1 systemd System and Service Manager
3 CHANGES WITH 241 in spe:
5 * The default locale can now be configured at compile time. Otherwise,
6 a suitable default will be selected automatically (one of C.UTF-8,
9 * The version string shown by systemd and other tools now includes the
10 git commit hash when built from git. An override may be specified
11 during compilation, which is intended to be used by distributions to
12 include the package release information.
14 * systemd-cat can now filter standard input and standard error streams
15 for different syslog priorities using the new --stderr-priority=
18 * systemd-journald and systemd-journal-remote reject entries which
19 contain too many fields (CVE-2018-16865) and set limits on the
20 process' command line length (CVE-2018-16864).
22 * $DBUS_SESSION_BUS_ADDRESS environment variable is set by pam_systemd
25 * A new network device NamePolicy "keep" is implemented for link files,
26 and used by default in 99-default.link (the fallback configuration
27 provided by systemd). With this policy, if the network device name
28 was already set by userspace, the device will not be renamed again.
29 This matches the naming scheme that was implemented before
30 systemd-240. If naming-scheme < 240 is specified, the "keep" policy
31 is also enabled by default, even if not specified. Effectively, this
32 means that if naming-scheme >= 240 is specified, network devices will
33 be renamed according to the configuration, even if they have been
34 renamed already, if "keep" is not specified as the naming policy in
35 the .link file. The 99-default.link file provided by systemd includes
36 "keep" for backwards compatibility, but it is recommended for user
37 installed .link files to *not* include it.
39 The "kernel" policy, which keeps kernel names declared to be
40 "persistent", now works again as documented.
42 * kernel-install script now optionally takes a path to an initrd file,
43 and passes it to all plugins.
45 * The mincore() system call has been dropped from the @system-service
46 system call filter group, as it is pretty exotic and may potentially
47 used for side-channel attacks.
49 * -fPIE is dropped from compiler and linker options. Please specify
50 -Db_pie=true option to meson to build position-independent
51 executables. Note that the meson option is supported since meson-0.49.
53 * The fs.protected_regular and fs.protected_fifos sysctls, which were
54 added in Linux 4.19 to make some data spoofing attacks harder, are
55 now enabled by default. While this will hopefully improve the
56 security of most installations, it is technically a backwards
57 incompatible change; to disable these sysctls again, place the
58 following lines in /etc/sysctl.d/60-protected.conf or a similar file:
60 fs.protected_regular = 0
61 fs.protected_fifos = 0
63 Note that the similar hardlink and symlink protection has been
64 enabled since v199, and may be disabled likewise.
68 * NoNewPrivileges=yes has been set for all long-running services
69 implemented by systemd. Previously, this was problematic due to
70 SELinux (as this would also prohibit the transition from PID1's label
71 to the service's label). This restriction has since been lifted, but
72 an SELinux policy update is required.
73 (See e.g. https://github.com/fedora-selinux/selinux-policy/pull/234.)
75 * DynamicUser=yes is dropped from systemd-networkd.service,
76 systemd-resolved.service and systemd-timesyncd.service, which was
77 enabled in v239 for systemd-networkd.service and systemd-resolved.service,
78 and since v236 for systemd-timesyncd.service. The users and groups
79 systemd-network, systemd-resolve and systemd-timesync are created
80 by systemd-sysusers again. Distributors or system administrators
81 may need to create these users and groups if they not exist (or need
82 to re-enable DynamicUser= for those units) while upgrading systemd.
83 Also, the clock file for systemd-timesyncd may need to move from
84 /var/lib/private/systemd/timesync/clock to /var/lib/systemd/timesync/clock.
86 * When unit files are loaded from disk, previously systemd would
87 sometimes (depending on the unit loading order) load units from the
88 target path of symlinks in .wants/ or .requires/ directories of other
89 units. This meant that unit could be loaded from different paths
90 depending on whether the unit was requested explicitly or as a
91 dependency of another unit, not honouring the priority of directories
92 in search path. It also meant that it was possible to successfully
93 load and start units which are not found in the unit search path, as
94 long as they were requested as a dependency and linked to from
95 .wants/ or .requires/. The target paths of those symlinks are not
96 used for loading units anymore and the unit file must be found in
99 * A new service type has been added: Type=exec. It's very similar to
100 Type=simple but ensures the service manager will wait for both fork()
101 and execve() of the main service binary to complete before proceeding
102 with follow-up units. This is primarily useful so that the manager
103 propagates any errors in the preparation phase of service execution
104 back to the job that requested the unit to be started. For example,
105 consider a service that has ExecStart= set to a file system binary
106 that doesn't exist. With Type=simple starting the unit would be
107 considered instantly successful, as only fork() has to complete
108 successfully and the manager does not wait for execve(), and hence
109 its failure is seen "too late". With the new Type=exec service type
110 starting the unit will fail, as the manager will wait for the
111 execve() and notice its failure, which is then propagated back to the
114 NOTE: with the next release 241 of systemd we intend to change the
115 systemd-run tool to default to Type=exec for transient services
116 started by it. This should be mostly safe, but in specific corner
117 cases might result in problems, as the systemd-run tool will then
118 block on NSS calls (such as user name look-ups due to User=) done
119 between the fork() and execve(), which under specific circumstances
120 might cause problems. It is recommended to specify "-p Type=simple"
121 explicitly in the few cases where this applies. For regular,
122 non-transient services (i.e. those defined with unit files on disk)
123 we will continue to default to Type=simple.
125 * The Linux kernel's current default RLIMIT_NOFILE resource limit for
126 userspace processes is set to 1024 (soft) and 4096
127 (hard). Previously, systemd passed this on unmodified to all
128 processes it forked off. With this systemd release the hard limit
129 systemd passes on is increased to 512K, overriding the kernel's
130 defaults and substantially increasing the number of simultaneous file
131 descriptors unprivileged userspace processes can allocate. Note that
132 the soft limit remains at 1024 for compatibility reasons: the
133 traditional UNIX select() call cannot deal with file descriptors >=
134 1024 and increasing the soft limit globally might thus result in
135 programs unexpectedly allocating a high file descriptor and thus
136 failing abnormally when attempting to use it with select() (of
137 course, programs shouldn't use select() anymore, and prefer
138 poll()/epoll, but the call unfortunately remains undeservedly popular
139 at this time). This change reflects the fact that file descriptor
140 handling in the Linux kernel has been optimized in more recent
141 kernels and allocating large numbers of them should be much cheaper
142 both in memory and in performance than it used to be. Programs that
143 want to take benefit of the increased limit have to "opt-in" into
144 high file descriptors explicitly by raising their soft limit. Of
145 course, when they do that they must acknowledge that they cannot use
146 select() anymore (and neither can any shared library they use — or
147 any shared library used by any shared library they use and so on).
148 Which default hard limit is most appropriate is of course hard to
149 decide. However, given reports that ~300K file descriptors are used
150 in real-life applications we believe 512K is sufficiently high as new
151 default for now. Note that there are also reports that using very
152 high hard limits (e.g. 1G) is problematic: some software allocates
153 large arrays with one element for each potential file descriptor
154 (Java, …) — a high hard limit thus triggers excessively large memory
155 allocations in these applications. Hopefully, the new default of 512K
156 is a good middle ground: higher than what real-life applications
157 currently need, and low enough for avoid triggering excessively large
158 allocations in problematic software. (And yes, somebody should fix
161 * The fs.nr_open and fs.file-max sysctls are now automatically bumped
162 to the highest possible values, as separate accounting of file
163 descriptors is no longer necessary, as memcg tracks them correctly as
164 part of the memory accounting anyway. Thus, from the four limits on
165 file descriptors currently enforced (fs.file-max, fs.nr_open,
166 RLIMIT_NOFILE hard, RLIMIT_NOFILE soft) we turn off the first two,
167 and keep only the latter two. A set of build-time options
168 (-Dbump-proc-sys-fs-file-max=no and -Dbump-proc-sys-fs-nr-open=no)
169 has been added to revert this change in behaviour, which might be
170 an option for systems that turn off memcg in the kernel.
172 * When no /etc/locale.conf file exists (and hence no locale settings
173 are in place), systemd will now use the "C.UTF-8" locale by default,
174 and set LANG= to it. This locale is supported by various
175 distributions including Fedora, with clear indications that upstream
176 glibc is going to make it available too. This locale enables UTF-8
177 mode by default, which appears appropriate for 2018.
179 * The "net.ipv4.conf.all.rp_filter" sysctl will now be set to 2 by
180 default. This effectively switches the RFC3704 Reverse Path filtering
181 from Strict mode to Loose mode. This is more appropriate for hosts
182 that have multiple links with routes to the same networks (e.g.
183 a client with a Wi-Fi and Ethernet both connected to the internet).
185 Consult the kernel documentation for details on this sysctl:
186 https://www.kernel.org/doc/Documentation/networking/ip-sysctl.txt
188 * CPUAccounting=yes no longer enables the CPU controller when using
189 kernel 4.15+ and the unified cgroup hierarchy, as required accounting
190 statistics are now provided independently from the CPU controller.
192 * Support for disabling a particular cgroup controller within a sub-tree
193 has been added through the DisableControllers= directive.
195 * cgroup_no_v1=all on the kernel command line now also implies
196 using the unified cgroup hierarchy, unless one explicitly passes
197 systemd.unified_cgroup_hierarchy=0 on the kernel command line.
199 * The new "MemoryMin=" unit file property may now be used to set the
200 memory usage protection limit of processes invoked by the unit. This
201 controls the cgroup v2 memory.min attribute. Similarly, the new
202 "IODeviceLatencyTargetSec=" property has been added, wrapping the new
203 cgroup v2 io.latency cgroup property for configuring per-service I/O
206 * systemd now supports the cgroup v2 devices BPF logic, as counterpart
207 to the cgroup v1 "devices" cgroup controller.
209 * systemd-escape now is able to combine --unescape with --template. It
210 also learnt a new option --instance for extracting and unescaping the
211 instance part of a unit name.
213 * sd-bus now provides the sd_bus_message_readv() which is similar to
214 sd_bus_message_read() but takes a va_list object. The pair
215 sd_bus_set_method_call_timeout() and sd_bus_get_method_call_timeout()
216 has been added for configuring the default method call timeout to
217 use. sd_bus_error_move() may be used to efficiently move the contents
218 from one sd_bus_error structure to another, invalidating the
219 source. sd_bus_set_close_on_exit() and sd_bus_get_close_on_exit() may
220 be used to control whether a bus connection object is automatically
221 flushed when an sd-event loop is exited.
223 * When processing classic BSD syslog log messages, journald will now
224 save the original time-stamp string supplied in the new
225 SYSLOG_TIMESTAMP= journal field. This permits consumers to
226 reconstruct the original BSD syslog message more correctly.
228 * StandardOutput=/StandardError= in service files gained support for
229 new "append:…" parameters, for connecting STDOUT/STDERR of a service
230 to a file, and appending to it.
