now allows optional interactive authorization via polkit for
4978 many of PID1's privileged operations such as unit file
4979 enabling and disabling.
4981 * "udevadm hwdb --update" learnt a new switch "--usr" for
4982 placing the rebuilt hardware database in /usr instead of
4983 /etc. When used only hardware database entries stored in
4984 /usr will be used, and any user database entries in /etc are
4985 ignored. This functionality is useful for vendors to ship a
4986 pre-built database on systems where local configuration is
4987 unnecessary or unlikely.
4989 * Calendar time specifications in .timer units now also
4990 understand the strings "semi-annually", "quarterly" and
4991 "minutely" as shortcuts (in addition to the preexisting
4992 "anually", "hourly", ...).
4994 * systemd-tmpfiles will now correctly create files in /dev
4995 at boot which are marked for creation only at boot. It is
4996 recommended to always create static device nodes with 'c!'
4997 and 'b!', so that they are created only at boot and not
4998 overwritten at runtime.
5000 * When the watchdog logic is used for a service (WatchdogSec=)
5001 and the watchdog timeout is hit the service will now be
5002 terminated with SIGABRT (instead of just SIGTERM), in order
5003 to make sure a proper coredump and backtrace is
5004 generated. This ensures that hanging services will result in
5005 similar coredump/backtrace behaviour as services that hit a
5008 Contributions from: Andreas Henriksson, Andrei Borzenkov,
5009 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
5010 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
5011 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
5012 Herrmann, David Sommerseth, David Strauss, Emil Renner
5013 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
5014 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
5015 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
5016 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
5017 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
5018 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
5019 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
5020 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
5021 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
5022 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
5023 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
5024 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
5025 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
5026 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
5027 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
5028 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
5031 — Berlin, 2014-10-28
5035 * timedated no longer reads NTP implementation unit names from
5036 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
5037 implementations should add a
5039 Conflicts=systemd-timesyncd.service
5041 to their unit files to take over and replace systemd's NTP
5042 default functionality.
5044 * systemd-sysusers gained a new line type "r" for configuring
5045 which UID/GID ranges to allocate system users/groups
5046 from. Lines of type "u" may now add an additional column
5047 that specifies the home directory for the system user to be
5048 created. Also, systemd-sysusers may now optionally read user
5049 information from STDIN instead of a file. This is useful for
5050 invoking it from RPM preinst scriptlets that need to create
5051 users before the first RPM file is installed since these
5052 files might need to be owned by them. A new
5053 %sysusers_create_inline RPM macro has been introduced to do
5054 just that. systemd-sysusers now updates the shadow files as
5055 well as the user/group databases, which should enhance
5056 compatibility with certain tools like grpck.
5058 * A number of bus APIs of PID 1 now optionally consult polkit to
5059 permit access for otherwise unprivileged clients under certain
5060 conditions. Note that this currently doesn't support
5061 interactive authentication yet, but this is expected to be
5062 added eventually, too.
5064 * /etc/machine-info now has new fields for configuring the
5065 deployment environment of the machine, as well as the
5066 location of the machine. hostnamectl has been updated with
5067 new command to update these fields.
5069 * systemd-timesyncd has been updated to automatically acquire
5070 NTP server information from systemd-networkd, which might
5071 have been discovered via DHCP.
5073 * systemd-resolved now includes a caching DNS stub resolver
5074 and a complete LLMNR name resolution implementation. A new
5075 NSS module "nss-resolve" has been added which can be used
5076 instead of glibc's own "nss-dns" to resolve hostnames via
5077 systemd-resolved. Hostnames, addresses and arbitrary RRs may
5078 be resolved via systemd-resolved D-Bus APIs. In contrast to
5079 the glibc internal resolver systemd-resolved is aware of
5080 multi-homed system, and keeps DNS server and caches separate
5081 and per-interface. Queries are sent simultaneously on all
5082 interfaces that have DNS servers configured, in order to
5083 properly handle VPNs and local LANs which might resolve
5084 separate sets of domain names. systemd-resolved may acquire
5085 DNS server information from systemd-networkd automatically,
5086 which in turn might have discovered them via DHCP. A tool
5087 "systemd-resolve-host" has been added that may be used to
5088 query the DNS logic in resolved. systemd-resolved implements
5089 IDNA and automatically uses IDNA or UTF-8 encoding depending
5090 on whether classic DNS or LLMNR is used as transport. In the
5091 next releases we intend to add a DNSSEC and mDNS/DNS-SD
5092 implementation to systemd-resolved.
5094 * A new NSS module nss-mymachines has been added, that
5095 automatically resolves the names of all local registered
5096 containers to their respective IP addresses.
5098 * A new client tool "networkctl" for systemd-networkd has been
5099 added. It currently is entirely passive and will query
5100 networking configuration from udev, rtnetlink and networkd,
5101 and present it to the user in a very friendly
5102 way. Eventually, we hope to extend it to become a full
5103 control utility for networkd.
5105 * .socket units gained a new DeferAcceptSec= setting that
5106 controls the kernels' TCP_DEFER_ACCEPT sockopt for
5107 TCP. Similarly, support for controlling TCP keep-alive
5108 settings has been added (KeepAliveTimeSec=,
5109 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
5110 turning off Nagle's algorithm on TCP has been added
5113 * logind learned a new session type "web", for use in projects
5114 like Cockpit which register web clients as PAM sessions.
5116 * timer units with at least one OnCalendar= setting will now
5117 be started only after time-sync.target has been
5118 reached. This way they will not elapse before the system
5119 clock has been corrected by a local NTP client or
5120 similar. This is particular useful on RTC-less embedded
5121 machines, that come up with an invalid system clock.
5123 * systemd-nspawn's --network-veth= switch should now result in
5124 stable MAC addresses for both the outer and the inner side
5127 * systemd-nspawn gained a new --volatile= switch for running
5128 container instances with /etc or /var unpopulated.
5130 * The kdbus client code has been updated to use the new Linux
5131 3.17 memfd subsystem instead of the old kdbus-specific one.
5133 * systemd-networkd's DHCP client and server now support
5134 FORCERENEW. There are also new configuration options to
5135 configure the vendor client identifier and broadcast mode
5138 * systemd will no longer inform the kernel about the current
5139 timezone, as this is necessarily incorrect and racy as the
5140 kernel has no understanding of DST and similar
5141 concepts. This hence means FAT timestamps will be always
5142 considered UTC, similar to what Android is already
5143 doing. Also, when the RTC is configured to the local time
5144 (rather than UTC) systemd will never synchronize back to it,
5145 as this might confuse Windows at a later boot.
5147 * systemd-analyze gained a new command "verify" for offline
5148 validation of unit files.
5150 * systemd-networkd gained support for a couple of additional
5151 settings for bonding networking setups. Also, the metric for
5152 statically configured routes may now be configured. For
5153 network interfaces where this is appropriate the peer IP
5154 address may now be configured.
5156 * systemd-networkd's DHCP client will no longer request
5157 broadcasting by default, as this tripped up some networks.
5158 For hardware where broadcast is required the feature should
5159 be switched back on using RequestBroadcast=yes.
5161 * systemd-networkd will now set up IPv4LL addresses (when
5162 enabled) even if DHCP is configured successfully.
5164 * udev will now default to respect network device names given
5165 by the kernel when the kernel indicates that these are
5166 predictable. This behavior can be tweaked by changing
5167 NamePolicy= in the relevant .link file.
5169 * A new library systemd-terminal has been added that
5170 implements full TTY stream parsing and rendering. This
5171 library is supposed to be used later on for implementing a
5172 full userspace VT subsystem, replacing the current kernel
5175 * A new tool systemd-journal-upload has been added to push
5176 journal data to a remote system running
5177 systemd-journal-remote.
5179 * journald will no longer forward all local data to another
5180 running syslog daemon. This change has been made because
5181 rsyslog (which appears to be the most commonly used syslog
5182 implementation these days) no longer makes use of this, and
5183 instead pulls the data out of the journal on its own. Since
5184 forwarding the messages to a non-existent syslog server is
5185 more expensive than we assumed we have now turned this
5186 off. If you run a syslog server that is not a recent rsyslog
5187 version, you have to turn this option on again
5188 (ForwardToSyslog= in journald.conf).
5190 * journald now optionally supports the LZ4 compressor for
5191 larger journal fields. This compressor should perform much
5192 better than XZ which was the previous default.
5194 * machinectl now shows the IP addresses of local containers,
5195 if it knows them, plus the interface name of the container.
5197 * A new tool "systemd-escape" has been added that makes it
5198 easy to escape strings to build unit names and similar.
5200 * sd_notify() messages may now include a new ERRNO= field
5201 which is parsed and collected by systemd and shown among the
5202 "systemctl status" output for a service.
5204 * A new component "systemd-firstboot" has been added that
5205 queries the most basic systemd information (timezone,
5206 hostname, root password) interactively on first
5207 boot. Alternatively it may also be used to provision these
5208 things offline on OS images installed into directories.
5210 * The default sysctl.d/ snippets will now set
5212 net.ipv4.conf.default.promote_secondaries=1
5214 This has the benefit of no flushing secondary IP addresses
5215 when primary addresses are removed.
5217 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
5218 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
5219 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
5220 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
5221 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
5222 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
5223 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
5224 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
5225 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
5226 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
5227 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
5228 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
5229 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
5230 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
5231 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
5233 — Berlin, 2014-08-19
5237 * A new tool systemd-sysusers has been added. This tool
5238 creates system users and groups in /etc/passwd and
5239 /etc/group, based on static declarative system user/group
5240 definitions in /usr/lib/sysusers.d/. This is useful to
5241 enable factory resets and volatile systems that boot up with
5242 an empty /etc directory, and thus need system users and
5243 groups created during early boot. systemd now also ships
5244 with two default sysusers.d/ files for the most basic
5245 users and groups systemd and the core operating system
5248 * A new tmpfiles snippet has been added that rebuilds the
5249 essential files in /etc on boot, should they be missing.
5251 * A directive for ensuring automatic clean-up of
5252 /var/cache/man/ has been removed from the default
5253 configuration. This line should now be shipped by the man
5254 implementation. The necessary change has been made to the
5255 man-db implementation. Note that you need to update your man
5256 implementation to one that ships this line, otherwise no
5257 automatic clean-up of /var/cache/man will take place.
5259 * A new condition ConditionNeedsUpdate= has been added that
5260 may conditionalize services to only run when /etc or /var
5261 are "older" than the vendor operating system resources in
5262 /usr. This is useful for reconstructing or updating /etc
5263 after an offline update of /usr or a factory reset, on the
5264 next reboot. Services that want to run once after such an
5265 update or reset should use this condition and order
5266 themselves before the new systemd-update-done.service, which
5267 will mark the two directories as fully updated. A number of
5268 service files have been added making use of this, to rebuild
5269 the udev hardware database, the journald message catalog and
5270 dynamic loader cache (ldconfig). The systemd-sysusers tool
5271 described above also makes use of this now. With this in
5272 place it is now possible to start up a minimal operating
5273 system with /etc empty cleanly. For more information on the
5274 concepts involved see this recent blog story:
5276 http://0pointer.de/blog/projects/stateless.html
5278 * A new system group "input" has been introduced, and all
5279 input device nodes get this group assigned. This is useful
5280 for system-level software to get access to input devices. It
5281 complements what is already done for "audio" and "video".
5283 * systemd-networkd learnt minimal DHCPv4 server support in
5284 addition to the existing DHCPv4 client support. It also
5285 learnt DHCPv6 client and IPv6 Router Solicitation client
5286 support. The DHCPv4 client gained support for static routes
5287 passed in from the server. Note that the [DHCPv4] section
5288 known in older systemd-networkd versions has been renamed to
5289 [DHCP] and is now also used by the DHCPv6 client. Existing
5290 .network files using settings of this section should be
5291 updated, though compatibility is maintained. Optionally, the
5292 client hostname may now be sent to the DHCP server.
5294 * networkd gained support for vxlan virtual networks as well
5295 as tun/tap and dummy devices.
5297 * networkd gained support for automatic allocation of address
5298 ranges for interfaces from a system-wide pool of
5299 addresses. This is useful for dynamically managing a large
5300 number of interfaces with a single network configuration
5301 file. In particular this is useful to easily assign
5302 appropriate IP addresses to the veth links of a large number
5303 of nspawn instances.
5305 * RPM macros for processing sysusers, sysctl and binfmt
5306 drop-in snippets at package installation time have been
5309 * The /etc/os-release file should now be placed in
5310 /usr/lib/os-release. The old location is automatically
5311 created as symlink. /usr/lib is the more appropriate
5312 location of this file, since it shall actually describe the
5313 vendor operating system shipped in /usr, and not the
5314 configuration stored in /etc.
5316 * .mount units gained a new boolean SloppyOptions= setting
5317 that maps to mount(8)'s -s option which enables permissive
5318 parsing of unknown mount options.
5320 * tmpfiles learnt a new "L+" directive which creates a symlink
5321 but (unlike "L") deletes a pre-existing file first, should
5322 it already exist and not already be the correct
5323 symlink. Similarly, "b+", "c+" and "p+" directives have been
5324 added as well, which create block and character devices, as
5325 well as fifos in the filesystem, possibly removing any
5326 pre-existing files of different types.
5328 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
5329 'argument' field (which so far specified the source to
5330 symlink/copy the files from) is now optional. If omitted the
5331 same file os copied from /usr/share/factory/ suffixed by the
5332 full destination path. This is useful for populating /etc
5333 with essential files, by copying them from vendor defaults
5334 shipped in /usr/share/factory/etc.
5336 * A new command "systemctl preset-all" has been added that
5337 applies the service preset settings to all installed unit
5338 files. A new switch --preset-mode= has been added that
5339 controls whether only enable or only disable operations
5342 * A new command "systemctl is-system-running" has been added
5343 that allows checking the overall state of the system, for
5344 example whether it is fully up and running.
5346 * When the system boots up with an empty /etc, the equivalent
5347 to "systemctl preset-all" is executed during early boot, to
5348 make sure all default services are enabled after a factory
5351 * systemd now contains a minimal preset file that enables the
5352 most basic services systemd ships by default.
5354 * Unit files' [Install] section gained a new DefaultInstance=
5355 field for defining the default instance to create if a
5356 template unit is enabled with no instance specified.
5358 * A new passive target cryptsetup-pre.target has been added
5359 that may be used by services that need to make they run and
5360 finish before the first LUKS cryptographic device is set up.
5362 * The /dev/loop-control and /dev/btrfs-control device nodes
5363 are now owned by the "disk" group by default, opening up
5364 access to this group.
5366 * systemd-coredump will now automatically generate a
5367 stack trace of all core dumps taking place on the system,
5368 based on elfutils' libdw library. This stack trace is logged
5371 * systemd-coredump may now optionally store coredumps directly
5372 on disk (in /var/lib/systemd/coredump, possibly compressed),
5373 instead of storing them unconditionally in the journal. This
5374 mode is the new default. A new configuration file
5375 /etc/systemd/coredump.conf has been added to configure this
5376 and other parameters of systemd-coredump.
5378 * coredumpctl gained a new "info" verb to show details about a
5379 specific coredump. A new switch "-1" has also been added
5380 that makes sure to only show information about the most
5381 recent entry instead of all entries. Also, as the tool is
5382 generally useful now the "systemd-" prefix of the binary
5383 name has been removed. Distributions that want to maintain
5384 compatibility with the old name should add a symlink from
5385 the old name to the new name.
5387 * journald's SplitMode= now defaults to "uid". This makes sure
5388 that unprivileged users can access their own coredumps with
5389 coredumpctl without restrictions.
5391 * New kernel command line options "systemd.wants=" (for
5392 pulling an additional unit during boot), "systemd.mask="
5393 (for masking a specific unit for the boot), and
5394 "systemd.debug-shell" (for enabling the debug shell on tty9)
5395 have been added. This is implemented in the new generator
5396 "systemd-debug-generator".
5398 * systemd-nspawn will now by default filter a couple of
5399 syscalls for containers, among them those required for
5400 kernel module loading, direct x86 IO port access, swap
5401 management, and kexec. Most importantly though
5402 open_by_handle_at() is now prohibited for containers,
5403 closing a hole similar to a recently discussed vulnerability
5404 in docker regarding access to files on file hierarchies the
5405 container should normally not have access to. Note that, for
5406 nspawn, we generally make no security claims anyway (and
5407 this is explicitly documented in the man page), so this is
5408 just a fix for one of the most obvious problems.
5410 * A new man page file-hierarchy(7) has been added that
5411 contains a minimized, modernized version of the file system
5412 layout systemd expects, similar in style to the FHS
5413 specification or hier(5). A new tool systemd-path(1) has
5414 been added to query many of these paths for the local
5417 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
5418 longer done. Since the directory now has a per-user size
5419 limit, and is cleaned on logout this appears unnecessary,
5420 in particular since this now brings the lifecycle of this
5421 directory closer in line with how IPC objects are handled.
5423 * systemd.pc now exports a number of additional directories,
5424 including $libdir (which is useful to identify the library
5425 path for the primary architecture of the system), and a
5426 couple of drop-in directories.
5428 * udev's predictable network interface names now use the dev_port
5429 sysfs attribute, introduced in linux 3.15 instead of dev_id to
5430 distinguish between ports of the same PCI function. dev_id should
5431 only be used for ports using the same HW address, hence the need
5434 * machined has been updated to export the OS version of a
5435 container (read from /etc/os-release and
5436 /usr/lib/os-release) on the bus. This is now shown in
5437 "machinectl status" for a machine.