232 * The signal to use as last step of killing of unit processes is now
233 configurable. Previously it was hard-coded to SIGKILL, which may now
234 be overridden with the new KillSignal= setting. Note that this is the
235 signal used when regular termination (i.e. SIGTERM) does not suffice.
236 Similarly, the signal used when aborting a program in case of a
237 watchdog timeout may now be configured too (WatchdogSignal=).
239 * The XDG_SESSION_DESKTOP environment variable may now be configured in
240 the pam_systemd argument line, using the new desktop= switch. This is
241 useful to initialize it properly from a display manager without
242 having to touch C code.
244 * Most configuration options that previously accepted percentage values
245 now also accept permille values with the '‰' suffix (instead of '%').
247 * systemd-resolved may now optionally use OpenSSL instead of GnuTLS for
250 * systemd-resolved's configuration file resolved.conf gained a new
251 option ReadEtcHosts= which may be used to turn off processing and
252 honoring /etc/hosts entries.
254 * The "--wait" switch may now be passed to "systemctl
255 is-system-running", in which case the tool will synchronously wait
256 until the system finished start-up.
258 * hostnamed gained a new bus call to determine the DMI product UUID.
260 * On x86-64 systemd will now prefer using the RDRAND processor
261 instruction over /dev/urandom whenever it requires randomness that
262 neither has to be crypto-grade nor should be reproducible. This
263 should substantially reduce the amount of entropy systemd requests
264 from the kernel during initialization on such systems, though not
265 reduce it to zero. (Why not zero? systemd still needs to allocate
266 UUIDs and such uniquely, which require high-quality randomness.)
268 * networkd gained support for Foo-Over-UDP, ERSPAN and ISATAP
269 tunnels. It also gained a new option ForceDHCPv6PDOtherInformation=
270 for forcing the "Other Information" bit in IPv6 RA messages. The
271 bonding logic gained four new options AdActorSystemPriority=,
272 AdUserPortKey=, AdActorSystem= for configuring various 802.3ad
273 aspects, and DynamicTransmitLoadBalancing= for enabling dynamic
274 shuffling of flows. The tunnel logic gained a new
275 IPv6RapidDeploymentPrefix= option for configuring IPv6 Rapid
276 Deployment. The policy rule logic gained four new options IPProtocol=,
277 SourcePort= and DestinationPort=, InvertRule=. The bridge logic gained
278 support for the MulticastToUnicast= option. networkd also gained
279 support for configuring static IPv4 ARP or IPv6 neighbor entries.
281 * .preset files (as read by 'systemctl preset') may now be used to
282 instantiate services.
284 * /etc/crypttab now understands the sector-size= option to configure
285 the sector size for an encrypted partition.
287 * Key material for encrypted disks may now be placed on a formatted
288 medium, and referenced from /etc/crypttab by the UUID of the file
289 system, followed by "=" suffixed by the path to the key file.
291 * The "collect" udev component has been removed without replacement, as
292 it is neither used nor maintained.
294 * When the RuntimeDirectory=, StateDirectory=, CacheDirectory=,
295 LogsDirectory=, ConfigurationDirectory= settings are used in a
296 service the executed processes will now receive a set of environment
297 variables containing the full paths of these directories.
298 Specifically, RUNTIME_DIRECTORY=, STATE_DIRECTORY, CACHE_DIRECTORY,
299 LOGS_DIRECTORY, CONFIGURATION_DIRECTORY are now set if these options
300 are used. Note that these options may be used multiple times per
301 service in which case the resulting paths will be concatenated and
304 * Predictable interface naming has been extended to cover InfiniBand
305 NICs. They will be exposed with an "ib" prefix.
307 * tmpfiles.d/ line types may now be suffixed with a '-' character, in
308 which case the respective line failing is ignored.
310 * .link files may now be used to configure the equivalent to the
311 "ethtool advertise" commands.
313 * The sd-device.h and sd-hwdb.h APIs are now exported, as an
314 alternative to libudev.h. Previously, the latter was just an internal
315 wrapper around the former, but now these two APIs are exposed
318 * sd-id128.h gained a new function sd_id128_get_boot_app_specific()
319 which calculates an app-specific boot ID similar to how
320 sd_id128_get_machine_app_specific() generates an app-specific machine
323 * A new tool systemd-id128 has been added that can be used to determine
324 and generate various 128bit IDs.
326 * /etc/os-release gained two new standardized fields DOCUMENTATION_URL=
329 * systemd-hibernate-resume-generator will now honor the "noresume"
330 kernel command line option, in which case it will bypass resuming
331 from any hibernated image.
333 * The systemd-sleep.conf configuration file gained new options
334 AllowSuspend=, AllowHibernation=, AllowSuspendThenHibernate=,
335 AllowHybridSleep= for prohibiting specific sleep modes even if the
338 * portablectl is now officially supported and has thus moved to
341 * bootctl learnt the two new commands "set-default" and "set-oneshot"
342 for setting the default boot loader item to boot to (either
343 persistently or only for the next boot). This is currently only
344 compatible with sd-boot, but may be implemented on other boot loaders
345 too, that follow the boot loader interface. The updated interface is
348 https://systemd.io/BOOT_LOADER_INTERFACE
350 * A new kernel command line option systemd.early_core_pattern= is now
351 understood which may be used to influence the core_pattern PID 1
352 installs during early boot.
354 * busctl learnt two new options -j and --json= for outputting method
355 call replies, properties and monitoring output in JSON.
357 * journalctl's JSON output now supports simple ANSI coloring as well as
358 a new "json-seq" mode for generating RFC7464 output.
360 * Unit files now support the %g/%G specifiers that resolve to the UNIX
361 group/GID of the service manager runs as, similar to the existing
362 %u/%U specifiers that resolve to the UNIX user/UID.
364 * systemd-logind learnt a new global configuration option
365 UserStopDelaySec= that may be set in logind.conf. It specifies how
366 long the systemd --user instance shall remain started after a user
367 logs out. This is useful to speed up repetitive re-connections of the
368 same user, as it means the user's service manager doesn't have to be
369 stopped/restarted on each iteration, but can be reused between
370 subsequent options. This setting defaults to 10s. systemd-logind also
371 exports two new properties on its Manager D-Bus objects indicating
372 whether the system's lid is currently closed, and whether the system
375 * systemd gained support for a generic boot counting logic, which
376 generically permits automatic reverting to older boot loader entries
377 if newer updated ones don't work. The boot loader side is implemented
378 in sd-boot, but is kept open for other boot loaders too. For details
381 https://systemd.io/AUTOMATIC_BOOT_ASSESSMENT
383 * The SuccessAction=/FailureAction= unit file settings now learnt two
384 new parameters: "exit" and "exit-force", which result in immediate
385 exiting of the service manager, and are only useful in systemd --user
386 and container environments.
388 * Unit files gained support for a pair of options
389 FailureActionExitStatus=/SuccessActionExitStatus= for configuring the
390 exit status to use as service manager exit status when
391 SuccessAction=/FailureAction= is set to exit or exit-force.
393 * A pair of LogRateLimitIntervalSec=/LogRateLimitBurst= per-service
394 options may now be used to configure the log rate limiting applied by
395 journald per-service.
397 * systemd-analyze gained a new verb "timespan" for parsing and
398 normalizing time span values (i.e. strings like "5min 7s 8us").
400 * systemd-analyze also gained a new verb "security" for analyzing the
401 security and sand-boxing settings of services in order to determine an
402 "exposure level" for them, indicating whether a service would benefit
403 from more sand-boxing options turned on for them.
405 * "systemd-analyze syscall-filter" will now also show system calls
406 supported by the local kernel but not included in any of the defined
409 * .nspawn files now understand the Ephemeral= setting, matching the
410 --ephemeral command line switch.
412 * sd-event gained the new APIs sd_event_source_get_floating() and
413 sd_event_source_set_floating() for controlling whether a specific
414 event source is "floating", i.e. destroyed along with the even loop
417 * Unit objects on D-Bus gained a new "Refs" property that lists all
418 clients that currently have a reference on the unit (to ensure it is
421 * The JoinControllers= option in system.conf is no longer supported, as
422 it didn't work correctly, is hard to support properly, is legacy (as
423 the concept only exists on cgroup v1) and apparently wasn't used.
425 * Journal messages that are generated whenever a unit enters the failed
426 state are now tagged with a unique MESSAGE_ID. Similarly, messages
427 generated whenever a service process exits are now made recognizable,
428 too. A taged message is also emitted whenever a unit enters the
429 "dead" state on success.
431 * systemd-run gained a new switch --working-directory= for configuring
432 the working directory of the service to start. A shortcut -d is
433 equivalent, setting the working directory of the service to the
434 current working directory of the invoking program. The new --shell
435 (or just -S) option has been added for invoking the $SHELL of the
436 caller as a service, and implies --pty --same-dir --wait --collect
437 --service-type=exec. Or in other words, "systemd-run -S" is now the
438 quickest way to quickly get an interactive in a fully clean and
439 well-defined system service context.
441 * machinectl gained a new verb "import-fs" for importing an OS tree
442 from a directory. Moreover, when a directory or tarball is imported
443 and single top-level directory found with the OS itself below the OS
444 tree is automatically mangled and moved one level up.
446 * systemd-importd will no longer set up an implicit btrfs loop-back
447 file system on /var/lib/machines. If one is already set up, it will
450 * A new generator "systemd-run-generator" has been added. It will
451 synthesize a unit from one or more program command lines included in
452 the kernel command line. This is very useful in container managers
455 # systemd-nspawn -i someimage.raw -b systemd.run='"some command line"'
457 This will run "systemd-nspawn" on an image, invoke the specified
458 command line and immediately shut down the container again, returning
459 the command line's exit code.
461 * The block device locking logic is now documented:
463 https://systemd.io/BLOCK_DEVICE_LOCKING
465 * loginctl and machinectl now optionally output the various tables in
466 JSON using the --output= switch. It is our intention to add similar
467 support to systemctl and all other commands.
469 * udevadm's query and trigger verb now optionally take a .device unit
472 * systemd-udevd's network naming logic now understands a new
473 net.naming-scheme= kernel command line switch, which may be used to
474 pick a specific version of the naming scheme. This helps stabilizing
475 interface names even as systemd/udev are updated and the naming logic
478 * sd-id128.h learnt two new auxiliary helpers: sd_id128_is_allf() and
479 SD_ID128_ALLF to test if a 128bit ID is set to all 0xFF bytes, and to
480 initialize one to all 0xFF.
482 * After loading the SELinux policy systemd will now recursively relabel
483 all files and directories listed in
484 /run/systemd/relabel-extra.d/*.relabel (which should be simple
485 newline separated lists of paths) in addition to the ones it already
486 implicitly relabels in /run, /dev and /sys. After the relabelling is
487 completed the *.relabel files (and /run/systemd/relabel-extra.d/) are
488 removed. This is useful to permit initrds (i.e. code running before
489 the SELinux policy is in effect) to generate files in the host
490 filesystem safely and ensure that the correct label is applied during
491 the transition to the host OS.