5439 * A new service setting RestartForceExitStatus= has been
5440 added. If configured to a set of exit signals or process
5441 return values, the service will be restarted when the main
5442 daemon process exits with any of them, regardless of the
5445 * systemctl's -H switch for connecting to remote systemd
5446 machines has been extended so that it may be used to
5447 directly connect to a specific container on the
5448 host. "systemctl -H root@foobar:waldi" will now connect as
5449 user "root" to host "foobar", and then proceed directly to
5450 the container named "waldi". Note that currently you have to
5451 authenticate as user "root" for this to work, as entering
5452 containers is a privileged operation.
5454 Contributions from: Andreas Henriksson, Benjamin Steinwender,
5455 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
5456 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
5457 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
5458 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
5459 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
5460 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
5461 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
5462 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
5463 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
5464 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
5465 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
5467 — Berlin, 2014-07-03
5471 * As an experimental feature, udev now tries to lock the
5472 disk device node (flock(LOCK_SH|LOCK_NB)) while it
5473 executes events for the disk or any of its partitions.
5474 Applications like partitioning programs can lock the
5475 disk device node (flock(LOCK_EX)) and claim temporary
5476 device ownership that way; udev will entirely skip all event
5477 handling for this disk and its partitions. If the disk
5478 was opened for writing, the close will trigger a partition
5479 table rescan in udev's "watch" facility, and if needed
5480 synthesize "change" events for the disk and all its partitions.
5481 This is now unconditionally enabled, and if it turns out to
5482 cause major problems, we might turn it on only for specific
5483 devices, or might need to disable it entirely. Device Mapper
5484 devices are excluded from this logic.
5486 * We temporarily dropped the "-l" switch for fsck invocations,
5487 since they collide with the flock() logic above. util-linux
5488 upstream has been changed already to avoid this conflict,
5489 and we will readd "-l" as soon as util-linux with this
5490 change has been released.
5492 * The dependency on libattr has been removed. Since a long
5493 time, the extended attribute calls have moved to glibc, and
5494 libattr is thus unnecessary.
5496 * Virtualization detection works without privileges now. This
5497 means the systemd-detect-virt binary no longer requires
5498 CAP_SYS_PTRACE file capabilities, and our daemons can run
5499 with fewer privileges.
5501 * systemd-networkd now runs under its own "systemd-network"
5502 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
5503 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
5504 loses the ability to write to files owned by root this way.
5506 * Similarly, systemd-resolved now runs under its own
5507 "systemd-resolve" user with no capabilities remaining.
5509 * Similarly, systemd-bus-proxyd now runs under its own
5510 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
5512 * systemd-networkd gained support for setting up "veth"
5513 virtual Ethernet devices for container connectivity, as well
5514 as GRE and VTI tunnels.
5516 * systemd-networkd will no longer automatically attempt to
5517 manually load kernel modules necessary for certain tunnel
5518 transports. Instead, it is assumed the kernel loads them
5519 automatically when required. This only works correctly on
5520 very new kernels. On older kernels, please consider adding
5521 the kernel modules to /etc/modules-load.d/ as a work-around.
5523 * The resolv.conf file systemd-resolved generates has been
5524 moved to /run/systemd/resolve/. If you have a symlink from
5525 /etc/resolv.conf, it might be necessary to correct it.
5527 * Two new service settings, ProtectHome= and ProtectSystem=,
5528 have been added. When enabled, they will make the user data
5529 (such as /home) inaccessible or read-only and the system
5530 (such as /usr) read-only, for specific services. This allows
5531 very light-weight per-service sandboxing to avoid
5532 modifications of user data or system files from
5533 services. These two new switches have been enabled for all
5534 of systemd's long-running services, where appropriate.
5536 * Socket units gained new SocketUser= and SocketGroup=
5537 settings to set the owner user and group of AF_UNIX sockets
5538 and FIFOs in the file system.
5540 * Socket units gained a new RemoveOnStop= setting. If enabled,
5541 all FIFOS and sockets in the file system will be removed
5542 when the specific socket unit is stopped.
5544 * Socket units gained a new Symlinks= setting. It takes a list
5545 of symlinks to create to file system sockets or FIFOs
5546 created by the specific Unix sockets. This is useful to
5547 manage symlinks to socket nodes with the same life-cycle as
5550 * The /dev/log socket and /dev/initctl FIFO have been moved to
5551 /run, and have been replaced by symlinks. This allows
5552 connecting to these facilities even if PrivateDevices=yes is
5553 used for a service (which makes /dev/log itself unavailable,
5554 but /run is left). This also has the benefit of ensuring
5555 that /dev only contains device nodes, directories and
5556 symlinks, and nothing else.
5558 * sd-daemon gained two new calls sd_pid_notify() and
5559 sd_pid_notifyf(). They are similar to sd_notify() and
5560 sd_notifyf(), but allow overriding of the source PID of
5561 notification messages if permissions permit this. This is
5562 useful to send notify messages on behalf of a different
5563 process (for example, the parent process). The
5564 systemd-notify tool has been updated to make use of this
5565 when sending messages (so that notification messages now
5566 originate from the shell script invoking systemd-notify and
5567 not the systemd-notify process itself. This should minimize
5568 a race where systemd fails to associate notification
5569 messages to services when the originating process already
5572 * A new "on-abnormal" setting for Restart= has been added. If
5573 set, it will result in automatic restarts on all "abnormal"
5574 reasons for a process to exit, which includes unclean
5575 signals, core dumps, timeouts and watchdog timeouts, but
5576 does not include clean and unclean exit codes or clean
5577 signals. Restart=on-abnormal is an alternative for
5578 Restart=on-failure for services that shall be able to
5579 terminate and avoid restarts on certain errors, by
5580 indicating so with an unclean exit code. Restart=on-failure
5581 or Restart=on-abnormal is now the recommended setting for
5582 all long-running services.
5584 * If the InaccessibleDirectories= service setting points to a
5585 mount point (or if there are any submounts contained within
5586 it), it is now attempted to completely unmount it, to make
5587 the file systems truly unavailable for the respective
5590 * The ReadOnlyDirectories= service setting and
5591 systemd-nspawn's --read-only parameter are now recursively
5592 applied to all submounts, too.
5594 * Mount units may now be created transiently via the bus APIs.
5596 * The support for SysV and LSB init scripts has been removed
5597 from the systemd daemon itself. Instead, it is now
5598 implemented as a generator that creates native systemd units
5599 from these scripts when needed. This enables us to remove a
5600 substantial amount of legacy code from PID 1, following the
5601 fact that many distributions only ship a very small number
5602 of LSB/SysV init scripts nowadays.
5604 * Privileged Xen (dom0) domains are not considered
5605 virtualization anymore by the virtualization detection
5606 logic. After all, they generally have unrestricted access to
5607 the hardware and usually are used to manage the unprivileged
5610 * systemd-tmpfiles gained a new "C" line type, for copying
5611 files or entire directories.
5613 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
5614 lines. So far, they have been non-globbing versions of the
5615 latter, and have thus been redundant. In future, it is
5616 recommended to only use "z". "m" has hence been removed
5617 from the documentation, even though it stays supported.
5619 * A tmpfiles snippet to recreate the most basic structure in
5620 /var has been added. This is enough to create the /var/run →
5621 /run symlink and create a couple of structural
5622 directories. This allows systems to boot up with an empty or
5623 volatile /var. Of course, while with this change, the core OS
5624 now is capable with dealing with a volatile /var, not all
5625 user services are ready for it. However, we hope that sooner
5626 or later, many service daemons will be changed upstream so
5627 that they are able to automatically create their necessary
5628 directories in /var at boot, should they be missing. This is
5629 the first step to allow state-less systems that only require
5630 the vendor image for /usr to boot.
5632 * systemd-nspawn has gained a new --tmpfs= switch to mount an
5633 empty tmpfs instance to a specific directory. This is
5634 particularly useful for making use of the automatic
5635 reconstruction of /var (see above), by passing --tmpfs=/var.
5637 * Access modes specified in tmpfiles snippets may now be
5638 prefixed with "~", which indicates that they shall be masked
5639 by whether the existing file or directory is currently
5640 writable, readable or executable at all. Also, if specified,
5641 the sgid/suid/sticky bits will be masked for all
5644 * A new passive target unit "network-pre.target" has been
5645 added which is useful for services that shall run before any
5646 network is configured, for example firewall scripts.
5648 * The "floppy" group that previously owned the /dev/fd*
5649 devices is no longer used. The "disk" group is now used
5650 instead. Distributions should probably deprecate usage of
5653 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
5654 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
5655 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
5656 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
5657 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
5658 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
5661 — Berlin, 2014-06-11
5665 * A new "systemd-timesyncd" daemon has been added for
5666 synchronizing the system clock across the network. It
5667 implements an SNTP client. In contrast to NTP
5668 implementations such as chrony or the NTP reference server,
5669 this only implements a client side, and does not bother with
5670 the full NTP complexity, focusing only on querying time from
5671 one remote server and synchronizing the local clock to
5672 it. Unless you intend to serve NTP to networked clients or
5673 want to connect to local hardware clocks, this simple NTP
5674 client should be more than appropriate for most
5675 installations. The daemon runs with minimal privileges, and
5676 has been hooked up with networkd to only operate when
5677 network connectivity is available. The daemon saves the
5678 current clock to disk every time a new NTP sync has been
5679 acquired, and uses this to possibly correct the system clock
5680 early at bootup, in order to accommodate for systems that
5681 lack an RTC such as the Raspberry Pi and embedded devices,
5682 and to make sure that time monotonically progresses on these
5683 systems, even if it is not always correct. To make use of
5684 this daemon, a new system user and group "systemd-timesync"
5685 needs to be created on installation of systemd.
5687 * The queue "seqnum" interface of libudev has been disabled, as
5688 it was generally incompatible with device namespacing as
5689 sequence numbers of devices go "missing" if the devices are
5690 part of a different namespace.
5692 * "systemctl list-timers" and "systemctl list-sockets" gained
5693 a --recursive switch for showing units of these types also
5694 for all local containers, similar in style to the already
5695 supported --recursive switch for "systemctl list-units".
5697 * A new RebootArgument= setting has been added for service
5698 units, which may be used to specify a kernel reboot argument
5699 to use when triggering reboots with StartLimitAction=.
5701 * A new FailureAction= setting has been added for service
5702 units which may be used to specify an operation to trigger
5703 when a service fails. This works similarly to
5704 StartLimitAction=, but unlike it, controls what is done
5705 immediately rather than only after several attempts to
5706 restart the service in question.
5708 * hostnamed got updated to also expose the kernel name,
5709 release, and version on the bus. This is useful for
5710 executing commands like hostnamectl with the -H switch.
5711 systemd-analyze makes use of this to properly display
5712 details when running non-locally.
5714 * The bootchart tool can now show cgroup information in the
5715 graphs it generates.
5717 * The CFS CPU quota cgroup attribute is now exposed for
5718 services. The new CPUQuota= switch has been added for this
5719 which takes a percentage value. Setting this will have the
5720 result that a service may never get more CPU time than the
5721 specified percentage, even if the machine is otherwise idle.
5723 * systemd-networkd learned IPIP and SIT tunnel support.
5725 * LSB init scripts exposing a dependency on $network will now
5726 get a dependency on network-online.target rather than simply
5727 network.target. This should bring LSB handling closer to
5728 what it was on SysV systems.
5730 * A new fsck.repair= kernel option has been added to control
5731 how fsck shall deal with unclean file systems at boot.
5733 * The (.ini) configuration file parser will now silently
5734 ignore sections whose name begins with "X-". This may be
5735 used to maintain application-specific extension sections in unit
5738 * machined gained a new API to query the IP addresses of
5739 registered containers. "machinectl status" has been updated
5740 to show these addresses in its output.
5742 * A new call sd_uid_get_display() has been added to the
5743 sd-login APIs for querying the "primary" session of a
5744 user. The "primary" session of the user is elected from the
5745 user's sessions and generally a graphical session is
5746 preferred over a text one.
5748 * A minimal systemd-resolved daemon has been added. It
5749 currently simply acts as a companion to systemd-networkd and
5750 manages resolv.conf based on per-interface DNS
5751 configuration, possibly supplied via DHCP. In the long run
5752 we hope to extend this into a local DNSSEC enabled DNS and
5755 * The systemd-networkd-wait-online tool is now enabled by
5756 default. It will delay network-online.target until a network
5757 connection has been configured. The tool primarily integrates
5758 with networkd, but will also make a best effort to make sense
5759 of network configuration performed in some other way.
5761 * Two new service options StartupCPUShares= and
5762 StartupBlockIOWeight= have been added that work similarly to
5763 CPUShares= and BlockIOWeight= however only apply during
5764 system startup. This is useful to prioritize certain services
5765 differently during bootup than during normal runtime.
5767 * hostnamed has been changed to prefer the statically
5768 configured hostname in /etc/hostname (unless set to
5769 'localhost' or empty) over any dynamic one supplied by
5770 dhcp. With this change, the rules for picking the hostname
5771 match more closely the rules of other configuration settings
5772 where the local administrator's configuration in /etc always
5773 overrides any other settings.
5775 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
5776 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
5777 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
5778 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
5779 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
5780 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
5781 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
5782 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
5783 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
5784 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
5785 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
5786 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
5787 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
5788 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
5789 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
5790 Lindskog, WaLyong Cho, Will Woods, Zbigniew
5793 — Beijing, 2014-05-28
5797 * When restoring the screen brightness at boot, stay away from
5798 the darkest setting or from the lowest 5% of the available
5799 range, depending on which is the larger value of both. This
5800 should effectively protect the user from rebooting into a
5801 black screen, should the brightness have been set to minimum
5804 * sd-login gained a new sd_machine_get_class() call to
5805 determine the class ("vm" or "container") of a machine
5806 registered with machined.
5808 * sd-login gained new calls
5809 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
5810 to query the identity of the peer of a local AF_UNIX
5811 connection. They operate similarly to their sd_pid_get_xyz()
5814 * PID 1 will now maintain a system-wide system state engine
5815 with the states "starting", "running", "degraded",
5816 "maintenance", "stopping". These states are bound to system
5817 startup, normal runtime, runtime with at least one failed
5818 service, rescue/emergency mode and system shutdown. This
5819 state is shown in the "systemctl status" output when no unit
5820 name is passed. It is useful to determine system state, in
5821 particularly when doing so for many systems or containers at
5824 * A new command "list-machines" has been added to "systemctl"
5825 that lists all local OS containers and shows their system
5826 state (see above), if systemd runs inside of them.
5828 * systemctl gained a new "-r" switch to recursively enumerate
5829 units on all local containers, when used with the
5830 "list-unit" command (which is the default one that is
5831 executed when no parameters are specified).
5833 * The GPT automatic partition discovery logic will now honour
5834 two GPT partition flags: one may be set on a partition to
5835 cause it to be mounted read-only, and the other may be set
5836 on a partition to ignore it during automatic discovery.
5838 * Two new GPT type UUIDs have been added for automatic root
5839 partition discovery, for 32-bit and 64-bit ARM. This is not
5840 particularly useful for discovering the root directory on
5841 these architectures during bare-metal boots (since UEFI is
5842 not common there), but still very useful to allow booting of
5843 ARM disk images in nspawn with the -i option.
5845 * MAC addresses of interfaces created with nspawn's
5846 --network-interface= switch will now be generated from the
5847 machine name, and thus be stable between multiple invocations
5850 * logind will now automatically remove all IPC objects owned
5851 by a user if she or he fully logs out. This makes sure that
5852 users who are logged out cannot continue to consume IPC
5853 resources. This covers SysV memory, semaphores and message
5854 queues as well as POSIX shared memory and message
5855 queues. Traditionally, SysV and POSIX IPC had no life-cycle
5856 limits. With this functionality, that is corrected. This may
5857 be turned off by using the RemoveIPC= switch of logind.conf.
5859 * The systemd-machine-id-setup and tmpfiles tools gained a
5860 --root= switch to operate on a specific root directory,
5863 * journald can now forward logged messages to the TTYs of all
5864 logged in users ("wall"). This is the default for all
5865 emergency messages now.
5867 * A new tool systemd-journal-remote has been added to stream
5868 journal log messages across the network.
5870 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
5871 controller trees are mounted into it. Note that the
5872 directories mounted beneath it are not read-only. This is a
5873 security measure and is particularly useful because glibc
5874 actually includes a search logic to pick any tmpfs it can
5875 find to implement shm_open() if /dev/shm is not available
5876 (which it might very well be in namespaced setups).
5878 * machinectl gained a new "poweroff" command to cleanly power
5879 down a local OS container.
5881 * The PrivateDevices= unit file setting will now also drop the
5882 CAP_MKNOD capability from the capability bound set, and
5883 imply DevicePolicy=closed.
5885 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
5886 comprehensively on all long-running systemd services where
5887 this is appropriate.
5889 * systemd-udevd will now run in a disassociated mount
5890 namespace. To mount directories from udev rules, make sure to
5891 pull in mount units via SYSTEMD_WANTS properties.
5893 * The kdbus support gained support for uploading policy into
5894 the kernel. sd-bus gained support for creating "monitoring"
5895 connections that can eavesdrop into all bus communication
5896 for debugging purposes.
5898 * Timestamps may now be specified in seconds since the UNIX
5899 epoch Jan 1st, 1970 by specifying "@" followed by the value
5902 * Native tcpwrap support in systemd has been removed. tcpwrap
5903 is old code, not really maintained anymore and has serious
5904 shortcomings, and better options such as firewalls
5905 exist. For setups that require tcpwrap usage, please
5906 consider invoking your socket-activated service via tcpd,
5907 like on traditional inetd.
5909 * A new system.conf configuration option
5910 DefaultTimerAccuracySec= has been added that controls the
5911 default AccuracySec= setting of .timer units.
5913 * Timer units gained a new WakeSystem= switch. If enabled,
5914 timers configured this way will cause the system to resume
5915 from system suspend (if the system supports that, which most
5918 * Timer units gained a new Persistent= switch. If enabled,
5919 timers configured this way will save to disk when they have
5920 been last triggered. This information is then used on next
5921 reboot to possible execute overdue timer events, that
5922 could not take place because the system was powered off.
5923 This enables simple anacron-like behaviour for timer units.
5925 * systemctl's "list-timers" will now also list the time a
5926 timer unit was last triggered in addition to the next time
5927 it will be triggered.
5929 * systemd-networkd will now assign predictable IPv4LL
5930 addresses to its local interfaces.
5932 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
5933 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
5934 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
5935 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
5936 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
5937 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
5938 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
5939 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
5942 — Berlin, 2014-03-25
5946 * A new unit file setting RestrictAddressFamilies= has been
5947 added to restrict which socket address families unit
5948 processes gain access to. This takes address family names
5949 like "AF_INET" or "AF_UNIX", and is useful to minimize the
5950 attack surface of services via exotic protocol stacks. This
5951 is built on seccomp system call filters.
5953 * Two new unit file settings RuntimeDirectory= and
5954 RuntimeDirectoryMode= have been added that may be used to
5955 manage a per-daemon runtime directories below /run. This is
5956 an alternative for setting up directory permissions with
5957 tmpfiles snippets, and has the advantage that the runtime
5958 directory's lifetime is bound to the daemon runtime and that
5959 the daemon starts up with an empty directory each time. This
5960 is particularly useful when writing services that drop
5961 privileges using the User= or Group= setting.
5963 * The DeviceAllow= unit setting now supports globbing for
5964 matching against device group names.
5966 * The systemd configuration file system.conf gained new
5967 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
5968 DefaultMemoryAccounting= to globally turn on/off accounting
5969 for specific resources (cgroups) for all units. These
5970 settings may still be overridden individually in each unit
5973 * systemd-gpt-auto-generator is now able to discover /srv and
5974 root partitions in addition to /home and swap partitions. It
5975 also supports LUKS-encrypted partitions now. With this in
5976 place, automatic discovery of partitions to mount following
5977 the Discoverable Partitions Specification
5978 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
5979 is now a lot more complete. This allows booting without
5980 /etc/fstab and without root= on the kernel command line on
5981 systems prepared appropriately.
5983 * systemd-nspawn gained a new --image= switch which allows
5984 booting up disk images and Linux installations on any block
5985 device that follow the Discoverable Partitions Specification
5986 (see above). This means that installations made with
5987 appropriately updated installers may now be started and
5988 deployed using container managers, completely
5989 unmodified. (We hope that libvirt-lxc will add support for
5990 this feature soon, too.)
5992 * systemd-nspawn gained a new --network-macvlan= setting to
5993 set up a private macvlan interface for the
5994 container. Similarly, systemd-networkd gained a new
5995 Kind=macvlan setting in .netdev files.
5997 * systemd-networkd now supports configuring local addresses
6000 * A new tool systemd-network-wait-online has been added to
6001 synchronously wait for network connectivity using
6004 * The sd-bus.h bus API gained a new sd_bus_track object for
6005 tracking the life-cycle of bus peers. Note that sd-bus.h is
6006 still not a public API though (unless you specify
6007 --enable-kdbus on the configure command line, which however
6008 voids your warranty and you get no API stability guarantee).
6010 * The $XDG_RUNTIME_DIR runtime directories for each user are
6011 now individual tmpfs instances, which has the benefit of
6012 introducing separate pools for each user, with individual
6013 size limits, and thus making sure that unprivileged clients
6014 can no longer negatively impact the system or other users by
6015 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
6016 RuntimeDirectorySize= has been introduced that allows
6017 controlling the default size limit for all users. It
6018 defaults to 10% of the available physical memory. This is no
6019 replacement for quotas on tmpfs though (which the kernel
6020 still does not support), as /dev/shm and /tmp are still
6021 shared resources used by both the system and unprivileged
6024 * logind will now automatically turn off automatic suspending
6025 on laptop lid close when more than one display is
6026 connected. This was previously expected to be implemented
6027 individually in desktop environments (such as GNOME),
6028 however has been added to logind now, in order to fix a
6029 boot-time race where a desktop environment might not have
6030 been started yet and thus not been able to take an inhibitor
6031 lock at the time where logind already suspends the system
6032 due to a closed lid.
6034 * logind will now wait at least 30s after each system
6035 suspend/resume cycle, and 3min after system boot before
6036 suspending the system due to a closed laptop lid. This
6037 should give USB docking stations and similar enough time to
6038 be probed and configured after system resume and boot in
6039 order to then act as suspend blocker.
6041 * systemd-run gained a new --property= setting which allows
6042 initialization of resource control properties (and others)
6043 for the created scope or service unit. Example: "systemd-run
6044 --property=BlockIOWeight=10 updatedb" may be used to run
6045 updatedb at a low block IO scheduling weight.
6047 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
6048 now also work in --scope mode.
6050 * When systemd is compiled with kdbus support, basic support
6051 for enforced policies is now in place. (Note that enabling
6052 kdbus still voids your warranty and no API compatibility
6055 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
6056 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
6057 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
6058 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
6059 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
6060 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
6061 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
6062 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
6063 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
6064 Zbigniew Jędrzejewski-Szmek
6066 — Berlin, 2014-03-12
6070 * systemd will now relabel /dev after loading the SMACK policy
6071 according to SMACK rules.
6073 * A new unit file option AppArmorProfile= has been added to
6074 set the AppArmor profile for the processes of a unit.
6076 * A new condition check ConditionArchitecture= has been added
6077 to conditionalize units based on the system architecture, as
6078 reported by uname()'s "machine" field.
6080 * systemd-networkd now supports matching on the system
6081 virtualization, architecture, kernel command line, host name
6084 * logind is now a lot more aggressive when suspending the
6085 machine due to a closed laptop lid. Instead of acting only
6086 on the lid close action, it will continuously watch the lid
6087 status and act on it. This is useful for laptops where the
6088 power button is on the outside of the chassis so that it can
6089 be reached without opening the lid (such as the Lenovo
6090 Yoga). On those machines, logind will now immediately
6091 re-suspend the machine if the power button has been
6092 accidentally pressed while the laptop was suspended and in a
6093 backpack or similar.
6095 * logind will now watch SW_DOCK switches and inhibit reaction
6096 to the lid switch if it is pressed. This means that logind
6097 will not suspend the machine anymore if the lid is closed
6098 and the system is docked, if the laptop supports SW_DOCK
6099 notifications via the input layer. Note that ACPI docking
6100 stations do not generate this currently. Also note that this
6101 logic is usually not fully sufficient and Desktop
6102 Environments should take a lid switch inhibitor lock when an
6103 external display is connected, as systemd will not watch
6106 * nspawn will now make use of the devices cgroup controller by
6107 default, and only permit creation of and access to the usual
6108 API device nodes like /dev/null or /dev/random, as well as
6109 access to (but not creation of) the pty devices.
6111 * We will now ship a default .network file for
6112 systemd-networkd that automatically configures DHCP for
6113 network interfaces created by nspawn's --network-veth or
6114 --network-bridge= switches.
6116 * systemd will now understand the usual M, K, G, T suffixes
6117 according to SI conventions (i.e. to the base 1000) when
6118 referring to throughput and hardware metrics. It will stay
6119 with IEC conventions (i.e. to the base 1024) for software
6120 metrics, according to what is customary according to
6121 Wikipedia. We explicitly document which base applies for
6122 each configuration option.
6124 * The DeviceAllow= setting in unit files now supports a syntax
6125 to whitelist an entire group of devices node majors at once,
6126 based on the /proc/devices listing. For example, with the
6127 string "char-pts", it is now possible to whitelist all
6128 current and future pseudo-TTYs at once.
6130 * sd-event learned a new "post" event source. Event sources of
6131 this type are triggered by the dispatching of any event
6132 source of a type that is not "post". This is useful for
6133 implementing clean-up and check event sources that are
6134 triggered by other work being done in the program.
6136 * systemd-networkd is no longer statically enabled, but uses
6137 the usual [Install] sections so that it can be
6138 enabled/disabled using systemctl. It still is enabled by
6141 * When creating a veth interface pair with systemd-nspawn, the
6142 host side will now be prefixed with "vb-" if
6143 --network-bridge= is used, and with "ve-" if --network-veth
6144 is used. This way, it is easy to distinguish these cases on
6145 the host, for example to apply different configuration to
6146 them with systemd-networkd.
6148 * The compatibility libraries for libsystemd-journal.so,
6149 libsystem-id128.so, libsystemd-login.so and
6150 libsystemd-daemon.so do not make use of IFUNC
6151 anymore. Instead, we now build libsystemd.so multiple times
6152 under these alternative names. This means that the footprint
6153 is drastically increased, but given that these are
6154 transitional compatibility libraries, this should not matter
6155 much. This change has been made necessary to support the ARM
6156 platform for these compatibility libraries, as the ARM
6157 toolchain is not really at the same level as the toolchain
6158 for other architectures like x86 and does not support
6159 IFUNC. Please make sure to use --enable-compat-libs only
6160 during a transitional period!
6162 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
6163 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
6164 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
6165 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
6166 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
6167 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
6168 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
6169 Zbigniew Jędrzejewski-Szmek
6171 — Berlin, 2014-02-24
6175 * A new component "systemd-networkd" has been added that can
6176 be used to configure local network interfaces statically or
6177 via DHCP. It is capable of bringing up bridges, VLANs, and
6178 bonding. Currently, no hook-ups for interactive network
6179 configuration are provided. Use this for your initrd,
6180 container, embedded, or server setup if you need a simple,
6181 yet powerful, network configuration solution. This
6182 configuration subsystem is quite nifty, as it allows wildcard
6183 hotplug matching in interfaces. For example, with a single
6184 configuration snippet, you can configure that all Ethernet
6185 interfaces showing up are automatically added to a bridge,
6186 or similar. It supports link-sensing and more.
6188 * A new tool "systemd-socket-proxyd" has been added which can
6189 act as a bidirectional proxy for TCP sockets. This is
6190 useful for adding socket activation support to services that
6191 do not actually support socket activation, including virtual
6192 machines and the like.
6194 * Add a new tool to save/restore rfkill state on
6197 * Save/restore state of keyboard backlights in addition to
6198 display backlights on shutdown/boot.
6200 * udev learned a new SECLABEL{} construct to label device
6201 nodes with a specific security label when they appear. For
6202 now, only SECLABEL{selinux} is supported, but the syntax is
6203 prepared for additional security frameworks.
6205 * udev gained a new scheme to configure link-level attributes
6206 from files in /etc/systemd/network/*.link. These files can
6207 match against MAC address, device path, driver name and type,
6208 and will apply attributes like the naming policy, link speed,
6209 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
6210 address assignment policy (randomized, ...).
6212 * The configuration of network interface naming rules for
6213 "permanent interface names" has changed: a new NamePolicy=
6214 setting in the [Link] section of .link files determines the
6215 priority of possible naming schemes (onboard, slot, MAC,
6216 path). The default value of this setting is determined by
6217 /usr/lib/net/links/99-default.link. Old
6218 80-net-name-slot.rules udev configuration file has been
6219 removed, so local configuration overriding this file should
6220 be adapted to override 99-default.link instead.
6222 * When the User= switch is used in a unit file, also
6223 initialize $SHELL= based on the user database entry.
6225 * systemd no longer depends on libdbus. All communication is
6226 now done with sd-bus, systemd's low-level bus library
6229 * kdbus support has been added to PID 1 itself. When kdbus is
6230 enabled, this causes PID 1 to set up the system bus and
6231 enable support for a new ".busname" unit type that
6232 encapsulates bus name activation on kdbus. It works a little
6233 bit like ".socket" units, except for bus names. A new
6234 generator has been added that converts classic dbus1 service
6235 activation files automatically into native systemd .busname
6238 * sd-bus: add a light-weight vtable implementation that allows
6239 defining objects on the bus with a simple static const
6240 vtable array of its methods, signals and properties.
6242 * systemd will not generate or install static dbus
6243 introspection data anymore to /usr/share/dbus-1/interfaces,
6244 as the precise format of these files is unclear, and
6245 nothing makes use of it.
6247 * A proxy daemon is now provided to proxy clients connecting
6248 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
6249 compatibility with classic D-Bus.
6251 * A bus driver implementation has been added that supports the
6252 classic D-Bus bus driver calls on kdbus, also for
6253 compatibility purposes.
6255 * A new API "sd-event.h" has been added that implements a
6256 minimal event loop API built around epoll. It provides a
6257 couple of features that direct epoll usage is lacking:
6258 prioritization of events, scales to large numbers of timer
6259 events, per-event timer slack (accuracy), system-wide
6260 coalescing of timer events, exit handlers, watchdog
6261 supervision support using systemd's sd_notify() API, child
6264 * A new API "sd-rntl.h" has been added that provides an API
6265 around the route netlink interface of the kernel, similar in
6266 style to "sd-bus.h".
6268 * A new API "sd-dhcp-client.h" has been added that provides a
6269 small DHCPv4 client-side implementation. This is used by
6272 * There is a new kernel command line option
6273 "systemd.restore_state=0|1". When set to "0", none of the
6274 systemd tools will restore saved runtime state to hardware
6275 devices. More specifically, the rfkill and backlight states
6278 * The FsckPassNo= compatibility option in mount/service units
6279 has been removed. The fstab generator will now add the
6280 necessary dependencies automatically, and does not require
6281 PID1's support for that anymore.
6283 * journalctl gained a new switch, --list-boots, that lists
6284 recent boots with their times and boot IDs.
6286 * The various tools like systemctl, loginctl, timedatectl,
6287 busctl, systemd-run, ... have gained a new switch "-M" to
6288 connect to a specific, local OS container (as direct
6289 connection, without requiring SSH). This works on any
6290 container that is registered with machined, such as those
6291 created by libvirt-lxc or nspawn.
6293 * systemd-run and systemd-analyze also gained support for "-H"
6294 to connect to remote hosts via SSH. This is particularly
6295 useful for systemd-run because it enables queuing of jobs
6296 onto remote systems.
6298 * machinectl gained a new command "login" to open a getty
6299 login in any local container. This works with any container
6300 that is registered with machined (such as those created by
6301 libvirt-lxc or nspawn), and which runs systemd inside.
6303 * machinectl gained a new "reboot" command that may be used to
6304 trigger a reboot on a specific container that is registered
6305 with machined. This works on any container that runs an init
6306 system of some kind.
6308 * systemctl gained a new "list-timers" command to print a nice
6309 listing of installed timer units with the times they elapse
6312 * Alternative reboot() parameters may now be specified on the
6313 "systemctl reboot" command line and are passed to the
6314 reboot() system call.
6316 * systemctl gained a new --job-mode= switch to configure the
6317 mode to queue a job with. This is a more generic version of
6318 --fail, --irreversible, and --ignore-dependencies, which are
6319 still available but not advertised anymore.
6321 * /etc/systemd/system.conf gained new settings to configure
6322 various default timeouts of units, as well as the default
6323 start limit interval and burst. These may still be overridden
6326 * PID1 will now export on the bus profile data of the security
6327 policy upload process (such as the SELinux policy upload to
6330 * journald: when forwarding logs to the console, include
6331 timestamps (following the setting in
6332 /sys/module/printk/parameters/time).
6334 * OnCalendar= in timer units now understands the special
6335 strings "yearly" and "annually". (Both are equivalent)
6337 * The accuracy of timer units is now configurable with the new
6338 AccuracySec= setting. It defaults to 1min.
6340 * A new dependency type JoinsNamespaceOf= has been added that
6341 allows running two services within the same /tmp and network
6342 namespace, if PrivateNetwork= or PrivateTmp= are used.
6344 * A new command "cat" has been added to systemctl. It outputs
6345 the original unit file of a unit, and concatenates the
6346 contents of additional "drop-in" unit file snippets, so that
6347 the full configuration is shown.
6349 * systemctl now supports globbing on the various "list-xyz"
6350 commands, like "list-units" or "list-sockets", as well as on
6351 those commands which take multiple unit names.
6353 * journalctl's --unit= switch gained support for globbing.
6355 * All systemd daemons now make use of the watchdog logic so
6356 that systemd automatically notices when they hang.
6358 * If the $container_ttys environment variable is set,
6359 getty-generator will automatically spawn a getty for each
6360 listed tty. This is useful for container managers to request
6361 login gettys to be spawned on as many ttys as needed.
6363 * %h, %s, %U specifier support is not available anymore when
6364 used in unit files for PID 1. This is because NSS calls are
6365 not safe from PID 1. They stay available for --user
6366 instances of systemd, and as special case for the root user.
6368 * loginctl gained a new "--no-legend" switch to turn off output
6371 * The "sd-login.h" API gained three new calls:
6372 sd_session_is_remote(), sd_session_get_remote_user(),
6373 sd_session_get_remote_host() to query information about
6376 * The udev hardware database now also carries vendor/product
6377 information of SDIO devices.
6379 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
6380 determine whether watchdog notifications are requested by
6383 * Socket-activated per-connection services now include a
6384 short description of the connection parameters in the
6387 * tmpfiles gained a new "--boot" option. When this is not used,
6388 only lines where the command character is not suffixed with
6389 "!" are executed. When this option is specified, those
6390 options are executed too. This partitions tmpfiles
6391 directives into those that can be safely executed at any
6392 time, and those which should be run only at boot (for
6393 example, a line that creates /run/nologin).