493 * KERNEL API BREAKAGE: Linux kernel 4.18 changed behaviour regarding
494 mknod() handling in user namespaces. Previously mknod() would always
495 fail with EPERM in user namespaces. Since 4.18 mknod() will succeed
496 but device nodes generated that way cannot be opened, and attempts to
497 open them result in EPERM. This breaks the "graceful fallback" logic
498 in systemd's PrivateDevices= sand-boxing option. This option is
499 implemented defensively, so that when systemd detects it runs in a
500 restricted environment (such as a user namespace, or an environment
501 where mknod() is blocked through seccomp or absence of CAP_SYS_MKNOD)
502 where device nodes cannot be created the effect of PrivateDevices= is
503 bypassed (following the logic that 2nd-level sand-boxing is not
504 essential if the system systemd runs in is itself already sand-boxed
505 as a whole). This logic breaks with 4.18 in container managers where
506 user namespacing is used: suddenly PrivateDevices= succeeds setting
507 up a private /dev/ file system containing devices nodes — but when
508 these are opened they don't work.
510 At this point is is recommended that container managers utilizing
511 user namespaces that intend to run systemd in the payload explicitly
512 block mknod() with seccomp or similar, so that the graceful fallback
515 We are very sorry for the breakage and the requirement to change
516 container configurations for newer kernels. It's purely caused by an
517 incompatible kernel change. The relevant kernel developers have been
518 notified about this userspace breakage quickly, but they chose to
521 * PermissionsStartOnly= setting is deprecated (but is still supported
522 for backwards compatibility). The same functionality is provided by
523 the more flexible "+", "!", and "!!" prefixes to ExecStart= and other
526 * $DBUS_SESSION_BUS_ADDRESS environment variable is not set by
529 * The naming scheme for network devices was changed to always rename
530 devices, even if they were already renamed by userspace. The "kernel"
531 policy was changed to only apply as a fallback, if no other naming
534 * The requirements to build systemd is bumped to meson-0.46 and
537 Contributions from: afg, Alan Jenkins, Aleksei Timofeyev, Alexander
538 Filippov, Alexander Kurtz, Alexey Bogdanenko, Andreas Henriksson,
539 Andrew Jorgensen, Anita Zhang, apnix-uk, Arkan49, Arseny Maslennikov,
540 asavah, Asbjørn Apeland, aszlig, Bastien Nocera, Ben Boeckel, Benedikt
541 Morbach, Benjamin Berg, Bruce Zhang, Carlo Caione, Cedric Viou, Chen
542 Qi, Chris Chiu, Chris Down, Chris Morin, Christian Rebischke, Claudius
543 Ellsel, Colin Guthrie, dana, Daniel, Daniele Medri, Daniel Kahn
544 Gillmor, Daniel Rusek, Daniel van Vugt, Dariusz Gadomski, Dave Reisner,
545 David Anderson, Davide Cavalca, David Leeds, David Malcolm, David
546 Strauss, David Tardon, Dimitri John Ledkov, Dmitry Torokhov, dj-kaktus,
547 Dongsu Park, Elias Probst, Emil Soleyman, Erik Kooistra, Ervin Peters,
548 Evgeni Golov, Evgeny Vereshchagin, Fabrice Fontaine, Faheel Ahmad,
549 Faizal Luthfi, Felix Yan, Filipe Brandenburger, Franck Bui, Frank
550 Schaefer, Frantisek Sumsal, Gautier Husson, Gianluca Boiano, Giuseppe
551 Scrivano, glitsj16, Hans de Goede, Harald Hoyer, Harry Mallon, Harshit
552 Jain, Helmut Grohne, Henry Tung, Hui Yiqun, imayoda, Insun Pyo, Iwan
553 Timmer, Jan Janssen, Jan Pokorný, Jan Synacek, Jason A. Donenfeld,
554 javitoom, Jérémy Nouhaud, Jeremy Su, Jiuyang Liu, João Paulo Rechi
555 Vita, Joe Hershberger, Joe Rayhawk, Joerg Behrmann, Joerg Steffens,
556 Jonas Dorel, Jon Ringle, Josh Soref, Julian Andres Klode, Jun Bo Bi,
557 Jürg Billeter, Keith Busch, Khem Raj, Kirill Marinushkin, Larry
558 Bernstone, Lennart Poettering, Lion Yang, Li Song, Lorenz
559 Hübschle-Schneider, Lubomir Rintel, Lucas Werkmeister, Ludwin Janvier,
560 Lukáš Nykrýn, Luke Shumaker, mal, Marc-Antoine Perennou, Marcin
561 Skarbek, Marco Trevisan (Treviño), Marian Cepok, Mario Hros, Marko
562 Myllynen, Markus Grimm, Martin Pitt, Martin Sobotka, Martin Wilck,
563 Mathieu Trudel-Lapierre, Matthew Leeds, Michael Biebl, Michael Olbrich,
564 Michael 'pbone' Pobega, Michael Scherer, Michal Koutný, Michal
565 Sekletar, Michal Soltys, Mike Gilbert, Mike Palmer, Muhammet Kara, Neal
566 Gompa, Neil Brown, Network Silence, Niklas Tibbling, Nikolas Nyby,
567 Nogisaka Sadata, Oliver Smith, Patrik Flykt, Pavel Hrdina, Paweł
568 Szewczyk, Peter Hutterer, Piotr Drąg, Ray Strode, Reinhold Mueller,
569 Renaud Métrich, Roman Gushchin, Ronny Chevalier, Rubén Suárez Alvarez,
570 Ruixin Bao, RussianNeuroMancer, Ryutaroh Matsumoto, Saleem Rashid, Sam
571 Morris, Samuel Morris, Sandy Carter, scootergrisen, Sébastien Bacher,
572 Sergey Ptashnick, Shawn Landden, Shengyao Xue, Shih-Yuan Lee
573 (FourDollars), Silvio Knizek, Sjoerd Simons, Stasiek Michalski, Stephen
574 Gallagher, Steven Allen, Steve Ramage, Susant Sahani, Sven Joachim,
575 Sylvain Plantefève, Tanu Kaskinen, Tejun Heo, Thiago Macieira, Thomas
576 Blume, Thomas Haller, Thomas H. P. Andersen, Tim Ruffing, TJ, Tobias
577 Jungel, Todd Walton, Tommi Rantala, Tomsod M, Tony Novak, Tore
578 Anderson, Trevonn, Victor Laskurain, Victor Tapia, Violet Halo, Vojtech
579 Trefny, welaq, William A. Kennington III, William Douglas, Wyatt Ward,
580 Xiang Fan, Xi Ruoyao, Xuanwo, Yann E. Morin, YmrDtnJu, Yu Watanabe,
581 Zbigniew Jędrzejewski-Szmek, Zhang Xianwei, Zsolt Dollenstein
587 * NETWORK INTERFACE DEVICE NAMING CHANGES: systemd-udevd's "net_id"
588 builtin will name network interfaces differently than in previous
589 versions for virtual network interfaces created with SR-IOV and NPAR
590 and for devices where the PCI network controller device does not have
591 a slot number associated.
593 SR-IOV virtual devices are now named based on the name of the parent
594 interface, with a suffix of "v<N>", where <N> is the virtual device
595 number. Previously those virtual devices were named as if completely
598 The ninth and later NPAR virtual devices will be named following the
599 scheme used for the first eight NPAR partitions. Previously those
600 devices were not renamed and the kernel default (eth<n>) was used.
602 "net_id" will also generate names for PCI devices where the PCI
603 network controller device does not have an associated slot number
604 itself, but one of its parents does. Previously those devices were
605 not renamed and the kernel default (eth<n>) was used.
607 * AF_INET and AF_INET6 are dropped from RestrictAddressFamilies= in
608 systemd-logind.service. Since v235, IPAddressDeny=any has been set to
609 the unit. So, it is expected that the default behavior of
610 systemd-logind is not changed. However, if distribution packagers or
611 administrators disabled or modified IPAddressDeny= setting by a
612 drop-in config file, then it may be necessary to update the file to
613 re-enable AF_INET and AF_INET6 to support network user name services,
616 * When the RestrictNamespaces= unit property is specified multiple
617 times, then the specified types are merged now. Previously, only the
618 last assignment was used. So, if distribution packagers or
619 administrators modified the setting by a drop-in config file, then it
620 may be necessary to update the file.
622 * When OnFailure= is used in combination with Restart= on a service
623 unit, then the specified units will no longer be triggered on
624 failures that result in restarting. Previously, the specified units
625 would be activated each time the unit failed, even when the unit was
626 going to be restarted automatically. This behaviour contradicted the
627 documentation. With this release the code is adjusted to match the
630 * systemd-tmpfiles will now print a notice whenever it encounters
631 tmpfiles.d/ lines referencing the /var/run/ directory. It will
632 recommend reworking them to use the /run/ directory instead (for
633 which /var/run/ is simply a symlinked compatibility alias). This way
634 systemd-tmpfiles can properly detect line conflicts and merge lines
635 referencing the same file by two paths, without having to access
638 * systemctl disable/unmask/preset/preset-all cannot be used with
639 --runtime. Previously this was allowed, but resulted in unintuitive
640 behaviour that wasn't useful. systemctl disable/unmask will now undo
641 both runtime and persistent enablement/masking, i.e. it will remove
642 any relevant symlinks both in /run and /etc.
644 * Note that all long-running system services shipped with systemd will
645 now default to a system call whitelist (rather than a blacklist, as
646 before). In particular, systemd-udevd will now enforce one too. For
647 most cases this should be safe, however downstream distributions
648 which disabled sandboxing of systemd-udevd (specifically the
649 MountFlags= setting), might want to disable this security feature
650 too, as the default whitelisting will prohibit all mount, swap,
651 reboot and clock changing operations from udev rules.
653 * sd-boot acquired new loader configuration settings to optionally turn
654 off Windows and MacOS boot partition discovery as well as
655 reboot-into-firmware menu items. It is also able to pick a better
656 screen resolution for HiDPI systems, and now provides loader
657 configuration settings to change the resolution explicitly.
659 * systemd-resolved now supports DNS-over-TLS. It's still
660 turned off by default, use DNSOverTLS=opportunistic to turn it on in
661 resolved.conf. We intend to make this the default as soon as couple
662 of additional techniques for optimizing the initial latency caused by
663 establishing a TLS/TCP connection are implemented.
665 * systemd-resolved.service and systemd-networkd.service now set
666 DynamicUser=yes. The users systemd-resolve and systemd-network are
667 not created by systemd-sysusers anymore.
669 NOTE: This has a chance of breaking nss-ldap and similar NSS modules
670 that embedd a network facing module into any process using getpwuid()
671 or related call: the dynamic allocation of the user ID for
672 systemd-resolved.service means the service manager has to check NSS
673 if the user name is already taken when forking off the service. Since
674 the user in the common case won't be defined in /etc/passwd the
675 lookup is likely to trigger nss-ldap which in turn might use NSS to
676 ask systemd-resolved for hostname lookups. This will hence result in
677 a deadlock: a user name lookup in order to start
678 systemd-resolved.service will result in a host name lookup for which
679 systemd-resolved.service needs to be started already. There are
680 multiple ways to work around this problem: pre-allocate the
681 "systemd-resolve" user on such systems, so that nss-ldap won't be
682 triggered; or use a different NSS package that doesn't do networking
683 in-process but provides a local asynchronous name cache; or configure
684 the NSS package to avoid lookups for UIDs in the range `pkg-config
685 systemd --variable=dynamicuidmin` … `pkg-config systemd
686 --variable=dynamicuidmax`, so that it does not consider itself
687 authoritative for the same UID range systemd allocates dynamic users
690 * The systemd-resolve tool has been renamed to resolvectl (it also
691 remains available under the old name, for compatibility), and its
692 interface is now verb-based, similar in style to the other <xyz>ctl
693 tools, such as systemctl or loginctl.
695 * The resolvectl/systemd-resolve tool also provides 'resolvconf'
696 compatibility. It may be symlinked under the 'resolvconf' name, in
697 which case it will take arguments and input compatible with the
698 Debian and FreeBSD resolvconf tool.
700 * Support for suspend-then-hibernate has been added, i.e. a sleep mode
701 where the system initially suspends, and after a timeout resumes and
704 * networkd's ClientIdentifier= now accepts a new option "duid-only". If
705 set the client will only send a DUID as client identifier.
707 * The nss-systemd glibc NSS module will now enumerate dynamic users and
708 groups in effect. Previously, it could resolve UIDs/GIDs to user
709 names/groups and vice versa, but did not support enumeration.
711 * journald's Compress= configuration setting now optionally accepts a
712 byte threshold value. All journal objects larger than this threshold
713 will be compressed, smaller ones will not. Previously this threshold
714 was not configurable and set to 512.
716 * A new system.conf setting NoNewPrivileges= is now available which may
717 be used to turn off acquisition of new privileges system-wide
718 (i.e. set Linux' PR_SET_NO_NEW_PRIVS for PID 1 itself, and thus also
719 for all its children). Note that turning this option on means setuid
720 binaries and file system capabilities lose their special powers.
721 While turning on this option is a big step towards a more secure
722 system, doing so is likely to break numerous pre-existing UNIX tools,
723 in particular su and sudo.
725 * A new service systemd-time-sync-wait.service has been added. If
726 enabled it will delay the time-sync.target unit at boot until time
727 synchronization has been received from the network. This
728 functionality is useful on systems lacking a local RTC or where it is
729 acceptable that the boot process shall be delayed by external network
732 * When hibernating, systemd will now inform the kernel of the image
733 write offset, on kernels new enough to support this. This means swap
734 files should work for hibernation now.
736 * When loading unit files, systemd will now look for drop-in unit files
737 extensions in additional places. Previously, for a unit file name
738 "foo-bar-baz.service" it would look for dropin files in
739 "foo-bar-baz.service.d/*.conf". Now, it will also look in
740 "foo-bar-.service.d/*.conf" and "foo-.service.d/", i.e. at the
741 service name truncated after all inner dashes. This scheme allows
742 writing drop-ins easily that apply to a whole set of unit files at
743 once. It's particularly useful for mount and slice units (as their
744 naming is prefix based), but is also useful for service and other
745 units, for packages that install multiple unit files at once,
746 following a strict naming regime of beginning the unit file name with
747 the package's name. Two new specifiers are now supported in unit
748 files to match this: %j and %J are replaced by the part of the unit
749 name following the last dash.
751 * Unit files and other configuration files that support specifier
752 expansion now understand another three new specifiers: %T and %V will
753 resolve to /tmp and /var/tmp respectively, or whatever temporary
754 directory has been set for the calling user. %E will expand to either
755 /etc (for system units) or $XDG_CONFIG_HOME (for user units).
757 * The ExecStart= lines of unit files are no longer required to
758 reference absolute paths. If non-absolute paths are specified the
759 specified binary name is searched within the service manager's
760 built-in $PATH, which may be queried with 'systemd-path
761 search-binaries-default'. It's generally recommended to continue to
762 use absolute paths for all binaries specified in unit files.
764 * Units gained a new load state "bad-setting", which is used when a
765 unit file was loaded, but contained fatal errors which prevent it
766 from being started (for example, a service unit has been defined
767 lacking both ExecStart= and ExecStop= lines).
769 * coredumpctl's "gdb" verb has been renamed to "debug", in order to
770 support alternative debuggers, for example lldb. The old name
771 continues to be available however, for compatibility reasons. Use the
772 new --debugger= switch or the $SYSTEMD_DEBUGGER environment variable
773 to pick an alternative debugger instead of the default gdb.
775 * systemctl and the other tools will now output escape sequences that
776 generate proper clickable hyperlinks in various terminal emulators
777 where useful (for example, in the "systemctl status" output you can
778 now click on the unit file name to quickly open it in the
779 editor/viewer of your choice). Note that not all terminal emulators
780 support this functionality yet, but many do. Unfortunately, the
781 "less" pager doesn't support this yet, hence this functionality is
782 currently automatically turned off when a pager is started (which
783 happens quite often due to auto-paging). We hope to remove this
784 limitation as soon as "less" learns these escape sequences. This new
785 behaviour may also be turned off explicitly with the $SYSTEMD_URLIFY
786 environment variable. For details on these escape sequences see:
787 https://gist.github.com/egmontkob/eb114294efbcd5adb1944c9f3cb5feda
789 * networkd's .network files now support a new IPv6MTUBytes= option for
790 setting the MTU used by IPv6 explicitly as well as a new MTUBytes=
791 option in the [Route] section to configure the MTU to use for
792 specific routes. It also gained support for configuration of the DHCP
793 "UserClass" option through the new UserClass= setting. It gained
794 three new options in the new [CAN] section for configuring CAN
795 networks. The MULTICAST and ALLMULTI interface flags may now be
796 controlled explicitly with the new Multicast= and AllMulticast=
799 * networkd will now automatically make use of the kernel's route
800 expiration feature, if it is available.
802 * udevd's .link files now support setting the number of receive and
803 transmit channels, using the RxChannels=, TxChannels=,
804 OtherChannels=, CombinedChannels= settings.
806 * Support for UDPSegmentationOffload= has been removed, given its
807 limited support in hardware, and waning software support.
809 * networkd's .netdev files now support creating "netdevsim" interfaces.
811 * PID 1 learnt a new bus call GetUnitByControlGroup() which may be used
812 to query the unit belonging to a specific kernel control group.
814 * systemd-analyze gained a new verb "cat-config", which may be used to
815 dump the contents of any configuration file, with all its matching
816 drop-in files added in, and honouring the usual search and masking
817 logic applied to systemd configuration files. For example use
818 "systemd-analyze cat-config systemd/system.conf" to get the complete
819 system configuration file of systemd how it would be loaded by PID 1
820 itself. Similar to this, various tools such as systemd-tmpfiles or
821 systemd-sysusers, gained a new option "--cat-config", which does the
822 corresponding operation for their own configuration settings. For
823 example, "systemd-tmpfiles --cat-config" will now output the full
824 list of tmpfiles.d/ lines in place.
826 * timedatectl gained three new verbs: "show" shows bus properties of
827 systemd-timedated, "timesync-status" shows the current NTP
828 synchronization state of systemd-timesyncd, and "show-timesync"
829 shows bus properties of systemd-timesyncd.
831 * systemd-timesyncd gained a bus interface on which it exposes details
834 * A new environment variable $SYSTEMD_TIMEDATED_NTP_SERVICES is now
835 understood by systemd-timedated. It takes a colon-separated list of
836 unit names of NTP client services. The list is used by
837 "timedatectl set-ntp".
839 * systemd-nspawn gained a new --rlimit= switch for setting initial
840 resource limits for the container payload. There's a new switch
841 --hostname= to explicitly override the container's hostname. A new
842 --no-new-privileges= switch may be used to control the
843 PR_SET_NO_NEW_PRIVS flag for the container payload. A new
844 --oom-score-adjust= switch controls the OOM scoring adjustment value
845 for the payload. The new --cpu-affinity= switch controls the CPU
846 affinity of the container payload. The new --resolv-conf= switch
847 allows more detailed control of /etc/resolv.conf handling of the
848 container. Similarly, the new --timezone= switch allows more detailed
849 control of /etc/localtime handling of the container.
851 * systemd-detect-virt gained a new --list switch, which will print a
852 list of all currently known VM and container environments.
854 * Support for "Portable Services" has been added, see
855 doc/PORTABLE_SERVICES.md for details. Currently, the support is still
856 experimental, but this is expected to change soon. Reflecting this
857 experimental state, the "portablectl" binary is not installed into
858 /usr/bin yet. The binary has to be called with the full path
859 /usr/lib/systemd/portablectl instead.
861 * journalctl's and systemctl's -o switch now knows a new log output
862 mode "with-unit". The output it generates is very similar to the
863 regular "short" mode, but displays the unit name instead of the
864 syslog tag for each log line. Also, the date is shown with timezone
865 information. This mode is probably more useful than the classic
866 "short" output mode for most purposes, except where pixel-perfect
867 compatibility with classic /var/log/messages formatting is required.
869 * A new --dump-bus-properties switch has been added to the systemd
870 binary, which may be used to dump all supported D-Bus properties.
871 (Options which are still supported, but are deprecated, are *not*
874 * sd-bus gained a set of new calls:
875 sd_bus_slot_set_floating()/sd_bus_slot_get_floating() may be used to
876 enable/disable the "floating" state of a bus slot object,
877 i.e. whether the slot object pins the bus it is allocated for into
878 memory or if the bus slot object gets disconnected when the bus goes
879 away. sd_bus_open_with_description(),
880 sd_bus_open_user_with_description(),
881 sd_bus_open_system_with_description() may be used to allocate bus
882 objects and set their description string already during allocation.
884 * sd-event gained support for watching inotify events from the event
885 loop, in an efficient way, sharing inotify handles between multiple
886 users. For this a new function sd_event_add_inotify() has been added.
888 * sd-event and sd-bus gained support for calling special user-supplied
889 destructor functions for userdata pointers associated with
890 sd_event_source, sd_bus_slot, and sd_bus_track objects. For this new
891 functions sd_bus_slot_set_destroy_callback,
892 sd_bus_slot_get_destroy_callback, sd_bus_track_set_destroy_callback,
893 sd_bus_track_get_destroy_callback,
894 sd_event_source_set_destroy_callback,
895 sd_event_source_get_destroy_callback have been added.