6395 * A new API "sd-resolve.h" has been added which provides a simple
6396 asynchronous wrapper around glibc NSS host name resolution
6397 calls, such as getaddrinfo(). In contrast to glibc's
6398 getaddrinfo_a(), it does not use signals. In contrast to most
6399 other asynchronous name resolution libraries, this one does
6400 not reimplement DNS, but reuses NSS, so that alternate
6401 host name resolution systems continue to work, such as mDNS,
6402 LDAP, etc. This API is based on libasyncns, but it has been
6403 cleaned up for inclusion in systemd.
6405 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
6406 "sd-daemon.h" are no longer found in individual libraries
6407 libsystemd-journal.so, libsystemd-login.so,
6408 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
6409 merged them into a single library, libsystemd.so, which
6410 provides all symbols. The reason for this is cyclic
6411 dependencies, as these libraries tend to use each other's
6412 symbols. So far, we have managed to workaround that by linking
6413 a copy of a good part of our code into each of these
6414 libraries again and again, which, however, makes certain
6415 things hard to do, like sharing static variables. Also, it
6416 substantially increases footprint. With this change, there
6417 is only one library for the basic APIs systemd
6418 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
6419 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
6420 library as well, however are subject to the --enable-kdbus
6421 switch (see below). Note that "sd-dhcp-client.h" is not part
6422 of this library (this is because it only consumes, never
6423 provides, services of/to other APIs). To make the transition
6424 easy from the separate libraries to the unified one, we
6425 provide the --enable-compat-libs compile-time switch which
6426 will generate stub libraries that are compatible with the
6427 old ones but redirect all calls to the new one.
6429 * All of the kdbus logic and the new APIs "sd-bus.h",
6430 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
6431 and "sd-utf8.h" are compile-time optional via the
6432 "--enable-kdbus" switch, and they are not compiled in by
6433 default. To make use of kdbus, you have to explicitly enable
6434 the switch. Note however, that neither the kernel nor the
6435 userspace API for all of this is considered stable yet. We
6436 want to maintain the freedom to still change the APIs for
6437 now. By specifying this build-time switch, you acknowledge
6438 that you are aware of the instability of the current
6441 * Also, note that while kdbus is pretty much complete,
6442 it lacks one thing: proper policy support. This means you
6443 can build a fully working system with all features; however,
6444 it will be highly insecure. Policy support will be added in
6445 one of the next releases, at the same time that we will
6446 declare the APIs stable.
6448 * When the kernel command line argument "kdbus" is specified,
6449 systemd will automatically load the kdbus.ko kernel module. At
6450 this stage of development, it is only useful for testing kdbus
6451 and should not be used in production. Note: if "--enable-kdbus"
6452 is specified, and the kdbus.ko kernel module is available, and
6453 "kdbus" is added to the kernel command line, the entire system
6454 runs with kdbus instead of dbus-daemon, with the above mentioned
6455 problem of missing the system policy enforcement. Also a future
6456 version of kdbus.ko or a newer systemd will not be compatible with
6457 each other, and will unlikely be able to boot the machine if only
6458 one of them is updated.
6460 * systemctl gained a new "import-environment" command which
6461 uploads the caller's environment (or parts thereof) into the
6462 service manager so that it is inherited by services started
6463 by the manager. This is useful to upload variables like
6464 $DISPLAY into the user service manager.
6466 * A new PrivateDevices= switch has been added to service units
6467 which allows running a service with a namespaced /dev
6468 directory that does not contain any device nodes for
6469 physical devices. More specifically, it only includes devices
6470 such as /dev/null, /dev/urandom, and /dev/zero which are API
6473 * logind has been extended to support behaviour like VT
6474 switching on seats that do not support a VT. This makes
6475 multi-session available on seats that are not the first seat
6476 (seat0), and on systems where kernel support for VTs has
6477 been disabled at compile-time.
6479 * If a process holds a delay lock for system sleep or shutdown
6480 and fails to release it in time, we will now log its
6481 identity. This makes it easier to identify processes that
6482 cause slow suspends or power-offs.
6484 * When parsing /etc/crypttab, support for a new key-slot=
6485 option as supported by Debian is added. It allows indicating
6486 which LUKS slot to use on disk, speeding up key loading.
6488 * The sd_journal_sendv() API call has been checked and
6489 officially declared to be async-signal-safe so that it may
6490 be invoked from signal handlers for logging purposes.
6492 * Boot-time status output is now enabled automatically after a
6493 short timeout if boot does not progress, in order to give
6494 the user an indication what she or he is waiting for.
6496 * The boot-time output has been improved to show how much time
6497 remains until jobs expire.
6499 * The KillMode= switch in service units gained a new possible
6500 value "mixed". If set, and the unit is shut down, then the
6501 initial SIGTERM signal is sent only to the main daemon
6502 process, while the following SIGKILL signal is sent to
6503 all remaining processes of the service.
6505 * When a scope unit is registered, a new property "Controller"
6506 may be set. If set to a valid bus name, systemd will send a
6507 RequestStop() signal to this name when it would like to shut
6508 down the scope. This may be used to hook manager logic into
6509 the shutdown logic of scope units. Also, scope units may now
6510 be put in a special "abandoned" state, in which case the
6511 manager process which created them takes no further
6512 responsibilities for it.
6514 * When reading unit files, systemd will now verify
6515 the access mode of these files, and warn about certain
6516 suspicious combinations. This has been added to make it
6517 easier to track down packaging bugs where unit files are
6518 marked executable or world-writable.
6520 * systemd-nspawn gained a new "--setenv=" switch to set
6521 container-wide environment variables. The similar option in
6522 systemd-activate was renamed from "--environment=" to
6523 "--setenv=" for consistency.
6525 * systemd-nspawn has been updated to create a new kdbus domain
6526 for each container that is invoked, thus allowing each
6527 container to have its own set of system and user buses,
6528 independent of the host.
6530 * systemd-nspawn gained a new --drop-capability= switch to run
6531 the container with less capabilities than the default. Both
6532 --drop-capability= and --capability= now take the special
6533 string "all" for dropping or keeping all capabilities.
6535 * systemd-nspawn gained new switches for executing containers
6536 with specific SELinux labels set.
6538 * systemd-nspawn gained a new --quiet switch to not generate
6539 any additional output but the container's own console
6542 * systemd-nspawn gained a new --share-system switch to run a
6543 container without PID namespacing enabled.
6545 * systemd-nspawn gained a new --register= switch to control
6546 whether the container is registered with systemd-machined or
6547 not. This is useful for containers that do not run full
6548 OS images, but only specific apps.
6550 * systemd-nspawn gained a new --keep-unit which may be used
6551 when invoked as the only program from a service unit, and
6552 results in registration of the unit service itself in
6553 systemd-machined, instead of a newly opened scope unit.
6555 * systemd-nspawn gained a new --network-interface= switch for
6556 moving arbitrary interfaces to the container. The new
6557 --network-veth switch creates a virtual Ethernet connection
6558 between host and container. The new --network-bridge=
6559 switch then allows assigning the host side of this virtual
6560 Ethernet connection to a bridge device.
6562 * systemd-nspawn gained a new --personality= switch for
6563 setting the kernel personality for the container. This is
6564 useful when running a 32-bit container on a 64-bit host. A
6565 similar option Personality= is now also available for service
6568 * logind will now also track a "Desktop" identifier for each
6569 session which encodes the desktop environment of it. This is
6570 useful for desktop environments that want to identify
6571 multiple running sessions of itself easily.
6573 * A new SELinuxContext= setting for service units has been
6574 added that allows setting a specific SELinux execution
6575 context for a service.
6577 * Most systemd client tools will now honour $SYSTEMD_LESS for
6578 settings of the "less" pager. By default, these tools will
6579 override $LESS to allow certain operations to work, such as
6580 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
6581 influence this logic.
6583 * systemd's "seccomp" hook-up has been changed to make use of
6584 the libseccomp library instead of using its own
6585 implementation. This has benefits for portability among
6588 * For usage together with SystemCallFilter=, a new
6589 SystemCallErrorNumber= setting has been introduced that
6590 allows configuration of a system error number to be returned
6591 on filtered system calls, instead of immediately killing the
6592 process. Also, SystemCallArchitectures= has been added to
6593 limit access to system calls of a particular architecture
6594 (in order to turn off support for unused secondary
6595 architectures). There is also a global
6596 SystemCallArchitectures= setting in system.conf now to turn
6597 off support for non-native system calls system-wide.
6599 * systemd requires a kernel with a working name_to_handle_at(),
6600 please see the kernel config requirements in the README file.
6602 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
6603 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
6604 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
6605 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
6606 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
6607 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
6608 Elia Pinto, Florian Weimer, George McCollister, Goffredo
6609 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
6610 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
6611 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
6612 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
6613 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
6614 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
6615 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
6616 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
6617 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
6618 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
6619 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
6620 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
6621 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
6622 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
6623 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
6624 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
6625 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
6627 — Berlin, 2014-02-20
6631 * logind has gained support for facilitating privileged input
6632 and drm device access for unprivileged clients. This work is
6633 useful to allow Wayland display servers (and similar
6634 programs, such as kmscon) to run under the user's ID and
6635 access input and drm devices which are normally
6636 protected. When this is used (and the kernel is new enough)
6637 logind will "mute" IO on the file descriptors passed to
6638 Wayland as long as it is in the background and "unmute" it
6639 if it returns into the foreground. This allows secure
6640 session switching without allowing background sessions to
6641 eavesdrop on input and display data. This also introduces
6642 session switching support if VT support is turned off in the
6643 kernel, and on seats that are not seat0.
6645 * A new kernel command line option luks.options= is understood
6646 now which allows specifying LUKS options for usage for LUKS
6647 encrypted partitions specified with luks.uuid=.
6649 * tmpfiles.d(5) snippets may now use specifier expansion in
6650 path names. More specifically %m, %b, %H, %v, are now
6651 replaced by the local machine id, boot id, hostname, and
6652 kernel version number.
6654 * A new tmpfiles.d(5) command "m" has been introduced which
6655 may be used to change the owner/group/access mode of a file
6656 or directory if it exists, but do nothing if it does not.
6658 * This release removes high-level support for the
6659 MemorySoftLimit= cgroup setting. The underlying kernel
6660 cgroup attribute memory.soft_limit= is currently badly
6661 designed and likely to be removed from the kernel API in its
6662 current form, hence we should not expose it for now.
6664 * The memory.use_hierarchy cgroup attribute is now enabled for
6665 all cgroups systemd creates in the memory cgroup
6666 hierarchy. This option is likely to be come the built-in
6667 default in the kernel anyway, and the non-hierarchical mode
6668 never made much sense in the intrinsically hierarchical
6671 * A new field _SYSTEMD_SLICE= is logged along with all journal
6672 messages containing the slice a message was generated
6673 from. This is useful to allow easy per-customer filtering of
6674 logs among other things.
6676 * systemd-journald will no longer adjust the group of journal
6677 files it creates to the "systemd-journal" group. Instead we
6678 rely on the journal directory to be owned by the
6679 "systemd-journal" group, and its setgid bit set, so that the
6680 kernel file system layer will automatically enforce that
6681 journal files inherit this group assignment. The reason for
6682 this change is that we cannot allow NSS look-ups from
6683 journald which would be necessary to resolve
6684 "systemd-journal" to a numeric GID, because this might
6685 create deadlocks if NSS involves synchronous queries to
6686 other daemons (such as nscd, or sssd) which in turn are
6687 logging clients of journald and might block on it, which
6688 would then dead lock. A tmpfiles.d(5) snippet included in
6689 systemd will make sure the setgid bit and group are
6690 properly set on the journal directory if it exists on every
6691 boot. However, we recommend adjusting it manually after
6692 upgrades too (or from RPM scriptlets), so that the change is
6693 not delayed until next reboot.
6695 * Backlight and random seed files in /var/lib/ have moved into
6696 the /var/lib/systemd/ directory, in order to centralize all
6697 systemd generated files in one directory.
6699 * Boot time performance measurements (as displayed by
6700 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
6701 performance information if that's available to determine how
6702 much time BIOS and boot loader initialization required. With
6703 a sufficiently new BIOS you hence no longer need to boot
6704 with Gummiboot to get access to such information.
6706 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
6707 Cristian Rodríguez, Dave Reisner, David Herrmann, David
6708 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
6709 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
6710 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
6711 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
6712 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6714 — Berlin, 2013-10-02
6718 * The Restart= option for services now understands a new
6719 on-watchdog setting, which will restart the service
6720 automatically if the service stops sending out watchdog keep
6721 alive messages (as configured with WatchdogSec=).
6723 * The getty generator (which is responsible for bringing up a
6724 getty on configured serial consoles) will no longer only
6725 start a getty on the primary kernel console but on all
6726 others, too. This makes the order in which console= is
6727 specified on the kernel command line less important.
6729 * libsystemd-logind gained a new sd_session_get_vt() call to
6730 retrieve the VT number of a session.
6732 * If the option "tries=0" is set for an entry of /etc/crypttab
6733 its passphrase is queried indefinitely instead of any
6734 maximum number of tries.
6736 * If a service with a configure PID file terminates its PID
6737 file will now be removed automatically if it still exists
6738 afterwards. This should put an end to stale PID files.
6740 * systemd-run will now also take relative binary path names
6741 for execution and no longer insists on absolute paths.
6743 * InaccessibleDirectories= and ReadOnlyDirectories= now take
6744 paths that are optionally prefixed with "-" to indicate that
6745 it should not be considered a failure if they do not exist.
6747 * journalctl -o (and similar commands) now understands a new
6748 output mode "short-precise", it is similar to "short" but
6749 shows timestamps with usec accuracy.
6751 * The option "discard" (as known from Debian) is now
6752 synonymous to "allow-discards" in /etc/crypttab. In fact,
6753 "discard" is preferred now (since it is easier to remember
6756 * Some licensing clean-ups were made, so that more code is now
6757 LGPL-2.1 licensed than before.
6759 * A minimal tool to save/restore the display backlight
6760 brightness across reboots has been added. It will store the
6761 backlight setting as late as possible at shutdown, and
6762 restore it as early as possible during reboot.
6764 * A logic to automatically discover and enable home and swap
6765 partitions on GPT disks has been added. With this in place
6766 /etc/fstab becomes optional for many setups as systemd can
6767 discover certain partitions located on the root disk
6768 automatically. Home partitions are recognized under their
6769 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
6770 partitions are recognized under their GPT type ID
6771 0657fd6da4ab43c484e50933c84b4f4f.
6773 * systemd will no longer pass any environment from the kernel
6774 or initrd to system services. If you want to set an
6775 environment for all services, do so via the kernel command
6776 line systemd.setenv= assignment.
6778 * The systemd-sysctl tool no longer natively reads the file
6779 /etc/sysctl.conf. If desired, the file should be symlinked
6780 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
6781 legacy support by a symlink rather than built-in code, it
6782 also makes the otherwise hidden order of application of the
6783 different files visible. (Note that this partly reverts to a
6784 pre-198 application order of sysctl knobs!)
6786 * The "systemctl set-log-level" and "systemctl dump" commands
6787 have been moved to systemd-analyze.
6789 * systemd-run learned the new --remain-after-exit switch,
6790 which causes the scope unit not to be cleaned up
6791 automatically after the process terminated.
6793 * tmpfiles learned a new --exclude-prefix= switch to exclude
6794 certain paths from operation.
6796 * journald will now automatically flush all messages to disk
6797 as soon as a message at the log level CRIT, ALERT or EMERG
6800 Contributions from: Andrew Cook, Brandon Philips, Christian
6801 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
6802 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
6803 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
6804 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
6805 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
6806 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
6807 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
6808 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
6809 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
6810 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
6811 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
6812 William Giokas, Zbigniew Jędrzejewski-Szmek
6814 — Berlin, 2013-09-13
6818 * The documentation has been updated to cover the various new
6819 concepts introduced with 205.
6821 * Unit files now understand the new %v specifier which
6822 resolves to the kernel version string as returned by "uname
6825 * systemctl now supports filtering the unit list output by
6826 load state, active state and sub state, using the new
6829 * "systemctl status" will now show the results of the
6830 condition checks (like ConditionPathExists= and similar) of
6831 the last start attempts of the unit. They are also logged to
6834 * "journalctl -b" may now be used to look for boot output of a
6835 specific boot. Try "journalctl -b -1" for the previous boot,
6836 but the syntax is substantially more powerful.
6838 * "journalctl --show-cursor" has been added which prints the
6839 cursor string the last shown log line. This may then be used
6840 with the new "journalctl --after-cursor=" switch to continue
6841 browsing logs from that point on.
6843 * "journalctl --force" may now be used to force regeneration
6846 * Creation of "dead" device nodes has been moved from udev
6847 into kmod and tmpfiles. Previously, udev would read the kmod
6848 databases to pre-generate dead device nodes based on meta
6849 information contained in kernel modules, so that these would
6850 be auto-loaded on access rather then at boot. As this
6851 does not really have much to do with the exposing actual
6852 kernel devices to userspace this has always been slightly
6853 alien in the udev codebase. Following the new scheme kmod
6854 will now generate a runtime snippet for tmpfiles from the
6855 module meta information and it now is tmpfiles' job to the
6856 create the nodes. This also allows overriding access and
6857 other parameters for the nodes using the usual tmpfiles
6858 facilities. As side effect this allows us to remove the
6859 CAP_SYS_MKNOD capability bit from udevd entirely.
6861 * logind's device ACLs may now be applied to these "dead"
6862 devices nodes too, thus finally allowing managed access to
6863 devices such as /dev/snd/sequencer without loading the
6864 backing module right-away.
6866 * A new RPM macro has been added that may be used to apply
6867 tmpfiles configuration during package installation.
6869 * systemd-detect-virt and ConditionVirtualization= now can
6870 detect User-Mode-Linux machines (UML).