897 * The "net.ipv4.tcp_ecn" sysctl will now be turned on by default.
899 * PID 1 will now automatically reschedule .timer units whenever the
900 local timezone changes. (They previously got rescheduled
901 automatically when the system clock changed.)
903 * New documentation has been added to document cgroups delegation,
904 portable services and the various code quality tools we have set up:
906 https://github.com/systemd/systemd/blob/master/docs/CGROUP_DELEGATION.md
907 https://github.com/systemd/systemd/blob/master/docs/PORTABLE_SERVICES.md
908 https://github.com/systemd/systemd/blob/master/docs/CODE_QUALITY.md
910 * The Boot Loader Specification has been added to the source tree.
912 https://github.com/systemd/systemd/blob/master/docs/BOOT_LOADER_SPECIFICATION.md
914 While moving it into our source tree we have updated it and further
915 changes are now accepted through the usual github PR workflow.
917 * pam_systemd will now look for PAM userdata fields systemd.memory_max,
918 systemd.tasks_max, systemd.cpu_weight, systemd.io_weight set by
919 earlier PAM modules. The data in these fields is used to initialize
920 the session scope's resource properties. Thus external PAM modules
921 may now configure per-session limits, for example sourced from
922 external user databases.
924 * socket units with Accept=yes will now maintain a "refused" counter in
925 addition to the existing "accepted" counter, counting connections
926 refused due to the enforced limits.
928 * The "systemd-path search-binaries-default" command may now be use to
929 query the default, built-in $PATH PID 1 will pass to the services it
932 * A new unit file setting PrivateMounts= has been added. It's a boolean
933 option. If enabled the unit's processes are invoked in their own file
934 system namespace. Note that this behaviour is also implied if any
935 other file system namespacing options (such as PrivateTmp=,
936 PrivateDevices=, ProtectSystem=, …) are used. This option is hence
937 primarily useful for services that do not use any of the other file
938 system namespacing options. One such service is systemd-udevd.service
939 wher this is now used by default.
941 * ConditionSecurity= gained a new value "uefi-secureboot" that is true
942 when the system is booted in UEFI "secure mode".
944 * A new unit "system-update-pre.target" is added, which defines an
945 optional synchronization point for offline system updates, as
946 implemented by the pre-existing "system-update.target" unit. It
947 allows ordering services before the service that executes the actual
948 update process in a generic way.
950 Contributions from: Adam Duskett, Alan Jenkins, Alessandro Casale,
951 Alexander Kurtz, Alex Gartrell, Anssi Hannula, Arnaud Rebillout, Brian
952 J. Murrell, Bruno Vernay, Chris Lamb, Chris Lesiak, Christian Brauner,
953 Christian Hesse, Christian Rebischke, Colin Guthrie, Daniel Dao, Daniel
954 Lin, Danylo Korostil, Davide Cavalca, David Tardon, Dimitri John
955 Ledkov, Dmitriy Geels, Douglas Christman, Elia Geretto, emelenas, Emil
956 Velikov, Evgeny Vereshchagin, Felipe Sateler, Feng Sun, Filipe
957 Brandenburger, Franck Bui, futpib, Giuseppe Scrivano, Guillem Jover,
958 guixxx, Hannes Reinecke, Hans de Goede, Harald Hoyer, Henrique Dante de
959 Almeida, Hiram van Paassen, Ian Miell, Igor Gnatenko, Ivan Shapovalov,
960 Iwan Timmer, James Cowgill, Jan Janssen, Jan Synacek, Jared Kazimir,
961 Jérémy Rosen, João Paulo Rechi Vita, Joost Heitbrink, Jui-Chi Ricky
962 Liang, Jürg Billeter, Kai-Heng Feng, Karol Augustin, Kay Sievers,
963 Krzysztof Nowicki, Lauri Tirkkonen, Lennart Poettering, Leonard König,
964 Long Li, Luca Boccassi, Lucas Werkmeister, Marcel Hoppe, Marc
965 Kleine-Budde, Mario Limonciello, Martin Jansa, Martin Wilck, Mathieu
966 Malaterre, Matteo F. Vescovi, Matthew McGinn, Matthias-Christian Ott,
967 Michael Biebl, Michael Olbrich, Michael Prokop, Michal Koutný, Michal
968 Sekletar, Mike Gilbert, Mikhail Kasimov, Milan Broz, Milan Pässler,
969 Mladen Pejaković, Muhammet Kara, Nicolas Boichat, Omer Katz, Paride
970 Legovini, Paul Menzel, Paul Milliken, Pavel Hrdina, Peter A. Bigot,
971 Peter D'Hoye, Peter Hutterer, Peter Jones, Philip Sequeira, Philip
972 Withnall, Piotr Drąg, Radostin Stoyanov, Ricardo Salveti de Araujo,
973 Ronny Chevalier, Rosen Penev, Rubén Suárez Alvarez, Ryan Gonzalez,
974 Salvo Tomaselli, Sebastian Reichel, Sergey Ptashnick, Sergio Lindo
975 Mansilla, Stefan Schweter, Stephen Hemminger, Stuart Hayes, Susant
976 Sahani, Sylvain Plantefève, Thomas H. P. Andersen, Tobias Jungel,
977 Tomasz Torcz, Vito Caputo, Will Dietz, Will Thompson, Wim van Mourik,
978 Yu Watanabe, Zbigniew Jędrzejewski-Szmek
984 * The MemoryAccounting= unit property now defaults to on. After
985 discussions with the upstream control group maintainers we learnt
986 that the negative impact of cgroup memory accounting on current
987 kernels is finally relatively minimal, so that it should be safe to
988 enable this by default without affecting system performance. Besides
989 memory accounting only task accounting is turned on by default, all
990 other forms of resource accounting (CPU, IO, IP) remain off for now,
991 because it's not clear yet that their impact is small enough to move
992 from opt-in to opt-out. We recommend downstreams to leave memory
993 accounting on by default if kernel 4.14 or higher is primarily
994 used. On very resource constrained systems or when support for old
995 kernels is a necessity, -Dmemory-accounting-default=false can be used
996 to revert this change.
998 * rpm scriptlets to update the udev hwdb and rules (%udev_hwdb_update,
999 %udev_rules_update) and the journal catalog (%journal_catalog_update)
1000 from the upgrade scriptlets of individual packages now do nothing.
1001 Transfiletriggers have been added which will perform those updates
1002 once at the end of the transaction.
1004 Similar transfiletriggers have been added to execute any sysctl.d
1005 and binfmt.d rules. Thus, it should be unnecessary to provide any
1006 scriptlets to execute this configuration from package installation
1009 * systemd-sysusers gained a mode where the configuration to execute is
1010 specified on the command line, but this configuration is not executed
1011 directly, but instead it is merged with the configuration on disk,
1012 and the result is executed. This is useful for package installation
1013 scripts which want to create the user before installing any files on
1014 disk (in case some of those files are owned by that user), while
1015 still allowing local admin overrides.
1017 This functionality is exposed to rpm scriptlets through a new
1018 %sysusers_create_package macro. Old %sysusers_create and
1019 %sysusers_create_inline macros are deprecated.
1021 A transfiletrigger for sysusers.d configuration is now installed,
1022 which means that it should be unnecessary to call systemd-sysusers from
1023 package installation scripts, unless the package installs any files
1024 owned by those newly-created users, in which case
1025 %sysusers_create_package should be used.
1027 * Analogous change has been done for systemd-tmpfiles: it gained a mode
1028 where the command-line configuration is merged with the configuration
1029 on disk. This is exposed as the new %tmpfiles_create_package macro,
1030 and %tmpfiles_create is deprecated. A transfiletrigger is installed
1031 for tmpfiles.d, hence it should be unnecessary to call systemd-tmpfiles
1032 from package installation scripts.
1034 * sysusers.d configuration for a user may now also specify the group
1035 number, in addition to the user number ("u username 123:456"), or
1036 without the user number ("u username -:456").
1038 * Configution items for systemd-sysusers can now be specified as
1039 positional arguments when the new --inline switch is used.
1041 * The login shell of users created through sysusers.d may now be
1042 specified (previously, it was always /bin/sh for root and
1043 /sbin/nologin for other users).
1045 * systemd-analyze gained a new --global switch to look at global user
1046 configuration. It also gained a unit-paths verb to list the unit load
1047 paths that are compiled into systemd (which can be used with
1048 --systemd, --user, or --global).
1050 * udevadm trigger gained a new --settle/-w option to wait for any
1051 triggered events to finish (but just those, and not any other events
1052 which are triggered meanwhile).
1054 * The action that systemd-logind takes when the lid is closed and the
1055 machine is connected to external power can now be configured using
1056 HandleLidSwitchExternalPower= in logind.conf. Previously, this action
1057 was determined by HandleLidSwitch=, and, for backwards compatibility,
1058 is still is, if HandleLidSwitchExternalPower= is not explicitly set.
1060 * journalctl will periodically call sd_journal_process() to make it
1061 resilient against inotify queue overruns when journal files are
1062 rotated very quickly.
1064 * Two new functions in libsystemd — sd_bus_get_n_queued_read and
1065 sd_bus_get_n_queued_write — may be used to check the number of
1066 pending bus messages.
1068 * systemd gained a new
1069 org.freedesktop.systemd1.Manager.AttachProcessesToUnit dbus call
1070 which can be used to migrate foreign processes to scope and service
1071 units. The primary user for this new API is systemd itself: the
1072 systemd --user instance uses this call of the systemd --system
1073 instance to migrate processes if it itself gets the request to
1074 migrate processes and the kernel refuses this due to access
1075 restrictions. Thanks to this "systemd-run --scope --user …" works
1076 again in pure cgroup v2 environments when invoked from the user
1079 * A new TemporaryFileSystem= setting can be used to mask out part of
1080 the real file system tree with tmpfs mounts. This may be combined
1081 with BindPaths= and BindReadOnlyPaths= to hide files or directories
1082 not relevant to the unit, while still allowing some paths lower in
1083 the tree to be accessed.
1085 ProtectHome=tmpfs may now be used to hide user home and runtime
1086 directories from units, in a way that is mostly equivalent to
1087 "TemporaryFileSystem=/home /run/user /root".
1089 * Non-service units are now started with KeyringMode=shared by default.
1090 This means that mount and swapon and other mount tools have access
1091 to keys in the main keyring.
1093 * /sys/fs/bpf is now mounted automatically.
1095 * QNX virtualization is now detected by systemd-detect-virt and may
1096 be used in ConditionVirtualization=.
1098 * IPAccounting= may now be enabled also for slice units.
1100 * A new -Dsplit-bin= build configuration switch may be used to specify
1101 whether bin and sbin directories are merged, or if they should be
1102 included separately in $PATH and various listings of executable
1103 directories. The build configuration scripts will try to autodetect
1104 the proper values of -Dsplit-usr= and -Dsplit-bin= based on build
1105 system, but distributions are encouraged to configure this
1108 * A new -Dok-color= build configuration switch may be used to change
1109 the colour of "OK" status messages.
1111 * UPGRADE ISSUE: serialization of units using JoinsNamespaceOf= with
1112 PrivateNetwork=yes was buggy in previous versions of systemd. This
1113 means that after the upgrade and daemon-reexec, any such units must
1116 * INCOMPATIBILITY: as announced in the NEWS for 237, systemd-tmpfiles
1117 will not exclude read-only files owned by root from cleanup.
1119 Contributions from: Alan Jenkins, Alexander F Rødseth, Alexis Jeandet,
1120 Andika Triwidada, Andrei Gherzan, Ansgar Burchardt, antizealot1337,
1121 Batuhan Osman Taşkaya, Beniamino Galvani, Bill Yodlowsky, Caio Marcelo
1122 de Oliveira Filho, CuBiC, Daniele Medri, Daniel Mouritzen, Daniel
1123 Rusek, Davide Cavalca, Dimitri John Ledkov, Douglas Christman, Evgeny
1124 Vereshchagin, Faalagorn, Filipe Brandenburger, Franck Bui, futpib,
1125 Giacomo Longo, Gunnar Hjalmarsson, Hans de Goede, Hermann Gausterer,
1126 Iago López Galeiras, Jakub Filak, Jan Synacek, Jason A. Donenfeld,
1127 Javier Martinez Canillas, Jérémy Rosen, Lennart Poettering, Lucas
1128 Werkmeister, Mao Huang, Marco Gulino, Michael Biebl, Michael Vogt,
1129 MilhouseVH, Neal Gompa (ニール・ゴンパ), Oleander Reis, Olof Mogren,
1130 Patrick Uiterwijk, Peter Hutterer, Peter Portante, Piotr Drąg, Robert
1131 Antoni Buj Gelonch, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
1132 Fowler, SjonHortensius, snorreflorre, Susant Sahani, Sylvain
1133 Plantefève, Thomas Blume, Thomas Haller, Vito Caputo, Yu Watanabe,
1134 Zbigniew Jędrzejewski-Szmek, Марко М. Костић (Marko M. Kostić)
1136 — Warsaw, 2018-03-05
1140 * Some keyboards come with a zoom see-saw or rocker which until now got
1141 mapped to the Linux "zoomin/out" keys in hwdb. However, these
1142 keycodes are not recognized by any major desktop. They now produce
1143 Up/Down key events so that they can be used for scrolling.
1145 * INCOMPATIBILITY: systemd-tmpfiles' "f" lines changed behaviour
1146 slightly: previously, if an argument was specified for lines of this
1147 type (i.e. the right-most column was set) this string was appended to
1148 existing files each time systemd-tmpfiles was run. This behaviour was
1149 different from what the documentation said, and not particularly
1150 useful, as repeated systemd-tmpfiles invocations would not be
1151 idempotent and grow such files without bounds. With this release
1152 behaviour has been altered to match what the documentation says:
1153 lines of this type only have an effect if the indicated files don't
1154 exist yet, and only then the argument string is written to the file.
1156 * FUTURE INCOMPATIBILITY: In systemd v238 we intend to slightly change
1157 systemd-tmpfiles behaviour: previously, read-only files owned by root
1158 were always excluded from the file "aging" algorithm (i.e. the
1159 automatic clean-up of directories like /tmp based on
1160 atime/mtime/ctime). We intend to drop this restriction, and age files
1161 by default even when owned by root and read-only. This behaviour was
1162 inherited from older tools, but there have been requests to remove
1163 it, and it's not obvious why this restriction was made in the first
1164 place. Please speak up now, if you are aware of software that reqires
1165 this behaviour, otherwise we'll remove the restriction in v238.
1167 * A new environment variable $SYSTEMD_OFFLINE is now understood by
1168 systemctl. It takes a boolean argument. If on, systemctl assumes it
1169 operates on an "offline" OS tree, and will not attempt to talk to the
1170 service manager. Previously, this mode was implicitly enabled if a
1171 chroot() environment was detected, and this new environment variable
1172 now provides explicit control.
1174 * .path and .socket units may now be created transiently, too.
1175 Previously only service, mount, automount and timer units were
1176 supported as transient units. The systemd-run tool has been updated
1177 to expose this new functionality, you may hence use it now to bind
1178 arbitrary commands to path or socket activation on-the-fly from the
1179 command line. Moreover, almost all properties are now exposed for the
1180 unit types that already supported transient operation.
1182 * The systemd-mount command gained support for a new --owner= parameter
1183 which takes a user name, which is then resolved and included in uid=
1184 and gid= mount options string of the file system to mount.
1186 * A new unit condition ConditionControlGroupController= has been added
1187 that checks whether a specific cgroup controller is available.
1189 * Unit files, udev's .link files, and systemd-networkd's .netdev and
1190 .network files all gained support for a new condition
1191 ConditionKernelVersion= for checking against specific kernel
1194 * In systemd-networkd, the [IPVLAN] section in .netdev files gained
1195 support for configuring device flags in the Flags= setting. In the
1196 same files, the [Tunnel] section gained support for configuring
1197 AllowLocalRemote=. The [Route] section in .network files gained
1198 support for configuring InitialCongestionWindow=,
1199 InitialAdvertisedReceiveWindow= and QuickAck=. The [DHCP] section now
1200 understands RapidCommit=.
1202 * systemd-networkd's DHCPv6 support gained support for Prefix
1205 * sd-bus gained support for a new "watch-bind" feature. When this
1206 feature is enabled, an sd_bus connection may be set up to connect to
1207 an AF_UNIX socket in the file system as soon as it is created. This
1208 functionality is useful for writing early-boot services that
1209 automatically connect to the system bus as soon as it is started,
1210 without ugly time-based polling. systemd-networkd and
1211 systemd-resolved have been updated to make use of this
1212 functionality. busctl exposes this functionality in a new
1213 --watch-bind= command line switch.
1215 * sd-bus will now optionally synthesize a local "Connected" signal as
1216 soon as a D-Bus connection is set up fully. This message mirrors the
1217 already existing "Disconnected" signal which is synthesized when the
1218 connection is terminated. This signal is generally useful but
1219 particularly handy in combination with the "watch-bind" feature
1220 described above. Synthesizing of this message has to be requested
1221 explicitly through the new API call sd_bus_set_connected_signal(). In
1222 addition a new call sd_bus_is_ready() has been added that checks
1223 whether a connection is fully set up (i.e. between the "Connected" and
1224 "Disconnected" signals).
1226 * sd-bus gained two new calls sd_bus_request_name_async() and
1227 sd_bus_release_name_async() for asynchronously registering bus
1228 names. Similar, there is now sd_bus_add_match_async() for installing
1229 a signal match asynchronously. All of systemd's own services have
1230 been updated to make use of these calls. Doing these operations
1231 asynchronously has two benefits: it reduces the risk of deadlocks in
1232 case of cyclic dependencies between bus services, and it speeds up
1233 service initialization since synchronization points for bus
1234 round-trips are removed.
1236 * sd-bus gained two new calls sd_bus_match_signal() and
1237 sd_bus_match_signal_async(), which are similar to sd_bus_add_match()
1238 and sd_bus_add_match_async() but instead of taking a D-Bus match
1239 string take match fields as normal function parameters.
1241 * sd-bus gained two new calls sd_bus_set_sender() and
1242 sd_bus_message_set_sender() for setting the sender name of outgoing
1243 messages (either for all outgoing messages or for just one specific
1244 one). These calls are only useful in direct connections as on
1245 brokered connections the broker fills in the sender anyway,
1246 overwriting whatever the client filled in.
1248 * sd-event gained a new pseudo-handle that may be specified on all API
1249 calls where an "sd_event*" object is expected: SD_EVENT_DEFAULT. When
1250 used this refers to the default event loop object of the calling
1251 thread. Note however that this does not implicitly allocate one —
1252 which has to be done prior by using sd_event_default(). Similarly
1253 sd-bus gained three new pseudo-handles SD_BUS_DEFAULT,
1254 SD_BUS_DEFAULT_USER, SD_BUS_DEFAULT_SYSTEM that may be used to refer
1255 to the default bus of the specified type of the calling thread. Here
1256 too this does not implicitly allocate bus connection objects, this
1257 has to be done prior with sd_bus_default() and friends.
1259 * sd-event gained a new call pair
1260 sd_event_source_{get|set}_io_fd_own(). This may be used to request
1261 automatic closure of the file descriptor an IO event source watches
1262 when the event source is destroyed.
1264 * systemd-networkd gained support for natively configuring WireGuard
1267 * In previous versions systemd synthesized user records both for the
1268 "nobody" (UID 65534) and "root" (UID 0) users in nss-systemd and
1269 internally. In order to simplify distribution-wide renames of the
1270 "nobody" user (like it is planned in Fedora: nfsnobody → nobody), a
1271 new transitional flag file has been added: if
1272 /etc/systemd/dont-synthesize-nobody exists synthesizing of the 65534
1273 user and group record within the systemd codebase is disabled.
1275 * systemd-notify gained a new --uid= option for selecting the source
1276 user/UID to use for notification messages sent to the service
1279 * journalctl gained a new --grep= option to list only entries in which
1280 the message matches a certain pattern. By default matching is case
1281 insensitive if the pattern is lowercase, and case sensitive
1282 otherwise. Option --case-sensitive=yes|no can be used to override
1283 this an specify case sensitivity or case insensitivity.
1285 * There's now a "systemd-analyze service-watchdogs" command for printing
1286 the current state of the service runtime watchdog, and optionally
1287 enabling or disabling the per-service watchdogs system-wide if given a
1288 boolean argument (i.e. the concept you configure in WatchdogSec=), for
1289 debugging purposes. There's also a kernel command line option
1290 systemd.service_watchdogs= for controlling the same.
1292 * Two new "log-level" and "log-target" options for systemd-analyze were
1293 added that merge the now deprecated get-log-level, set-log-level and
1294 get-log-target, set-log-target pairs. The deprecated options are still
1295 understood for backwards compatibility. The two new options print the
1296 current value when no arguments are given, and set them when a
1297 level/target is given as an argument.
1299 * sysusers.d's "u" lines now optionally accept both a UID and a GID
1300 specification, separated by a ":" character, in order to create users
1301 where UID and GID do not match.