6872 * journald will now implicitly log the effective capabilities
6873 set of processes in the message metadata.
6875 * systemd-cryptsetup has gained support for TrueCrypt volumes.
6877 * The initrd interface has been simplified (more specifically,
6878 support for passing performance data via environment
6879 variables and fsck results via files in /run has been
6880 removed). These features were non-essential, and are
6881 nowadays available in a much nicer way by having systemd in
6882 the initrd serialize its state and have the hosts systemd
6883 deserialize it again.
6885 * The udev "keymap" data files and tools to apply keyboard
6886 specific mappings of scan to key codes, and force-release
6887 scan code lists have been entirely replaced by a udev
6888 "keyboard" builtin and a hwdb data file.
6890 * systemd will now honour the kernel's "quiet" command line
6891 argument also during late shutdown, resulting in a
6892 completely silent shutdown when used.
6894 * There's now an option to control the SO_REUSEPORT socket
6895 option in .socket units.
6897 * Instance units will now automatically get a per-template
6898 subslice of system.slice unless something else is explicitly
6899 configured. For example, instances of sshd@.service will now
6900 implicitly be placed in system-sshd.slice rather than
6901 system.slice as before.
6903 * Test coverage support may now be enabled at build time.
6905 Contributions from: Dave Reisner, Frederic Crozat, Harald
6906 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
6907 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
6908 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
6909 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
6910 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
6911 Giokas, Zbigniew Jędrzejewski-Szmek
6913 — Berlin, 2013-07-23
6917 * Two new unit types have been introduced:
6919 Scope units are very similar to service units, however, are
6920 created out of pre-existing processes — instead of PID 1
6921 forking off the processes. By using scope units it is
6922 possible for system services and applications to group their
6923 own child processes (worker processes) in a powerful way
6924 which then maybe used to organize them, or kill them
6925 together, or apply resource limits on them.
6927 Slice units may be used to partition system resources in an
6928 hierarchical fashion and then assign other units to them. By
6929 default there are now three slices: system.slice (for all
6930 system services), user.slice (for all user sessions),
6931 machine.slice (for VMs and containers).
6933 Slices and scopes have been introduced primarily in
6934 context of the work to move cgroup handling to a
6935 single-writer scheme, where only PID 1
6936 creates/removes/manages cgroups.
6938 * There's a new concept of "transient" units. In contrast to
6939 normal units these units are created via an API at runtime,
6940 not from configuration from disk. More specifically this
6941 means it is now possible to run arbitrary programs as
6942 independent services, with all execution parameters passed
6943 in via bus APIs rather than read from disk. Transient units
6944 make systemd substantially more dynamic then it ever was,
6945 and useful as a general batch manager.
6947 * logind has been updated to make use of scope and slice units
6948 for managing user sessions. As a user logs in he will get
6949 his own private slice unit, to which all sessions are added
6950 as scope units. We also added support for automatically
6951 adding an instance of user@.service for the user into the
6952 slice. Effectively logind will no longer create cgroup
6953 hierarchies on its own now, it will defer entirely to PID 1
6954 for this by means of scope, service and slice units. Since
6955 user sessions this way become entities managed by PID 1
6956 the output of "systemctl" is now a lot more comprehensive.
6958 * A new mini-daemon "systemd-machined" has been added which
6959 may be used by virtualization managers to register local
6960 VMs/containers. nspawn has been updated accordingly, and
6961 libvirt will be updated shortly. machined will collect a bit
6962 of meta information about the VMs/containers, and assign
6963 them their own scope unit (see above). The collected
6964 meta-data is then made available via the "machinectl" tool,
6965 and exposed in "ps" and similar tools. machined/machinectl
6966 is compile-time optional.
6968 * As discussed earlier, the low-level cgroup configuration
6969 options ControlGroup=, ControlGroupModify=,
6970 ControlGroupPersistent=, ControlGroupAttribute= have been
6971 removed. Please use high-level attribute settings instead as
6972 well as slice units.
6974 * A new bus call SetUnitProperties() has been added to alter
6975 various runtime parameters of a unit. This is primarily
6976 useful to alter cgroup parameters dynamically in a nice way,
6977 but will be extended later on to make more properties
6978 modifiable at runtime. systemctl gained a new set-properties
6979 command that wraps this call.
6981 * A new tool "systemd-run" has been added which can be used to
6982 run arbitrary command lines as transient services or scopes,
6983 while configuring a number of settings via the command
6984 line. This tool is currently very basic, however already
6985 very useful. We plan to extend this tool to even allow
6986 queuing of execution jobs with time triggers from the
6987 command line, similar in fashion to "at".
6989 * nspawn will now inform the user explicitly that kernels with
6990 audit enabled break containers, and suggest the user to turn
6993 * Support for detecting the IMA and AppArmor security
6994 frameworks with ConditionSecurity= has been added.
6996 * journalctl gained a new "-k" switch for showing only kernel
6997 messages, mimicking dmesg output; in addition to "--user"
6998 and "--system" switches for showing only user's own logs
7001 * systemd-delta can now show information about drop-in
7002 snippets extending unit files.
7004 * libsystemd-bus has been substantially updated but is still
7005 not available as public API.
7007 * systemd will now look for the "debug" argument on the kernel
7008 command line and enable debug logging, similar to what
7009 "systemd.log_level=debug" already did before.
7011 * "systemctl set-default", "systemctl get-default" has been
7012 added to configure the default.target symlink, which
7013 controls what to boot into by default.
7015 * "systemctl set-log-level" has been added as a convenient
7016 way to raise and lower systemd logging threshold.
7018 * "systemd-analyze plot" will now show the time the various
7019 generators needed for execution, as well as information
7020 about the unit file loading.
7022 * libsystemd-journal gained a new sd_journal_open_files() call
7023 for opening specific journal files. journactl also gained a
7024 new switch to expose this new functionality. Previously we
7025 only supported opening all files from a directory, or all
7026 files from the system, as opening individual files only is
7027 racy due to journal file rotation.
7029 * systemd gained the new DefaultEnvironment= setting in
7030 /etc/systemd/system.conf to set environment variables for
7033 * If a privileged process logs a journal message with the
7034 OBJECT_PID= field set, then journald will automatically
7035 augment this with additional OBJECT_UID=, OBJECT_GID=,
7036 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
7037 system services want to log events about specific client
7038 processes. journactl/systemctl has been updated to make use
7039 of this information if all log messages regarding a specific
7042 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
7043 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
7044 Reisner, David Coppa, David King, David Strauss, Eelco
7045 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
7046 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
7047 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
7048 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
7049 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
7050 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
7051 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
7052 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
7053 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
7054 Łukasz Stelmach, 장동준
7058 * The Python bindings gained some minimal support for the APIs
7059 exposed by libsystemd-logind.
7061 * ConditionSecurity= gained support for detecting SMACK. Since
7062 this condition already supports SELinux and AppArmor we only
7063 miss IMA for this. Patches welcome!
7065 Contributions from: Karol Lewandowski, Lennart Poettering,
7066 Zbigniew Jędrzejewski-Szmek
7070 * systemd-nspawn will now create /etc/resolv.conf if
7071 necessary, before bind-mounting the host's file onto it.
7073 * systemd-nspawn will now store meta information about a
7074 container on the container's cgroup as extended attribute
7075 fields, including the root directory.
7077 * The cgroup hierarchy has been reworked in many ways. All
7078 objects any of the components systemd creates in the cgroup
7079 tree are now suffixed. More specifically, user sessions are
7080 now placed in cgroups suffixed with ".session", users in
7081 cgroups suffixed with ".user", and nspawn containers in
7082 cgroups suffixed with ".nspawn". Furthermore, all cgroup
7083 names are now escaped in a simple scheme to avoid collision
7084 of userspace object names with kernel filenames. This work
7085 is preparation for making these objects relocatable in the
7086 cgroup tree, in order to allow easy resource partitioning of
7087 these objects without causing naming conflicts.
7089 * systemctl list-dependencies gained the new switches
7090 --plain, --reverse, --after and --before.
7092 * systemd-inhibit now shows the process name of processes that
7093 have taken an inhibitor lock.
7095 * nss-myhostname will now also resolve "localhost"
7096 implicitly. This makes /etc/hosts an optional file and
7097 nicely handles that on IPv6 ::1 maps to both "localhost" and
7100 * libsystemd-logind.so gained a new call
7101 sd_get_machine_names() to enumerate running containers and
7102 VMs (currently only supported by very new libvirt and
7103 nspawn). sd_login_monitor can now be used to watch
7104 VMs/containers coming and going.
7106 * .include is not allowed recursively anymore, and only in
7107 unit files. Usually it is better to use drop-in snippets in
7108 .d/*.conf anyway, as introduced with systemd 198.
7110 * systemd-analyze gained a new "critical-chain" command that
7111 determines the slowest chain of units run during system
7112 boot-up. It is very useful for tracking down where
7113 optimizing boot time is the most beneficial.
7115 * systemd will no longer allow manipulating service paths in
7116 the name=systemd:/system cgroup tree using ControlGroup= in
7117 units. (But is still fine with it in all other dirs.)
7119 * There's a new systemd-nspawn@.service service file that may
7120 be used to easily run nspawn containers as system
7121 services. With the container's root directory in
7122 /var/lib/container/foobar it is now sufficient to run
7123 "systemctl start systemd-nspawn@foobar.service" to boot it.
7125 * systemd-cgls gained a new parameter "--machine" to list only
7126 the processes within a certain container.
7128 * ConditionSecurity= now can check for "apparmor". We still
7129 are lacking checks for SMACK and IMA for this condition
7130 check though. Patches welcome!
7132 * A new configuration file /etc/systemd/sleep.conf has been
7133 added that may be used to configure which kernel operation
7134 systemd is supposed to execute when "suspend", "hibernate"
7135 or "hybrid-sleep" is requested. This makes the new kernel
7136 "freeze" state accessible to the user.
7138 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
7139 the passed argument if applicable.
7141 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
7142 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
7143 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
7144 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
7145 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
7146 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
7147 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
7152 * The output of 'systemctl list-jobs' got some polishing. The
7153 '--type=' argument may now be passed more than once. A new
7154 command 'systemctl list-sockets' has been added which shows
7155 a list of kernel sockets systemd is listening on with the
7156 socket units they belong to, plus the units these socket
7159 * The experimental libsystemd-bus library got substantial
7160 updates to work in conjunction with the (also experimental)
7161 kdbus kernel project. It works well enough to exchange
7162 messages with some sophistication. Note that kdbus is not
7163 ready yet, and the library is mostly an elaborate test case
7164 for now, and not installable.
7166 * systemd gained a new unit 'systemd-static-nodes.service'
7167 that generates static device nodes earlier during boot, and
7168 can run in conjunction with udev.
7170 * libsystemd-login gained a new call sd_pid_get_user_unit()
7171 to retrieve the user systemd unit a process is running
7172 in. This is useful for systems where systemd is used as
7175 * systemd-nspawn now places all containers in the new /machine
7176 top-level cgroup directory in the name=systemd
7177 hierarchy. libvirt will soon do the same, so that we get a
7178 uniform separation of /system, /user and /machine for system
7179 services, user processes and containers/virtual
7180 machines. This new cgroup hierarchy is also useful to stick
7181 stable names to specific container instances, which can be
7182 recognized later this way (this name may be controlled
7183 via systemd-nspawn's new -M switch). libsystemd-login also
7184 gained a new call sd_pid_get_machine_name() to retrieve the
7185 name of the container/VM a specific process belongs to.
7187 * bootchart can now store its data in the journal.
7189 * libsystemd-journal gained a new call
7190 sd_journal_add_conjunction() for AND expressions to the
7191 matching logic. This can be used to express more complex
7192 logical expressions.
7194 * journactl can now take multiple --unit= and --user-unit=
7197 * The cryptsetup logic now understands the "luks.key=" kernel
7198 command line switch for specifying a file to read the
7199 decryption key from. Also, if a configured key file is not
7200 found the tool will now automatically fall back to prompting
7203 * Python systemd.journal module was updated to wrap recently
7204 added functions from libsystemd-journal. The interface was
7205 changed to bring the low level interface in s.j._Reader
7206 closer to the C API, and the high level interface in
7207 s.j.Reader was updated to wrap and convert all data about
7210 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
7211 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
7212 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
7213 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
7214 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
7215 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
7219 * journalctl --update-catalog now understands a new --root=
7220 option to operate on catalogs found in a different root
7223 * During shutdown after systemd has terminated all running
7224 services a final killing loop kills all remaining left-over
7225 processes. We will now print the name of these processes
7226 when we send SIGKILL to them, since this usually indicates a
7229 * If /etc/crypttab refers to password files stored on
7230 configured mount points automatic dependencies will now be
7231 generated to ensure the specific mount is established first
7232 before the key file is attempted to be read.
7234 * 'systemctl status' will now show information about the
7235 network sockets a socket unit is listening on.
7237 * 'systemctl status' will also shown information about any
7238 drop-in configuration file for units. (Drop-In configuration
7239 files in this context are files such as
7240 /etc/systemd/systemd/foobar.service.d/*.conf)
7242 * systemd-cgtop now optionally shows summed up CPU times of
7243 cgroups. Press '%' while running cgtop to switch between
7244 percentage and absolute mode. This is useful to determine
7245 which cgroups use up the most CPU time over the entire
7246 runtime of the system. systemd-cgtop has also been updated
7247 to be 'pipeable' for processing with further shell tools.
7249 * 'hostnamectl set-hostname' will now allow setting of FQDN
7252 * The formatting and parsing of time span values has been
7253 changed. The parser now understands fractional expressions
7254 such as "5.5h". The formatter will now output fractional
7255 expressions for all time spans under 1min, i.e. "5.123456s"
7256 rather than "5s 123ms 456us". For time spans under 1s
7257 millisecond values are shown, for those under 1ms
7258 microsecond values are shown. This should greatly improve
7259 all time-related output of systemd.
7261 * libsystemd-login and libsystemd-journal gained new
7262 functions for querying the poll() events mask and poll()
7263 timeout value for integration into arbitrary event
7266 * localectl gained the ability to list available X11 keymaps
7267 (models, layouts, variants, options).
7269 * 'systemd-analyze dot' gained the ability to filter for
7270 specific units via shell-style globs, to create smaller,
7271 more useful graphs. I.e. it is now possible to create simple
7272 graphs of all the dependencies between only target units, or
7273 of all units that Avahi has dependencies with.
7275 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
7276 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
7277 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
7278 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
7279 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
7280 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
7281 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
7285 * The boot-time readahead implementation for rotating media
7286 will now read the read-ahead data in multiple passes which
7287 consist of all read requests made in equidistant time
7288 intervals. This means instead of strictly reading read-ahead
7289 data in its physical order on disk we now try to find a
7290 middle ground between physical and access time order.
7292 * /etc/os-release files gained a new BUILD_ID= field for usage
7293 on operating systems that provide continuous builds of OS
7296 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
7297 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
7298 William Douglas, Zbigniew Jędrzejewski-Szmek
7302 * systemd-python gained an API exposing libsystemd-daemon.
7304 * The SMACK setup logic gained support for uploading CIPSO
7307 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
7308 ReadOnlyDirectories= and InaccessibleDirectories= has
7309 changed. The private /tmp and /var/tmp directories are now
7310 shared by all processes of a service (which means
7311 ExecStartPre= may now leave data in /tmp that ExecStart= of
7312 the same service can still access). When a service is
7313 stopped its temporary directories are immediately deleted
7314 (normal clean-up with tmpfiles is still done in addition to
7317 * By default, systemd will now set a couple of sysctl
7318 variables in the kernel: the safe sysrq options are turned
7319 on, IP route verification is turned on, and source routing
7320 disabled. The recently added hardlink and softlink
7321 protection of the kernel is turned on. These settings should
7322 be reasonably safe, and good defaults for all new systems.
7324 * The predictable network naming logic may now be turned off
7325 with a new kernel command line switch: net.ifnames=0.
7327 * A new libsystemd-bus module has been added that implements a
7328 pretty complete D-Bus client library. For details see:
7330 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
7332 * journald will now explicitly flush the journal files to disk
7333 at the latest 5min after each write. The file will then also
7334 be marked offline until the next write. This should increase
7335 reliability in case of a crash. The synchronization delay
7336 can be configured via SyncIntervalSec= in journald.conf.
7338 * There's a new remote-fs-setup.target unit that can be used
7339 to pull in specific services when at least one remote file
7340 system is to be mounted.
7342 * There are new targets timers.target and paths.target as
7343 canonical targets to pull user timer and path units in
7344 from. This complements sockets.target with a similar
7345 purpose for socket units.
7347 * libudev gained a new call udev_device_set_attribute_value()
7348 to set sysfs attributes of a device.
7350 * The udev daemon now sets the default number of worker
7351 processes executed in parallel based on the number of available
7352 CPUs instead of the amount of available RAM. This is supposed
7353 to provide a more reliable default and limit a too aggressive
7354 parallelism for setups with 1000s of devices connected.