1303 Contributions from: Adam Duskett, Alan Jenkins, Alexander Kuleshov,
1304 Alexis Deruelle, Andrew Jeddeloh, Armin Widegreen, Batuhan Osman
1305 Taşkaya, Björn Esser, bleep_blop, Bruce A. Johnson, Chris Down, Clinton
1306 Roy, Colin Walters, Daniel Rusek, Dimitri John Ledkov, Dmitry Rozhkov,
1307 Evgeny Vereshchagin, Ewout van Mansom, Felipe Sateler, Franck Bui,
1308 Frantisek Sumsal, George Gaydarov, Gianluca Boiano, Hans-Christian
1309 Noren Egtvedt, Hans de Goede, Henrik Grindal Bakken, Jan Alexander
1310 Steffens, Jan Klötzke, Jason A. Donenfeld, jdkbx, Jérémy Rosen,
1311 Jerónimo Borque, John Lin, John Paul Herold, Jonathan Rudenberg, Jörg
1312 Thalheim, Ken (Bitsko) MacLeod, Larry Bernstone, Lennart Poettering,
1313 Lucas Werkmeister, Maciej S. Szmigiero, Marek Čermák, Martin Pitt,
1314 Mathieu Malaterre, Matthew Thode, Matthias-Christian Ott, Max Harmathy,
1315 Michael Biebl, Michael Vogt, Michal Koutný, Michal Sekletar, Michał
1316 Szczepański, Mike Gilbert, Nathaniel McCallum, Nicolas Chauvet, Olaf
1317 Hering, Olivier Schwander, Patrik Flykt, Paul Cercueil, Peter Hutterer,
1318 Piotr Drąg, Raphael Vogelgsang, Reverend Homer, Robert Kolchmeyer,
1319 Samuel Dionne-Riel, Sergey Ptashnick, Shawn Landden, Susant Sahani,
1320 Sylvain Plantefève, Thomas H. P. Andersen, Thomas Huth, Tomasz
1321 Bachorski, Vladislav Vishnyakov, Wieland Hoffmann, Yu Watanabe, Zachary
1322 Winnerman, Zbigniew Jędrzejewski-Szmek, Дамјан Георгиевски, Дилян
1329 * The modprobe.d/ drop-in for the bonding.ko kernel module introduced
1330 in v235 has been extended to also set the dummy.ko module option
1331 numdummies=0, preventing the kernel from automatically creating
1332 dummy0. All dummy interfaces must now be explicitly created.
1334 * Unknown '%' specifiers in configuration files are now rejected. This
1335 applies to units and tmpfiles.d configuration. Any percent characters
1336 that are followed by a letter or digit that are not supposed to be
1337 interpreted as the beginning of a specifier should be escaped by
1338 doubling ("%%"). (So "size=5%" is still accepted, as well as
1339 "size=5%,foo=bar", but not "LABEL=x%y%z" since %y and %z are not
1340 valid specifiers today.)
1342 * systemd-resolved now maintains a new dynamic
1343 /run/systemd/resolve/stub-resolv.conf compatibility file. It is
1344 recommended to make /etc/resolv.conf a symlink to it. This file
1345 points at the systemd-resolved stub DNS 127.0.0.53 resolver and
1346 includes dynamically acquired search domains, achieving more correct
1347 DNS resolution by software that bypasses local DNS APIs such as NSS.
1349 * The "uaccess" udev tag has been dropped from /dev/kvm and
1350 /dev/dri/renderD*. These devices now have the 0666 permissions by
1351 default (but this may be changed at build-time). /dev/dri/renderD*
1352 will now be owned by the "render" group along with /dev/kfd.
1354 * "DynamicUser=yes" has been enabled for systemd-timesyncd.service,
1355 systemd-journal-gatewayd.service and
1356 systemd-journal-upload.service. This means "nss-systemd" must be
1357 enabled in /etc/nsswitch.conf to ensure the UIDs assigned to these
1358 services are resolved properly.
1360 * In /etc/fstab two new mount options are now understood:
1361 x-systemd.makefs and x-systemd.growfs. The former has the effect that
1362 the configured file system is formatted before it is mounted, the
1363 latter that the file system is resized to the full block device size
1364 after it is mounted (i.e. if the file system is smaller than the
1365 partition it resides on, it's grown). This is similar to the fsck
1366 logic in /etc/fstab, and pulls in systemd-makefs@.service and
1367 systemd-growfs@.service as necessary, similar to
1368 systemd-fsck@.service. Resizing is currently only supported on ext4
1371 * In systemd-networkd, the IPv6 RA logic now optionally may announce
1372 DNS server and domain information.
1374 * Support for the LUKS2 on-disk format for encrypted partitions has
1375 been added. This requires libcryptsetup2 during compilation and
1378 * The systemd --user instance will now signal "readiness" when its
1379 basic.target unit has been reached, instead of when the run queue ran
1380 empty for the first time.
1382 * Tmpfiles.d with user configuration are now also supported.
1383 systemd-tmpfiles gained a new --user switch, and snippets placed in
1384 ~/.config/user-tmpfiles.d/ and corresponding directories will be
1385 executed by systemd-tmpfiles --user running in the new
1386 systemd-tmpfiles-setup.service and systemd-tmpfiles-clean.service
1387 running in the user session.
1389 * Unit files and tmpfiles.d snippets learnt three new % specifiers:
1390 %S resolves to the top-level state directory (/var/lib for the system
1391 instance, $XDG_CONFIG_HOME for the user instance), %C resolves to the
1392 top-level cache directory (/var/cache for the system instance,
1393 $XDG_CACHE_HOME for the user instance), %L resolves to the top-level
1394 logs directory (/var/log for the system instance,
1395 $XDG_CONFIG_HOME/log/ for the user instance). This matches the
1396 existing %t specifier, that resolves to the top-level runtime
1397 directory (/run for the system instance, and $XDG_RUNTIME_DIR for the
1400 * journalctl learnt a new parameter --output-fields= for limiting the
1401 set of journal fields to output in verbose and JSON output modes.
1403 * systemd-timesyncd's configuration file gained a new option
1404 RootDistanceMaxSec= for setting the maximum root distance of servers
1405 it'll use, as well as the new options PollIntervalMinSec= and
1406 PollIntervalMaxSec= to tweak the minimum and maximum poll interval.
1408 * bootctl gained a new command "list" for listing all available boot
1409 menu items on systems that follow the boot loader specification.
1411 * systemctl gained a new --dry-run switch that shows what would be done
1412 instead of doing it, and is currently supported by the shutdown and
1415 * ConditionSecurity= can now detect the TOMOYO security module.
1417 * Unit file [Install] sections are now also respected in unit drop-in
1418 files. This is intended to be used by drop-ins under /usr/lib/.
1420 * systemd-firstboot may now also set the initial keyboard mapping.
1422 * Udev "changed" events for devices which are exposed as systemd
1423 .device units are now propagated to units specified in
1424 ReloadPropagatedFrom= as reload requests.
1426 * If a udev device has a SYSTEMD_WANTS= property containing a systemd
1427 unit template name (i.e. a name in the form of 'foobar@.service',
1428 without the instance component between the '@' and - the '.'), then
1429 the escaped sysfs path of the device is automatically used as the
1432 * SystemCallFilter= in unit files has been extended so that an "errno"
1433 can be specified individually for each system call. Example:
1434 SystemCallFilter=~uname:EILSEQ.
1436 * The cgroup delegation logic has been substantially updated. Delegate=
1437 now optionally takes a list of controllers (instead of a boolean, as
1438 before), which lists the controllers to delegate at least.
1440 * The networkd DHCPv6 client now implements the FQDN option (RFC 4704).
1442 * A new LogLevelMax= setting configures the maximum log level any
1443 process of the service may log at (i.e. anything with a lesser
1444 priority than what is specified is automatically dropped). A new
1445 LogExtraFields= setting allows configuration of additional journal
1446 fields to attach to all log records generated by any of the unit's
1449 * New StandardInputData= and StandardInputText= settings along with the
1450 new option StandardInput=data may be used to configure textual or
1451 binary data that shall be passed to the executed service process via
1452 standard input, encoded in-line in the unit file.
1454 * StandardInput=, StandardOutput= and StandardError= may now be used to
1455 connect stdin/stdout/stderr of executed processes directly with a
1456 file or AF_UNIX socket in the file system, using the new "file:" option.
1458 * A new unit file option CollectMode= has been added, that allows
1459 tweaking the garbage collection logic for units. It may be used to
1460 tell systemd to garbage collect units that have failed automatically
1461 (normally it only GCs units that exited successfully). systemd-run
1462 and systemd-mount expose this new functionality with a new -G option.
1464 * "machinectl bind" may now be used to bind mount non-directories
1465 (i.e. regularfiles, devices, fifos, sockets).
1467 * systemd-analyze gained a new verb "calendar" for validating and
1468 testing calendar time specifications to use for OnCalendar= in timer
1469 units. Besides validating the expression it will calculate the next
1470 time the specified expression would elapse.
1472 * In addition to the pre-existing FailureAction= unit file setting
1473 there's now SuccessAction=, for configuring a shutdown action to
1474 execute when a unit completes successfully. This is useful in
1475 particular inside containers that shall terminate after some workload
1476 has been completed. Also, both options are now supported for all unit
1477 types, not just services.
1479 * networkds's IP rule support gained two new options
1480 IncomingInterface= and OutgoingInterface= for configuring the incoming
1481 and outgoing interfaces of configured rules. systemd-networkd also
1482 gained support for "vxcan" network devices.
1484 * networkd gained a new setting RequiredForOnline=, taking a
1485 boolean. If set, systemd-wait-online will take it into consideration
1486 when determining that the system is up, otherwise it will ignore the
1487 interface for this purpose.
1489 * The sd_notify() protocol gained support for a new operation: with
1490 FDSTOREREMOVE=1 file descriptors may be removed from the per-service
1491 store again, ahead of POLLHUP or POLLERR when they are removed
1494 * A new document doc/UIDS-GIDS.md has been added to the source tree,
1495 that documents the UID/GID range and assignment assumptions and
1496 requirements of systemd.
1498 * The watchdog device PID 1 will ping may now be configured through the
1499 WatchdogDevice= configuration file setting, or by setting the
1500 systemd.watchdog_service= kernel commandline option.
1502 * systemd-resolved's gained support for registering DNS-SD services on
1503 the local network using MulticastDNS. Services may either be
1504 registered by dropping in a .dnssd file in /etc/systemd/dnssd/ (or
1505 the same dir below /run, /usr/lib), or through its D-Bus API.
1507 * The sd_notify() protocol can now with EXTEND_TIMEOUT_USEC=microsecond
1508 extend the effective start, runtime, and stop time. The service must
1509 continue to send EXTEND_TIMEOUT_USEC within the period specified to
1510 prevent the service manager from making the service as timedout.
1512 * systemd-resolved's DNSSEC support gained support for RFC 8080
1513 (Ed25519 keys and signatures).
1515 * The systemd-resolve command line tool gained a new set of options
1516 --set-dns=, --set-domain=, --set-llmnr=, --set-mdns=, --set-dnssec=,
1517 --set-nta= and --revert to configure per-interface DNS configuration
1518 dynamically during runtime. It's useful for pushing DNS information
1519 into systemd-resolved from DNS hook scripts that various interface
1520 managing software supports (such as pppd).