7356 Contributions from: Auke Kok, Colin Walters, Cristian
7357 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
7358 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
7359 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
7360 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
7361 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
7362 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
7363 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
7364 Zbigniew Jędrzejewski-Szmek
7368 * Configuration of unit files may now be extended via drop-in
7369 files without having to edit/override the unit files
7370 themselves. More specifically, if the administrator wants to
7371 change one value for a service file foobar.service he can
7372 now do so by dropping in a configuration snippet into
7373 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
7374 will load all these snippets and apply them on top of the
7375 main unit configuration file, possibly extending or
7376 overriding its settings. Using these drop-in snippets is
7377 generally nicer than the two earlier options for changing
7378 unit files locally: copying the files from
7379 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
7380 them there; or creating a new file in /etc/systemd/system/
7381 that incorporates the original one via ".include". Drop-in
7382 snippets into these .d/ directories can be placed in any
7383 directory systemd looks for units in, and the usual
7384 overriding semantics between /usr/lib, /etc and /run apply
7387 * Most unit file settings which take lists of items can now be
7388 reset by assigning the empty string to them. For example,
7389 normally, settings such as Environment=FOO=BAR append a new
7390 environment variable assignment to the environment block,
7391 each time they are used. By assigning Environment= the empty
7392 string the environment block can be reset to empty. This is
7393 particularly useful with the .d/*.conf drop-in snippets
7394 mentioned above, since this adds the ability to reset list
7395 settings from vendor unit files via these drop-ins.
7397 * systemctl gained a new "list-dependencies" command for
7398 listing the dependencies of a unit recursively.
7400 * Inhibitors are now honored and listed by "systemctl
7401 suspend", "systemctl poweroff" (and similar) too, not only
7402 GNOME. These commands will also list active sessions by
7405 * Resource limits (as exposed by the various control group
7406 controllers) can now be controlled dynamically at runtime
7407 for all units. More specifically, you can now use a command
7408 like "systemctl set-cgroup-attr foobar.service cpu.shares
7409 2000" to alter the CPU shares a specific service gets. These
7410 settings are stored persistently on disk, and thus allow the
7411 administrator to easily adjust the resource usage of
7412 services with a few simple commands. This dynamic resource
7413 management logic is also available to other programs via the
7414 bus. Almost any kernel cgroup attribute and controller is
7417 * systemd-vconsole-setup will now copy all font settings to
7418 all allocated VTs, where it previously applied them only to
7421 * libsystemd-login gained the new sd_session_get_tty() API
7424 * This release drops support for a few legacy or
7425 distribution-specific LSB facility names when parsing init
7426 scripts: $x-display-manager, $mail-transfer-agent,
7427 $mail-transport-agent, $mail-transfer-agent, $smtp,
7428 $null. Also, the mail-transfer-agent.target unit backing
7429 this has been removed. Distributions which want to retain
7430 compatibility with this should carry the burden for
7431 supporting this themselves and patch support for these back
7432 in, if they really need to. Also, the facilities $syslog and
7433 $local_fs are now ignored, since systemd does not support
7434 early-boot LSB init scripts anymore, and these facilities
7435 are implied anyway for normal services. syslog.target has
7438 * There are new bus calls on PID1's Manager object for
7439 cancelling jobs, and removing snapshot units. Previously,
7440 both calls were only available on the Job and Snapshot
7443 * systemd-journal-gatewayd gained SSL support.
7445 * The various "environment" files, such as /etc/locale.conf
7446 now support continuation lines with a backslash ("\") as
7447 last character in the line, similarly in style (but different)
7448 to how this is supported in shells.
7450 * For normal user processes the _SYSTEMD_USER_UNIT= field is
7451 now implicitly appended to every log entry logged. systemctl
7452 has been updated to filter by this field when operating on a
7453 user systemd instance.
7455 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
7456 CAP_AUDIT_CONTROL capabilities to the capabilities set for
7457 the container. This makes it easier to boot unmodified
7458 Fedora systems in a container, which however still requires
7459 audit=0 to be passed on the kernel command line. Auditing in
7460 kernel and userspace is unfortunately still too broken in
7461 context of containers, hence we recommend compiling it out
7462 of the kernel or using audit=0. Hopefully this will be fixed
7463 one day for good in the kernel.
7465 * nspawn gained the new --bind= and --bind-ro= parameters to
7466 bind mount specific directories from the host into the
7469 * nspawn will now mount its own devpts file system instance
7470 into the container, in order not to leak pty devices from
7471 the host into the container.
7473 * systemd will now read the firmware boot time performance
7474 information from the EFI variables, if the used boot loader
7475 supports this, and takes it into account for boot performance
7476 analysis via "systemd-analyze". This is currently supported
7477 only in conjunction with Gummiboot, but could be supported
7478 by other boot loaders too. For details see:
7480 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
7482 * A new generator has been added that automatically mounts the
7483 EFI System Partition (ESP) to /boot, if that directory
7484 exists, is empty, and no other file system has been
7485 configured to be mounted there.
7487 * logind will now send out PrepareForSleep(false) out
7488 unconditionally, after coming back from suspend. This may be
7489 used by applications as asynchronous notification for
7490 system resume events.
7492 * "systemctl unlock-sessions" has been added, that allows
7493 unlocking the screens of all user sessions at once, similar
7494 to how "systemctl lock-sessions" already locked all users
7495 sessions. This is backed by a new D-Bus call UnlockSessions().
7497 * "loginctl seat-status" will now show the master device of a
7498 seat. (i.e. the device of a seat that needs to be around for
7499 the seat to be considered available, usually the graphics
7502 * tmpfiles gained a new "X" line type, that allows
7503 configuration of files and directories (with wildcards) that
7504 shall be excluded from automatic cleanup ("aging").
7506 * udev default rules set the device node permissions now only
7507 at "add" events, and do not change them any longer with a
7508 later "change" event.
7510 * The log messages for lid events and power/sleep keypresses
7511 now carry a message ID.
7513 * We now have a substantially larger unit test suite, but this
7514 continues to be work in progress.
7516 * udevadm hwdb gained a new --root= parameter to change the
7517 root directory to operate relative to.
7519 * logind will now issue a background sync() request to the kernel
7520 early at shutdown, so that dirty buffers are flushed to disk early
7521 instead of at the last moment, in order to optimize shutdown
7524 * A new bootctl tool has been added that is an interface for
7525 certain boot loader operations. This is currently a preview
7526 and is likely to be extended into a small mechanism daemon
7527 like timedated, localed, hostnamed, and can be used by
7528 graphical UIs to enumerate available boot options, and
7529 request boot into firmware operations.
7531 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
7532 the rest of the package. It also has been updated to work
7533 correctly in initrds.
7535 * polkit previously has been runtime optional, and is now also
7536 compile time optional via a configure switch.
7538 * systemd-analyze has been reimplemented in C. Also "systemctl
7539 dot" has moved into systemd-analyze.
7541 * "systemctl status" with no further parameters will now print
7542 the status of all active or failed units.
7544 * Operations such as "systemctl start" can now be executed
7545 with a new mode "--irreversible" which may be used to queue
7546 operations that cannot accidentally be reversed by a later
7547 job queuing. This is by default used to make shutdown
7548 requests more robust.
7550 * The Python API of systemd now gained a new module for
7551 reading journal files.
7553 * A new tool kernel-install has been added that can install
7554 kernel images according to the Boot Loader Specification:
7556 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
7558 * Boot time console output has been improved to provide
7559 animated boot time output for hanging jobs.
7561 * A new tool systemd-activate has been added which can be used
7562 to test socket activation with, directly from the command
7563 line. This should make it much easier to test and debug
7564 socket activation in daemons.
7566 * journalctl gained a new "--reverse" (or -r) option to show
7567 journal output in reverse order (i.e. newest line first).
7569 * journalctl gained a new "--pager-end" (or -e) option to jump
7570 to immediately jump to the end of the journal in the
7571 pager. This is only supported in conjunction with "less".
7573 * journalctl gained a new "--user-unit=" option, that works
7574 similarly to "--unit=" but filters for user units rather than
7577 * A number of unit files to ease adoption of systemd in
7578 initrds has been added. This moves some minimal logic from
7579 the various initrd implementations into systemd proper.
7581 * The journal files are now owned by a new group
7582 "systemd-journal", which exists specifically to allow access
7583 to the journal, and nothing else. Previously, we used the
7584 "adm" group for that, which however possibly covers more
7585 than just journal/log file access. This new group is now
7586 already used by systemd-journal-gatewayd to ensure this
7587 daemon gets access to the journal files and as little else
7588 as possible. Note that "make install" will also set FS ACLs
7589 up for /var/log/journal to give "adm" and "wheel" read
7590 access to it, in addition to "systemd-journal" which owns
7591 the journal files. We recommend that packaging scripts also
7592 add read access to "adm" + "wheel" to /var/log/journal, and
7593 all existing/future journal files. To normal users and
7594 administrators little changes, however packagers need to
7595 ensure to create the "systemd-journal" system group at
7596 package installation time.
7598 * The systemd-journal-gatewayd now runs as unprivileged user
7599 systemd-journal-gateway:systemd-journal-gateway. Packaging
7600 scripts need to create these system user/group at
7603 * timedated now exposes a new boolean property CanNTP that
7604 indicates whether a local NTP service is available or not.
7606 * systemd-detect-virt will now also detect xen PVs
7608 * The pstore file system is now mounted by default, if it is
7611 * In addition to the SELinux and IMA policies we will now also
7612 load SMACK policies at early boot.
7614 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
7615 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
7616 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
7617 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
7618 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
7619 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
7620 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
7621 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
7622 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
7623 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
7624 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
7625 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
7626 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
7627 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
7631 * Timer units now support calendar time events in addition to
7632 monotonic time events. That means you can now trigger a unit
7633 based on a calendar time specification such as "Thu,Fri
7634 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
7635 or fifth day of any month of the year 2013, given that it is
7636 a thursday or friday. This brings timer event support
7637 considerably closer to cron's capabilities. For details on
7638 the supported calendar time specification language see
7641 * udev now supports a number of different naming policies for
7642 network interfaces for predictable names, and a combination
7643 of these policies is now the default. Please see this wiki
7644 document for details:
7646 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
7648 * Auke Kok's bootchart implementation has been added to the
7649 systemd tree. It is an optional component that can graph the
7650 boot in quite some detail. It is one of the best bootchart
7651 implementations around and minimal in its code and
7654 * nss-myhostname has been integrated into the systemd source
7655 tree. nss-myhostname guarantees that the local hostname
7656 always stays resolvable via NSS. It has been a weak
7657 requirement of systemd-hostnamed since a long time, and
7658 since its code is actually trivial we decided to just
7659 include it in systemd's source tree. It can be turned off
7660 with a configure switch.
7662 * The read-ahead logic is now capable of properly detecting
7663 whether a btrfs file system is on SSD or rotating media, in
7664 order to optimize the read-ahead scheme. Previously, it was
7665 only capable of detecting this on traditional file systems
7668 * In udev, additional device properties are now read from the
7669 IAB in addition to the OUI database. Also, Bluetooth company
7670 identities are attached to the devices as well.
7672 * In service files %U may be used as specifier that is
7673 replaced by the configured user name of the service.
7675 * nspawn may now be invoked without a controlling TTY. This
7676 makes it suitable for invocation as its own service. This
7677 may be used to set up a simple containerized server system
7678 using only core OS tools.
7680 * systemd and nspawn can now accept socket file descriptors
7681 when they are started for socket activation. This enables
7682 implementation of socket activated nspawn
7683 containers. i.e. think about autospawning an entire OS image
7684 when the first SSH or HTTP connection is received. We expect
7685 that similar functionality will also be added to libvirt-lxc
7688 * journalctl will now suppress ANSI color codes when
7689 presenting log data.
7691 * systemctl will no longer show control group information for
7692 a unit if the control group is empty anyway.
7694 * logind can now automatically suspend/hibernate/shutdown the
7697 * /etc/machine-info and hostnamed now also expose the chassis
7698 type of the system. This can be used to determine whether
7699 the local system is a laptop, desktop, handset or
7700 tablet. This information may either be configured by the
7701 user/vendor or is automatically determined from ACPI and DMI
7702 information if possible.
7704 * A number of polkit actions are now bound together with "imply"
7705 rules. This should simplify creating UIs because many actions
7706 will now authenticate similar ones as well.
7708 * Unit files learnt a new condition ConditionACPower= which
7709 may be used to conditionalize a unit depending on whether an
7710 AC power source is connected or not, of whether the system
7711 is running on battery power.
7713 * systemctl gained a new "is-failed" verb that may be used in
7714 shell scripts and suchlike to check whether a specific unit
7715 is in the "failed" state.
7717 * The EnvironmentFile= setting in unit files now supports file
7718 globbing, and can hence be used to easily read a number of
7719 environment files at once.
7721 * systemd will no longer detect and recognize specific
7722 distributions. All distribution-specific #ifdeffery has been
7723 removed, systemd is now fully generic and
7724 distribution-agnostic. Effectively, not too much is lost as
7725 a lot of the code is still accessible via explicit configure
7726 switches. However, support for some distribution specific
7727 legacy configuration file formats has been dropped. We
7728 recommend distributions to simply adopt the configuration
7729 files everybody else uses now and convert the old
7730 configuration from packaging scripts. Most distributions
7731 already did that. If that's not possible or desirable,
7732 distributions are welcome to forward port the specific
7733 pieces of code locally from the git history.
7735 * When logging a message about a unit systemd will now always
7736 log the unit name in the message meta data.
7738 * localectl will now also discover system locale data that is
7739 not stored in locale archives, but directly unpacked.
7741 * logind will no longer unconditionally use framebuffer
7742 devices as seat masters, i.e. as devices that are required
7743 to be existing before a seat is considered preset. Instead,
7744 it will now look for all devices that are tagged as
7745 "seat-master" in udev. By default, framebuffer devices will
7746 be marked as such, but depending on local systems, other
7747 devices might be marked as well. This may be used to
7748 integrate graphics cards using closed source drivers (such
7749 as NVidia ones) more nicely into logind. Note however, that
7750 we recommend using the open source NVidia drivers instead,
7751 and no udev rules for the closed-source drivers will be
7752 shipped from us upstream.
7754 Contributions from: Adam Williamson, Alessandro Crismani, Auke
7755 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
7756 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
7757 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
7758 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
7759 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
7760 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
7761 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
7762 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
7763 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
7764 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
7765 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
7770 * udev gained support for loading additional device properties
7771 from an indexed database that is keyed by vendor/product IDs
7772 and similar device identifiers. For the beginning this
7773 "hwdb" is populated with data from the well-known PCI and
7774 USB database, but also includes PNP, ACPI and OID data. In
7775 the longer run this indexed database shall grow into
7776 becoming the one central database for non-essential
7777 userspace device metadata. Previously, data from the PCI/USB
7778 database was only attached to select devices, since the
7779 lookup was a relatively expensive operation due to O(n) time
7780 complexity (with n being the number of entries in the
7781 database). Since this is now O(1), we decided to add in this
7782 data for all devices where this is available, by
7783 default. Note that the indexed database needs to be rebuilt
7784 when new data files are installed. To achieve this you need
7785 to update your packaging scripts to invoke "udevadm hwdb
7786 --update" after installation of hwdb data files. For
7787 RPM-based distributions we introduced the new
7788 %udev_hwdb_update macro for this purpose.
7790 * The Journal gained support for the "Message Catalog", an
7791 indexed database to link up additional information with
7792 journal entries. For further details please check:
7794 https://www.freedesktop.org/wiki/Software/systemd/catalog
7796 The indexed message catalog database also needs to be
7797 rebuilt after installation of message catalog files. Use
7798 "journalctl --update-catalog" for this. For RPM-based
7799 distributions we introduced the %journal_catalog_update
7800 macro for this purpose.
7802 * The Python Journal bindings gained support for the standard
7803 Python logging framework.
7805 * The Journal API gained new functions for checking whether
7806 the underlying file system of a journal file is capable of
7807 properly reporting file change notifications, or whether
7808 applications that want to reflect journal changes "live"
7809 need to recheck journal files continuously in appropriate
7812 * It is now possible to set the "age" field for tmpfiles
7813 entries to 0, indicating that files matching this entry
7814 shall always be removed when the directories are cleaned up.
7816 * coredumpctl gained a new "gdb" verb which invokes gdb
7817 right-away on the selected coredump.
7819 * There's now support for "hybrid sleep" on kernels that
7820 support this, in addition to "suspend" and "hibernate". Use
7821 "systemctl hybrid-sleep" to make use of this.
7823 * logind's HandleSuspendKey= setting (and related settings)
7824 now gained support for a new "lock" setting to simply
7825 request the screen lock on all local sessions, instead of
7826 actually executing a suspend or hibernation.
7828 * systemd will now mount the EFI variables file system by
7831 * Socket units now gained support for configuration of the
7832 SMACK security label.
7834 * timedatectl will now output the time of the last and next
7835 daylight saving change.
7837 * We dropped support for various legacy and distro-specific
7838 concepts, such as insserv, early-boot SysV services
7839 (i.e. those for non-standard runlevels such as 'b' or 'S')
7840 or ArchLinux /etc/rc.conf support. We recommend the
7841 distributions who still need support this to either continue
7842 to maintain the necessary patches downstream, or find a
7843 different solution. (Talk to us if you have questions!)
7845 * Various systemd components will now bypass polkit checks for
7846 root and otherwise handle properly if polkit is not found to
7847 be around. This should fix most issues for polkit-less
7848 systems. Quite frankly this should have been this way since
7849 day one. It is absolutely our intention to make systemd work
7850 fine on polkit-less systems, and we consider it a bug if
7851 something does not work as it should if polkit is not around.
7853 * For embedded systems it is now possible to build udev and
7854 systemd without blkid and/or kmod support.
7856 * "systemctl switch-root" is now capable of switching root
7857 more than once. I.e. in addition to transitions from the
7858 initrd to the host OS it is now possible to transition to
7859 further OS images from the host. This is useful to implement
7860 offline updating tools.
7862 * Various other additions have been made to the RPM macros
7863 shipped with systemd. Use %udev_rules_update() after
7864 installing new udev rules files. %_udevhwdbdir,
7865 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
7866 %_sysctldir are now available which resolve to the right
7867 directories for packages to place various data files in.