1522 * systemd-nspawn gained a new --network-namespace-path= command line
1523 option, which may be used to make a container join an existing
1524 network namespace, by specifying a path to a "netns" file.
1526 Contributions from: Alan Jenkins, Alan Robertson, Alessandro Ghedini,
1527 Andrew Jeddeloh, Antonio Rojas, Ari, asavah, bleep_blop, Carsten
1528 Strotmann, Christian Brauner, Christian Hesse, Clinton Roy, Collin
1529 Eggert, Cong Wang, Daniel Black, Daniel Lockyer, Daniel Rusek, Dimitri
1530 John Ledkov, Dmitry Rozhkov, Dongsu Park, Edward A. James, Evgeny
1531 Vereshchagin, Florian Klink, Franck Bui, Gwendal Grignou, Hans de
1532 Goede, Harald Hoyer, Hristo Venev, Iago López Galeiras, Ikey Doherty,
1533 Jakub Wilk, Jérémy Rosen, Jiahui Xie, John Lin, José Bollo, Josef
1534 Andersson, juga0, Krzysztof Nowicki, Kyle Walker, Lars Karlitski, Lars
1535 Kellogg-Stedman, Lauri Tirkkonen, Lennart Poettering, Lubomir Rintel,
1536 Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn, Lukáš Říha, Lukasz
1537 Rubaszewski, Maciej S. Szmigiero, Mantas Mikulėnas, Marcus Folkesson,
1538 Martin Steuer, Mathieu Trudel-Lapierre, Matija Skala,
1539 Matthias-Christian Ott, Max Resch, Michael Biebl, Michael Vogt, Michal
1540 Koutný, Michal Sekletar, Mike Gilbert, Muhammet Kara, Neil Brown, Olaf
1541 Hering, Ondrej Kozina, Patrik Flykt, Patryk Kocielnik, Peter Hutterer,
1542 Piotr Drąg, Razvan Cojocaru, Robin McCorkell, Roland Hieber, Saran
1543 Tunyasuvunakool, Sergey Ptashnick, Shawn Landden, Shuang Liu, Simon
1544 Arlott, Simon Peeters, Stanislav Angelovič, Stefan Agner, Susant
1545 Sahani, Sylvain Plantefève, Thomas Blume, Thomas Haller, Tiago Salem
1546 Herrmann, Tinu Weber, Tom Stellard, Topi Miettinen, Torsten Hilbrich,
1547 Vito Caputo, Vladislav Vishnyakov, WaLyong Cho, Yu Watanabe, Zbigniew
1548 Jędrzejewski-Szmek, Zeal Jagannatha
1550 — Berlin, 2017-12-14
1554 * INCOMPATIBILITY: systemd-logind.service and other long-running
1555 services now run inside an IPv4/IPv6 sandbox, prohibiting them any IP
1556 communication with the outside. This generally improves security of
1557 the system, and is in almost all cases a safe and good choice, as
1558 these services do not and should not provide any network-facing
1559 functionality. However, systemd-logind uses the glibc NSS API to
1560 query the user database. This creates problems on systems where NSS
1561 is set up to directly consult network services for user database
1562 lookups. In particular, this creates incompatibilities with the
1563 "nss-nis" module, which attempts to directly contact the NIS/YP
1564 network servers it is configured for, and will now consistently
1565 fail. In such cases, it is possible to turn off IP sandboxing for
1566 systemd-logind.service (set IPAddressDeny= in its [Service] section
1567 to the empty string, via a .d/ unit file drop-in). Downstream
1568 distributions might want to update their nss-nis packaging to include
1569 such a drop-in snippet, accordingly, to hide this incompatibility
1570 from the user. Another option is to make use of glibc's nscd service
1571 to proxy such network requests through a privilege-separated, minimal
1572 local caching daemon, or to switch to more modern technologies such
1573 sssd, whose NSS hook-ups generally do not involve direct network
1574 access. In general, we think it's definitely time to question the
1575 implementation choices of nss-nis, i.e. whether it's a good idea
1576 today to embed a network-facing loadable module into all local
1577 processes that need to query the user database, including the most
1578 trivial and benign ones, such as "ls". For more details about
1579 IPAddressDeny= see below.
1581 * A new modprobe.d drop-in is now shipped by default that sets the
1582 bonding module option max_bonds=0. This overrides the kernel default,
1583 to avoid conflicts and ambiguity as to whether or not bond0 should be
1584 managed by systemd-networkd or not. This resolves multiple issues
1585 with bond0 properties not being applied, when bond0 is configured
1586 with systemd-networkd. Distributors may choose to not package this,
1587 however in that case users will be prevented from correctly managing
1588 bond0 interface using systemd-networkd.
1590 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
1591 which print the logging level and target of the system manager. They
1592 complement the existing "set-log-level" and "set-log-target" verbs
1593 used to change those values.
1595 * journald.conf gained a new boolean setting ReadKMsg= which defaults
1596 to on. If turned off kernel log messages will not be read by
1597 systemd-journald or included in the logs. It also gained a new
1598 setting LineMax= for configuring the maximum line length in
1599 STDOUT/STDERR log streams. The new default for this value is 48K, up
1600 from the previous hardcoded 2048.
1602 * A new unit setting RuntimeDirectoryPreserve= has been added, which
1603 allows more detailed control of what to do with a runtime directory
1604 configured with RuntimeDirectory= (i.e. a directory below /run or
1605 $XDG_RUNTIME_DIR) after a unit is stopped.
1607 * The RuntimeDirectory= setting for units gained support for creating
1608 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
1609 one top-level directory.
1611 * Units gained new options StateDirectory=, CacheDirectory=,
1612 LogsDirectory= and ConfigurationDirectory= which are closely related
1613 to RuntimeDirectory= but manage per-service directories below
1614 /var/lib, /var/cache, /var/log and /etc. By making use of them it is
1615 possible to write unit files which when activated automatically gain
1616 properly owned service specific directories in these locations, thus
1617 making unit files self-contained and increasing compatibility with
1618 stateless systems and factory reset where /etc or /var are
1619 unpopulated at boot. Matching these new settings there's also
1620 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
1621 ConfigurationDirectoryMode= for configuring the access mode of these
1622 directories. These settings are particularly useful in combination
1623 with DynamicUser=yes as they provide secure, properly-owned,
1624 writable, and stateful locations for storage, excluded from the
1625 sandbox that such services live in otherwise.
1627 * Automake support has been removed from this release. systemd is now
1630 * systemd-journald will now aggressively cache client metadata during
1631 runtime, speeding up log write performance under pressure. This comes
1632 at a small price though: as much of the metadata is read
1633 asynchronously from /proc/ (and isn't implicitly attached to log
1634 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
1635 metadata stored alongside a log entry might be slightly
1636 out-of-date. Previously it could only be slightly newer than the log
1637 message. The time window is small however, and given that the kernel
1638 is unlikely to be improved anytime soon in this regard, this appears
1641 * nss-myhostname/systemd-resolved will now by default synthesize an
1642 A/AAAA resource record for the "_gateway" hostname, pointing to the
1643 current default IP gateway. Previously it did that for the "gateway"
1644 name, hampering adoption, as some distributions wanted to leave that
1645 host name open for local use. The old behaviour may still be
1646 requested at build time.
1648 * systemd-networkd's [Address] section in .network files gained a new
1649 Scope= setting for configuring the IP address scope. The [Network]
1650 section gained a new boolean setting ConfigureWithoutCarrier= that
1651 tells systemd-networkd to ignore link sensing when configuring the
1652 device. The [DHCP] section gained a new Anonymize= boolean option for
1653 turning on a number of options suggested in RFC 7844. A new
1654 [RoutingPolicyRule] section has been added for configuring the IP
1655 routing policy. The [Route] section has gained support for a new
1656 Type= setting which permits configuring
1657 blackhole/unreachable/prohibit routes.
1659 * The [VRF] section in .netdev files gained a new Table= setting for
1660 configuring the routing table to use. The [Tunnel] section gained a
1661 new Independent= boolean field for configuring tunnels independent of
1662 an underlying network interface. The [Bridge] section gained a new
1663 GroupForwardMask= option for configuration of propagation of link
1664 local frames between bridge ports.
1666 * The WakeOnLan= setting in .link files gained support for a number of
1667 new modes. A new TCP6SegmentationOffload= setting has been added for
1668 configuring TCP/IPv6 hardware segmentation offload.
1670 * The IPv6 RA sender implementation may now optionally send out RDNSS
1671 and RDNSSL records to supply DNS configuration to peers.
1673 * systemd-nspawn gained support for a new --system-call-filter= command
1674 line option for adding and removing entries in the default system
1675 call filter it applies. Moreover systemd-nspawn has been changed to
1676 implement a system call whitelist instead of a blacklist.
1678 * systemd-run gained support for a new --pipe command line option. If
1679 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
1680 are directly passed on to the activated transient service
1681 executable. This allows invoking arbitrary processes as systemd
1682 services (for example to take benefit of dependency management,
1683 accounting management, resource management or log management that is
1684 done automatically for services) — while still allowing them to be
1685 integrated in a classic UNIX shell pipeline.
1687 * When a service sends RELOAD=1 via sd_notify() and reload propagation
1688 using ReloadPropagationTo= is configured, a reload is now propagated
1689 to configured units. (Previously this was only done on explicitly
1690 requested reloads, using "systemctl reload" or an equivalent
1693 * For each service unit a restart counter is now kept: it is increased
1694 each time the service is restarted due to Restart=, and may be
1695 queried using "systemctl show -p NRestarts …".
1697 * New system call filter groups @aio, @sync, @chown, @setuid, @memlock,
1698 @signal and @timer have been added, for usage with SystemCallFilter=
1699 in unit files and the new --system-call-filter= command line option
1700 of systemd-nspawn (see above).
1702 * ExecStart= lines in unit files gained two new modifiers: when a
1703 command line is prefixed with "!" the command will be executed as
1704 configured, except for the credentials applied by
1705 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
1706 "+", but does still apply namespacing options unlike "+". There's
1707 also "!!" now, which is mostly identical, but becomes a NOP on
1708 systems that support ambient capabilities. This is useful to write
1709 unit files that work with ambient capabilities where possible but
1710 automatically fall back to traditional privilege dropping mechanisms
1711 on systems where this is not supported.
1713 * ListenNetlink= settings in socket units now support RDMA netlink
1716 * A new unit file setting LockPersonality= has been added which permits
1717 locking down the chosen execution domain ("personality") of a service
1720 * A new special target "getty-pre.target" has been added, which is
1721 ordered before all text logins, and may be used to order services
1722 before textual logins acquire access to the console.
1724 * systemd will now attempt to load the virtio-rng.ko kernel module very
1725 early on if a VM environment supporting this is detected. This should
1726 improve entropy during early boot in virtualized environments.
1728 * A _netdev option is now supported in /etc/crypttab that operates in