7869 * journalctl gained the new --full switch (in addition to
7870 --all, to disable ellipsation for long messages.
7872 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
7873 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
7874 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
7875 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
7876 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
7877 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
7878 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
7879 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
7880 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
7884 * journalctl gained new --since= and --until= switches to
7885 filter by time. It also now supports nice filtering for
7886 units via --unit=/-u.
7888 * Type=oneshot services may use ExecReload= and do the
7891 * The journal daemon now supports time-based rotation and
7892 vacuuming, in addition to the usual disk-space based
7895 * The journal will now index the available field values for
7896 each field name. This enables clients to show pretty drop
7897 downs of available match values when filtering. The bash
7898 completion of journalctl has been updated
7899 accordingly. journalctl gained a new switch -F to list all
7900 values a certain field takes in the journal database.
7902 * More service events are now written as structured messages
7903 to the journal, and made recognizable via message IDs.
7905 * The timedated, localed and hostnamed mini-services which
7906 previously only provided support for changing time, locale
7907 and hostname settings from graphical DEs such as GNOME now
7908 also have a minimal (but very useful) text-based client
7909 utility each. This is probably the nicest way to changing
7910 these settings from the command line now, especially since
7911 it lists available options and is fully integrated with bash
7914 * There's now a new tool "systemd-coredumpctl" to list and
7915 extract coredumps from the journal.
7917 * We now install a README each in /var/log/ and
7918 /etc/rc.d/init.d explaining where the system logs and init
7919 scripts went. This hopefully should help folks who go to
7920 that dirs and look into the otherwise now empty void and
7921 scratch their heads.
7923 * When user-services are invoked (by systemd --user) the
7924 $MANAGERPID env var is set to the PID of systemd.
7926 * SIGRTMIN+24 when sent to a --user instance will now result
7927 in immediate termination of systemd.
7929 * gatewayd received numerous feature additions such as a
7930 "follow" mode, for live syncing and filtering.
7932 * browse.html now allows filtering and showing detailed
7933 information on specific entries. Keyboard navigation and
7934 mouse screen support has been added.
7936 * gatewayd/journalctl now supports HTML5/JSON
7937 Server-Sent-Events as output.
7939 * The SysV init script compatibility logic will now
7940 heuristically determine whether a script supports the
7941 "reload" verb, and only then make this available as
7944 * "systemctl status --follow" has been removed, use "journalctl
7947 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
7948 have been removed since they are hardly useful to be
7951 * And I'd like to take the opportunity to specifically mention
7952 Zbigniew for his great contributions. Zbigniew, you rock!
7954 Contributions from: Andrew Eikum, Christian Hesse, Colin
7955 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
7956 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
7957 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
7958 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
7959 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
7960 Jędrzejewski-Szmek, Сковорода Никита Андреевич
7964 * If /etc/vconsole.conf is non-existent or empty we will no
7965 longer load any console font or key map at boot by
7966 default. Instead the kernel defaults will be left
7967 intact. This is definitely the right thing to do, as no
7968 configuration should mean no configuration, and hard-coding
7969 font names that are different on all archs is probably a bad
7970 idea. Also, the kernel default key map and font should be
7971 good enough for most cases anyway, and mostly identical to
7972 the userspace fonts/key maps we previously overloaded them
7973 with. If distributions want to continue to default to a
7974 non-kernel font or key map they should ship a default
7975 /etc/vconsole.conf with the appropriate contents.
7977 Contributions from: Colin Walters, Daniel J Walsh, Dave
7978 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
7979 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
7983 * journalctl gained a new --cursor= switch to show entries
7984 starting from the specified location in the journal.
7986 * We now enforce a size limit on journal entry fields exported
7987 with "-o json" in journalctl. Fields larger than 4K will be
7988 assigned null. This can be turned off with --all.
7990 * An (optional) journal gateway daemon is now available as
7991 "systemd-journal-gatewayd.service". This service provides
7992 access to the journal via HTTP and JSON. This functionality
7993 will be used to implement live log synchronization in both
7994 pull and push modes, but has various other users too, such
7995 as easy log access for debugging of embedded devices. Right
7996 now it is already useful to retrieve the journal via HTTP:
7998 # systemctl start systemd-journal-gatewayd.service
7999 # wget http://localhost:19531/entries
8001 This will download the journal contents in a
8002 /var/log/messages compatible format. The same as JSON:
8004 # curl -H"Accept: application/json" http://localhost:19531/entries
8006 This service is also accessible via a web browser where a
8007 single static HTML5 app is served that uses the JSON logic
8008 to enable the user to do some basic browsing of the
8009 journal. This will be extended later on. Here's an example
8010 screenshot of this app in its current state:
8012 http://0pointer.de/public/journal-gatewayd
8014 Contributions from: Kay Sievers, Lennart Poettering, Robert
8015 Milasan, Tom Gundersen
8019 * The bash completion logic is now available for journalctl
8022 * We do not mount the "cpuset" controller anymore together with
8023 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
8024 started if no parameters are assigned to it. "cpuset" hence
8025 broke code that assumed it could create "cpu" groups and
8028 * journalctl -f will now subscribe to terminal size changes,
8029 and line break accordingly.
8031 Contributions from: Dave Reisner, Kay Sievers, Lennart
8032 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
8036 * nspawn will now create a symlink /etc/localtime in the
8037 container environment, copying the host's timezone
8038 setting. Previously this has been done via a bind mount, but
8039 since symlinks cannot be bind mounted this has now been
8040 changed to create/update the appropriate symlink.
8042 * journalctl -n's line number argument is now optional, and
8043 will default to 10 if omitted.
8045 * journald will now log the maximum size the journal files may
8046 take up on disk. This is particularly useful if the default
8047 built-in logic of determining this parameter from the file
8048 system size is used. Use "systemctl status
8049 systemd-journald.service" to see this information.
8051 * The multi-seat X wrapper tool has been stripped down. As X
8052 is now capable of enumerating graphics devices via udev in a
8053 seat-aware way the wrapper is not strictly necessary
8054 anymore. A stripped down temporary stop-gap is still shipped
8055 until the upstream display managers have been updated to
8056 fully support the new X logic. Expect this wrapper to be
8057 removed entirely in one of the next releases.
8059 * HandleSleepKey= in logind.conf has been split up into
8060 HandleSuspendKey= and HandleHibernateKey=. The old setting
8061 is not available anymore. X11 and the kernel are
8062 distinguishing between these keys and we should too. This
8063 also means the inhibition lock for these keys has been split
8066 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
8067 Poettering, Lukas Nykryn, Václav Pavlín
8071 * Whenever a unit changes state we will now log this to the
8072 journal and show along the unit's own log output in
8075 * ConditionPathIsMountPoint= can now properly detect bind
8076 mount points too. (Previously, a bind mount of one file
8077 system to another place in the same file system could not be
8078 detected as mount, since they shared struct stat's st_dev
8081 * We will now mount the cgroup controllers cpu, cpuacct,
8082 cpuset and the controllers net_cls, net_prio together by
8085 * nspawn containers will now have a virtualized boot
8086 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
8087 over with a randomized ID at container initialization). This
8088 has the effect of making "journalctl -b" do the right thing
8091 * The JSON output journal serialization has been updated not
8092 to generate "endless" list objects anymore, but rather one
8093 JSON object per line. This is more in line how most JSON
8094 parsers expect JSON objects. The new output mode
8095 "json-pretty" has been added to provide similar output, but
8096 neatly aligned for readability by humans.
8098 * We dropped all explicit sync() invocations in the shutdown
8099 code. The kernel does this implicitly anyway in the kernel
8100 reboot() syscall. halt(8)'s -n option is now a compatibility
8103 * We now support virtualized reboot() in containers, as
8104 supported by newer kernels. We will fall back to exit() if
8105 CAP_SYS_REBOOT is not available to the container. Also,
8106 nspawn makes use of this now and will actually reboot the
8107 container if the containerized OS asks for that.
8109 * journalctl will only show local log output by default
8110 now. Use --merge (-m) to show remote log output, too.
8112 * libsystemd-journal gained the new sd_journal_get_usage()
8113 call to determine the current disk usage of all journal
8114 files. This is exposed in the new "journalctl --disk-usage"
8117 * journald gained a new configuration setting SplitMode= in
8118 journald.conf which may be used to control how user journals
8119 are split off. See journald.conf(5) for details.
8121 * A new condition type ConditionFileNotEmpty= has been added.
8123 * tmpfiles' "w" lines now support file globbing, to write
8124 multiple files at once.
8126 * We added Python bindings for the journal submission
8127 APIs. More Python APIs for a number of selected APIs will
8128 likely follow. Note that we intend to add native bindings
8129 only for the Python language, as we consider it common
8130 enough to deserve bindings shipped within systemd. There are
8131 various projects outside of systemd that provide bindings
8132 for languages such as PHP or Lua.
8134 * Many conditions will now resolve specifiers such as %i. In
8135 addition, PathChanged= and related directives of .path units
8136 now support specifiers as well.
8138 * There's now a new RPM macro definition for the system preset
8141 * journald will now warn if it ca not forward a message to the
8142 syslog daemon because its socket is full.
8144 * timedated will no longer write or process /etc/timezone,
8145 except on Debian. As we do not support late mounted /usr
8146 anymore /etc/localtime always being a symlink is now safe,
8147 and hence the information in /etc/timezone is not necessary
8150 * logind will now always reserve one VT for a text getty (VT6
8151 by default). Previously if more than 6 X sessions where
8152 started they took up all the VTs with auto-spawned gettys,
8153 so that no text gettys were available anymore.
8155 * udev will now automatically inform the btrfs kernel logic
8156 about btrfs RAID components showing up. This should make
8157 simple hotplug based btrfs RAID assembly work.
8159 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
8160 (but not for its children which will stay at the kernel
8161 default). This should allow setups with a lot more listening
8164 * systemd will now always pass the configured timezone to the
8165 kernel at boot. timedated will do the same when the timezone
8168 * logind's inhibition logic has been updated. By default,
8169 logind will now handle the lid switch, the power and sleep
8170 keys all the time, even in graphical sessions. If DEs want
8171 to handle these events on their own they should take the new
8172 handle-power-key, handle-sleep-key and handle-lid-switch
8173 inhibitors during their runtime. A simple way to achieve
8174 that is to invoke the DE wrapped in an invocation of:
8176 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
8178 * Access to unit operations is now checked via SELinux taking
8179 the unit file label and client process label into account.
8181 * systemd will now notify the administrator in the journal
8182 when he over-mounts a non-empty directory.
8184 * There are new specifiers that are resolved in unit files,
8185 for the host name (%H), the machine ID (%m) and the boot ID
8188 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
8189 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
8190 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
8191 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
8192 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
8193 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
8194 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
8198 * Support for reading structured kernel messages from
8199 /dev/kmsg has now been added and is enabled by default.
8201 * Support for reading kernel messages from /proc/kmsg has now
8202 been removed. If you want kernel messages in the journal
8203 make sure to run a recent kernel (>= 3.5) that supports
8204 reading structured messages from /dev/kmsg (see
8205 above). /proc/kmsg is now exclusive property of classic
8206 syslog daemons again.
8208 * The libudev API gained the new
8209 udev_device_new_from_device_id() call.
8211 * The logic for file system namespace (ReadOnlyDirectory=,
8212 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
8213 require pivot_root() anymore. This means fewer temporary
8214 directories are created below /tmp for this feature.
8216 * nspawn containers will now see and receive all submounts
8217 made on the host OS below the root file system of the
8220 * Forward Secure Sealing is now supported for Journal files,
8221 which provide cryptographical sealing of journal files so
8222 that attackers cannot alter log history anymore without this
8223 being detectable. Lennart will soon post a blog story about
8224 this explaining it in more detail.
8226 * There are two new service settings RestartPreventExitStatus=
8227 and SuccessExitStatus= which allow configuration of exit
8228 status (exit code or signal) which will be excepted from the
8229 restart logic, resp. consider successful.
8231 * journalctl gained the new --verify switch that can be used
8232 to check the integrity of the structure of journal files and
8233 (if Forward Secure Sealing is enabled) the contents of
8236 * nspawn containers will now be run with /dev/stdin, /dev/fd/
8237 and similar symlinks pre-created. This makes running shells
8238 as container init process a lot more fun.
8240 * The fstab support can now handle PARTUUID= and PARTLABEL=
8243 * A new ConditionHost= condition has been added to match
8244 against the hostname (with globs) and machine ID. This is
8245 useful for clusters where a single OS image is used to
8246 provision a large number of hosts which shall run slightly
8247 different sets of services.
8249 * Services which hit the restart limit will now be placed in a
8252 Contributions from: Bertram Poettering, Dave Reisner, Huang
8253 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
8254 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
8258 * When running in --user mode systemd will now become a
8259 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
8260 tree a lot more organized.
8262 * A new PartOf= unit dependency type has been introduced that
8263 may be used to group services in a natural way.
8265 * "systemctl enable" may now be used to enable instances of
8268 * journalctl now prints error log levels in red, and
8269 warning/notice log levels in bright white. It also supports
8270 filtering by log level now.
8272 * cgtop gained a new -n switch (similar to top), to configure
8273 the maximum number of iterations to run for. It also gained
8274 -b, to run in batch mode (accepting no input).
8276 * The suffix ".service" may now be omitted on most systemctl
8277 command lines involving service unit names.
8279 * There's a new bus call in logind to lock all sessions, as
8280 well as a loginctl verb for it "lock-sessions".
8282 * libsystemd-logind.so gained a new call sd_journal_perror()
8283 that works similar to libc perror() but logs to the journal
8284 and encodes structured information about the error number.
8286 * /etc/crypttab entries now understand the new keyfile-size=
8289 * shutdown(8) now can send a (configurable) wall message when
8290 a shutdown is cancelled.
8292 * The mount propagation mode for the root file system will now
8293 default to "shared", which is useful to make containers work
8294 nicely out-of-the-box so that they receive new mounts from
8295 the host. This can be undone locally by running "mount
8296 --make-rprivate /" if needed.
8298 * The prefdm.service file has been removed. Distributions
8299 should maintain this unit downstream if they intend to keep
8300 it around. However, we recommend writing normal unit files
8301 for display managers instead.
8303 * Since systemd is a crucial part of the OS we will now
8304 default to a number of compiler switches that improve
8305 security (hardening) such as read-only relocations, stack
8306 protection, and suchlike.
8308 * The TimeoutSec= setting for services is now split into
8309 TimeoutStartSec= and TimeoutStopSec= to allow configuration
8310 of individual time outs for the start and the stop phase of
8313 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
8314 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
8315 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
8316 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
8317 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
8318 Gundersen, Zbigniew Jędrzejewski-Szmek
8322 * The journal and id128 C APIs are now fully documented as man
8325 * Extra safety checks have been added when transitioning from
8326 the initial RAM disk to the main system to avoid accidental
8329 * /etc/crypttab entries now understand the new keyfile-offset=
8332 * systemctl -t can now be used to filter by unit load state.
8334 * The journal C API gained the new sd_journal_wait() call to
8335 make writing synchronous journal clients easier.
8337 * journalctl gained the new -D switch to show journals from a
8340 * journalctl now displays a special marker between log
8341 messages of two different boots.
8343 * The journal is now explicitly flushed to /var via a service
8344 systemd-journal-flush.service, rather than implicitly simply
8345 by seeing /var/log/journal to be writable.
8347 * journalctl (and the journal C APIs) can now match for much
8348 more complex expressions, with alternatives and
8351 * When transitioning from the initial RAM disk to the main
8352 system we will now kill all processes in a killing spree to
8353 ensure no processes stay around by accident.
8355 * Three new specifiers may be used in unit files: %u, %h, %s
8356 resolve to the user name, user home directory resp. user
8357 shell. This is useful for running systemd user instances.
8359 * We now automatically rotate journal files if their data
8360 object hash table gets a fill level > 75%. We also size the
8361 hash table based on the configured maximum file size. This
8362 together should lower hash collisions drastically and thus
8363 speed things up a bit.
8365 * journalctl gained the new "--header" switch to introspect
8366 header data of journal files.
8368 * A new setting SystemCallFilters= has been added to services
8369 which may be used to apply blacklists or whitelists to
8370 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
8372 * nspawn gained a new --link-journal= switch (and quicker: -j)
8373 to link the container journal with the host. This makes it
8374 very easy to centralize log viewing on the host for all
8375 guests while still keeping the journal files separated.
8377 * Many bugfixes and optimizations
8379 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
8380 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
8381 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
8386 * Several tools now understand kernel command line arguments,
8387 which are only read when run in an initial RAM disk. They
8388 usually follow closely their normal counterparts, but are
8391 * There's a new tool to analyze the readahead files that are
8392 automatically generated at boot. Use:
8394 /usr/lib/systemd/systemd-readahead analyze /.readahead
8396 * We now provide an early debug shell on tty9 if this enabled. Use:
8398 systemctl enable debug-shell.service
8400 * All plymouth related units have been moved into the Plymouth
8401 package. Please make sure to upgrade your Plymouth version
8404 * systemd-tmpfiles now supports getting passed the basename of
8405 a configuration file only, in which case it will look for it
8406 in all appropriate directories automatically.
8408 * udevadm info now takes a /dev or /sys path as argument, and
8409 does the right thing. Example:
8411 udevadm info /dev/sda
8412 udevadm info /sys/class/block/sda
8414 * systemctl now prints a warning if a unit is stopped but a
8415 unit that might trigger it continues to run. Example: a
8416 service is stopped but the socket that activates it is left
8419 * "systemctl status" will now mention if the log output was
8420 shortened due to rotation since a service has been started.
8422 * The journal API now exposes functions to determine the
8423 "cutoff" times due to rotation.
8425 * journald now understands SIGUSR1 and SIGUSR2 for triggering
8426 immediately flushing of runtime logs to /var if possible,
8427 resp. for triggering immediate rotation of the journal
8430 * It is now considered an error if a service is attempted to
8431 be stopped that is not loaded.
8433 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
8435 * systemd-analyze now supports Python 3
8437 * tmpfiles now supports cleaning up directories via aging
8438 where the first level dirs are always kept around but
8439 directories beneath it automatically aged. This is enabled
8440 by prefixing the age field with '~'.
8442 * Seat objects now expose CanGraphical, CanTTY properties
8443 which is required to deal with very fast bootups where the
8444 display manager might be running before the graphics drivers
8445 completed initialization.
8447 * Seat objects now expose a State property.
8449 * We now include RPM macros for service enabling/disabling
8450 based on the preset logic. We recommend RPM based
8451 distributions to make use of these macros if possible. This
8452 makes it simpler to reuse RPM spec files across
8455 * We now make sure that the collected systemd unit name is
8456 always valid when services log to the journal via
8459 * There's a new man page kernel-command-line(7) detailing all
8460 command line options we understand.
8462 * The fstab generator may now be disabled at boot by passing
8463 fstab=0 on the kernel command line.
8465 * A new kernel command line option modules-load= is now understood
8466 to load a specific kernel module statically, early at boot.
8468 * Unit names specified on the systemctl command line are now
8469 automatically escaped as needed. Also, if file system or
8470 device paths are specified they are automatically turned
8471 into the appropriate mount or device unit names. Example:
8473 systemctl status /home
8474 systemctl status /dev/sda
8476 * The SysVConsole= configuration option has been removed from
8477 system.conf parsing.
8479 * The SysV search path is no longer exported on the D-Bus
8482 * The Names= option has been removed from unit file parsing.
8484 * There's a new man page bootup(7) detailing the boot process.
8486 * Every unit and every generator we ship with systemd now
8487 comes with full documentation. The self-explanatory boot is
8490 * A couple of services gained "systemd-" prefixes in their
8491 name if they wrap systemd code, rather than only external
8492 code. Among them fsck@.service which is now
8493 systemd-fsck@.service.
8495 * The HaveWatchdog property has been removed from the D-Bus
8498 * systemd.confirm_spawn= on the kernel command line should now
8501 * There's a new man page crypttab(5) which details all options
8502 we actually understand.
8504 * systemd-nspawn gained a new --capability= switch to pass
8505 additional capabilities to the container.
8507 * timedated will now read known NTP implementation unit names
8508 from /usr/lib/systemd/ntp-units.d/*.list,
8509 systemd-timedated-ntp.target has been removed.
8511 * journalctl gained a new switch "-b" that lists log data of
8512 the current boot only.
8514 * The notify socket is in the abstract namespace again, in
8515 order to support daemons which chroot() at start-up.
8517 * There is a new Storage= configuration option for journald
8518 which allows configuration of where log data should go. This
8519 also provides a way to disable journal logging entirely, so
8520 that data collected is only forwarded to the console, the
8521 kernel log buffer or another syslog implementation.
8523 * Many bugfixes and optimizations
8525 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
8526 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
8527 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
8528 Shawn Landden, Tom Gundersen
8532 * "systemctl help <unit>" now shows the man page if one is
8535 * Several new man pages have been added.
8537 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
8538 MaxLevelConsole= can now be specified in
8539 journald.conf. These options allow reducing the amount of
8540 data stored on disk or forwarded by the log level.
8542 * TimerSlackNSec= can now be specified in system.conf for
8543 PID1. This allows system-wide power savings.
8545 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
8546 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
8551 * logind is now capable of (optionally) handling power and
8552 sleep keys as well as the lid switch.
8554 * journalctl now understands the syntax "journalctl
8555 /usr/bin/avahi-daemon" to get all log output of a specific
8558 * CapabilityBoundingSet= in system.conf now also influences
8559 the capability bound set of usermode helpers of the kernel.
8561 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
8562 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
8563 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
8564 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
8568 * Note that we skipped 139 releases here in order to set the
8569 new version to something that is greater than both udev's
8570 and systemd's most recent version number.
8572 * udev: all udev sources are merged into the systemd source tree now.
8573 All future udev development will happen in the systemd tree. It
8574 is still fully supported to use the udev daemon and tools without
8575 systemd running, like in initramfs or other init systems. Building
8576 udev though, will require the *build* of the systemd tree, but
8577 udev can be properly *run* without systemd.
8579 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
8580 should be used to create dead device nodes as workarounds for broken
8583 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
8584 no longer supported. udev_monitor_new_from_netlink() needs to be
8585 used to subscribe to events.
8587 * udev: when udevd is started by systemd, processes which are left
8588 behind by forking them off of udev rules, are unconditionally cleaned
8589 up and killed now after the event handling has finished. Services or
8590 daemons must be started as systemd services. Services can be
8591 pulled-in by udev to get started, but they can no longer be directly
8592 forked by udev rules.
8594 * udev: the daemon binary is called systemd-udevd now and installed
8595 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
8596 to adapt to that, create symlink, or rename the binary after building
8599 * libudev no longer provides these symbols:
8600 udev_monitor_from_socket()
8601 udev_queue_get_failed_list_entry()
8602 udev_get_{dev,sys,run}_path()
8603 The versions number was bumped and symbol versioning introduced.
8605 * systemd-loginctl and systemd-journalctl have been renamed
8606 to loginctl and journalctl to match systemctl.
8608 * The config files: /etc/systemd/systemd-logind.conf and
8609 /etc/systemd/systemd-journald.conf have been renamed to
8610 logind.conf and journald.conf. Package updates should rename
8611 the files to the new names on upgrade.
8613 * For almost all files the license is now LGPL2.1+, changed
8614 from the previous GPL2.0+. Exceptions are some minor stuff
8615 of udev (which will be changed to LGPL2.1 eventually, too),
8616 and the MIT licensed sd-daemon.[ch] library that is suitable
8617 to be used as drop-in files.
8619 * systemd and logind now handle system sleep states, in
8620 particular suspending and hibernating.
8622 * logind now implements a sleep/shutdown/idle inhibiting logic
8623 suitable for a variety of uses. Soonishly Lennart will blog
8624 about this in more detail.
8626 * var-run.mount and var-lock.mount are no longer provided
8627 (which previously bind mounted these directories to their new
8628 places). Distributions which have not converted these
8629 directories to symlinks should consider stealing these files
8630 from git history and add them downstream.
8632 * We introduced the Documentation= field for units and added
8633 this to all our shipped units. This is useful to make it
8634 easier to explore the boot and the purpose of the various
8637 * All smaller setup units (such as
8638 systemd-vconsole-setup.service) now detect properly if they
8639 are run in a container and are skipped when
8640 appropriate. This guarantees an entirely noise-free boot in
8641 Linux container environments such as systemd-nspawn.
8643 * A framework for implementing offline system updates is now
8644 integrated, for details see:
8645 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
8647 * A new service type Type=idle is available now which helps us
8648 avoiding ugly interleaving of getty output and boot status
8651 * There's now a system-wide CapabilityBoundingSet= option to
8652 globally reduce the set of capabilities for the
8653 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
8654 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
8655 even CAP_NET_ADMIN system-wide for secure systems.
8657 * There are now system-wide DefaultLimitXXX= options to
8658 globally change the defaults of the various resource limits
8659 for all units started by PID 1.
8661 * Harald Hoyer's systemd test suite has been integrated into
8662 systemd which allows easy testing of systemd builds in qemu
8663 and nspawn. (This is really awesome! Ask us for details!)
8665 * The fstab parser is now implemented as generator, not inside
8668 * systemctl will now warn you if .mount units generated from
8669 /etc/fstab are out of date due to changes in fstab that
8670 have not been read by systemd yet.
8672 * systemd is now suitable for usage in initrds. Dracut has
8673 already been updated to make use of this. With this in place
8674 initrds get a slight bit faster but primarily are much
8675 easier to introspect and debug since "systemctl status" in
8676 the host system can be used to introspect initrd services,
8677 and the journal from the initrd is kept around too.
8679 * systemd-delta has been added, a tool to explore differences
8680 between user/admin configuration and vendor defaults.
8682 * PrivateTmp= now affects both /tmp and /var/tmp.
8684 * Boot time status messages are now much prettier and feature
8685 proper english language. Booting up systemd has never been
8688 * Read-ahead pack files now include the inode number of all
8689 files to pre-cache. When the inode changes the pre-caching
8690 is not attempted. This should be nicer to deal with updated
8691 packages which might result in changes of read-ahead
8694 * We now temporaritly lower the kernel's read_ahead_kb variable
8695 when collecting read-ahead data to ensure the kernel's
8696 built-in read-ahead does not add noise to our measurements
8697 of necessary blocks to pre-cache.
8699 * There's now RequiresMountsFor= to add automatic dependencies
8700 for all mounts necessary for a specific file system path.
8702 * MountAuto= and SwapAuto= have been removed from
8703 system.conf. Mounting file systems at boot has to take place
8706 * nspawn now learned a new switch --uuid= to set the machine
8707 ID on the command line.
8709 * nspawn now learned the -b switch to automatically search
8712 * vt102 is now the default TERM for serial TTYs, upgraded from
8715 * systemd-logind now works on VT-less systems.
8717 * The build tree has been reorganized. The individual
8718 components now have directories of their own.
8720 * A new condition type ConditionPathIsReadWrite= is now available.
8722 * nspawn learned the new -C switch to create cgroups for the
8723 container in other hierarchies.
8725 * We now have support for hardware watchdogs, configurable in
8728 * The scheduled shutdown logic now has a public API.
8730 * We now mount /tmp as tmpfs by default, but this can be
8731 masked and /etc/fstab can override it.
8733 * Since udisks does not make use of /media anymore we are not
8734 mounting a tmpfs on it anymore.
8736 * journalctl gained a new --local switch to only interleave
8737 locally generated journal files.
8739 * We can now load the IMA policy at boot automatically.
8741 * The GTK tools have been split off into a systemd-ui.
8743 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
8744 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
8745 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
8746 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
8747 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
8748 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
8749 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
8750 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
8751 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
8756 * This is mostly a bugfix release
8758 * Support optional initialization of the machine ID from the
8759 KVM or container configured UUID.
8761 * Support immediate reboots with "systemctl reboot -ff"
8763 * Show /etc/os-release data in systemd-analyze output
8765 * Many bugfixes for the journal, including endianness fixes and
8766 ensuring that disk space enforcement works
8768 * sd-login.h is C++ compatible again
8770 * Extend the /etc/os-release format on request of the Debian
8773 * We now refuse non-UTF8 strings used in various configuration
8774 and unit files. This is done to ensure we do not pass invalid
8775 data over D-Bus or expose it elsewhere.
8777 * Register Mimo USB Screens as suitable for automatic seat
8780 * Read SELinux client context from journal clients in a race
8783 * Reorder configuration file lookup order. /etc now always
8784 overrides /run in order to allow the administrator to always
8785 and unconditionally override vendor-supplied or
8786 automatically generated data.
8788 * The various user visible bits of the journal now have man
8789 pages. We still lack man pages for the journal API calls
8792 * We now ship all man pages in HTML format again in the
8795 Contributions from: Dave Reisner, Dirk Eibach, Frederic
8796 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
8797 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
8802 * This is mostly a bugfix release
8804 * systems lacking /etc/os-release are no longer supported.
8806 * Various functionality updates to libsystemd-login.so
8808 * Track class of PAM logins to distinguish greeters from
8811 Contributions from: Kay Sievers, Lennart Poettering, Michael
8816 * This is an important bugfix release for v41.
8818 * Building man pages is now optional which should be useful
8819 for those building systemd from git but unwilling to install
8822 * Watchdog support for supervising services is now usable. In
8823 a future release support for hardware watchdogs
8824 (i.e. /dev/watchdog) will be added building on this.
8826 * Service start rate limiting is now configurable and can be
8827 turned off per service. When a start rate limit is hit a
8828 reboot can automatically be triggered.
8830 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
8832 Contributions from: Benjamin Franzke, Bill Nottingham,
8833 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
8834 Schmidt, Michał Górny, Piotr Drąg
8838 * The systemd binary is installed /usr/lib/systemd/systemd now;
8839 An existing /sbin/init symlink needs to be adapted with the
8842 * The code that loads kernel modules has been ported to invoke
8843 libkmod directly, instead of modprobe. This means we do not
8844 support systems with module-init-tools anymore.
8846 * Watchdog support is now already useful, but still not
8849 * A new kernel command line option systemd.setenv= is
8850 understood to set system wide environment variables
8851 dynamically at boot.
8853 * We now limit the set of capabilities of systemd-journald.
8855 * We now set SIGPIPE to ignore by default, since it only is
8856 useful in shell pipelines, and has little use in general
8857 code. This can be disabled with IgnoreSIPIPE=no in unit
8860 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
8861 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
8866 * This is mostly a bugfix release
8868 * We now expose the reason why a service failed in the
8869 "Result" D-Bus property.
8871 * Rudimentary service watchdog support (will be completed over
8872 the next few releases.)
8874 * When systemd forks off in order execute some service we will
8875 now immediately changes its argv[0] to reflect which process
8876 it will execute. This is useful to minimize the time window
8877 with a generic argv[0], which makes bootcharts more useful
8879 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
8880 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
8881 Mike Kazantsev, Ray Strode
8885 * This is mostly a test release, but incorporates many
8888 * New systemd-cgtop tool to show control groups by their
8891 * Linking against libacl for ACLs is optional again. If
8892 disabled, support tracking device access for active logins
8893 goes becomes unavailable, and so does access to the user
8894 journals by the respective users.
8896 * If a group "adm" exists, journal files are automatically
8897 owned by them, thus allow members of this group full access
8898 to the system journal as well as all user journals.
8900 * The journal now stores the SELinux context of the logging
8901 client for all entries.
8903 * Add C++ inclusion guards to all public headers
8905 * New output mode "cat" in the journal to print only text
8906 messages, without any meta data like date or time.
8908 * Include tiny X server wrapper as a temporary stop-gap to
8909 teach XOrg udev display enumeration. This is used by display
8910 managers such as gdm, and will go away as soon as XOrg
8911 learned native udev hotplugging for display devices.
8913 * Add new systemd-cat tool for executing arbitrary programs
8914 with STDERR/STDOUT connected to the journal. Can also act as
8915 BSD logger replacement, and does so by default.
8917 * Optionally store all locally generated coredumps in the
8918 journal along with meta data.
8920 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
8921 writing short strings to files (for usage for /sys), and for
8922 creating symlinks, character and block device nodes.
8924 * New unit file option ControlGroupPersistent= to make cgroups
8925 persistent, following the mechanisms outlined in
8926 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
8928 * Support multiple local RTCs in a sane way
8930 * No longer monopolize IO when replaying readahead data on
8931 rotating disks, since we might starve non-file-system IO to
8932 death, since fanotify() will not see accesses done by blkid,
8935 * Do not show kernel threads in systemd-cgls anymore, unless
8936 requested with new -k switch.
8938 Contributions from: Dan Horák, Kay Sievers, Lennart
8939 Poettering, Michal Schmidt
8943 * This is mostly a test release, but incorporates many
8946 * The git repository moved to:
8947 git://anongit.freedesktop.org/systemd/systemd
8948 ssh://git.freedesktop.org/git/systemd/systemd
8950 * First release with the journal
8951 http://0pointer.de/blog/projects/the-journal.html
8953 * The journal replaces both systemd-kmsg-syslogd and
8954 systemd-stdout-bridge.
8956 * New sd_pid_get_unit() API call in libsystemd-logind
8958 * Many systemadm clean-ups
8960 * Introduce remote-fs-pre.target which is ordered before all
8961 remote mounts and may be used to start services before all
8964 * Added Mageia support
8966 * Add bash completion for systemd-loginctl
8968 * Actively monitor PID file creation for daemons which exit in
8969 the parent process before having finished writing the PID
8970 file in the daemon process. Daemons which do this need to be
8971 fixed (i.e. PID file creation must have finished before the
8972 parent exits), but we now react a bit more gracefully to them.
8974 * Add colourful boot output, mimicking the well-known output
8975 of existing distributions.
8977 * New option PassCredentials= for socket units, for
8978 compatibility with a recent kernel ABI breakage.
8980 * /etc/rc.local is now hooked in via a generator binary, and
8981 thus will no longer act as synchronization point during
8984 * systemctl list-unit-files now supports --root=.
8986 * systemd-tmpfiles now understands two new commands: z, Z for
8987 relabelling files according to the SELinux database. This is
8988 useful to apply SELinux labels to specific files in /sys,
8991 * Output of SysV services is now forwarded to both the console
8992 and the journal by default, not only just the console.
8994 * New man pages for all APIs from libsystemd-login.
8996 * The build tree got reorganized and the build system is a
8997 lot more modular allowing embedded setups to specifically
8998 select the components of systemd they are interested in.
9000 * Support for Linux systems lacking the kernel VT subsystem is
9003 * configure's --with-rootdir= got renamed to
9004 --with-rootprefix= to follow the naming used by udev and
9007 * Unless specified otherwise we will now install to /usr instead
9008 of /usr/local by default.
9010 * Processes with '@' in argv[0][0] are now excluded from the
9011 final shut-down killing spree, following the logic explained
9013 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
9015 * All processes remaining in a service cgroup when we enter
9016 the START or START_PRE states are now killed with
9017 SIGKILL. That means it is no longer possible to spawn
9018 background processes from ExecStart= lines (which was never
9019 supported anyway, and bad style).
9021 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
9022 reloading of units together.
9024 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
9025 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
9026 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
9027 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
9028 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek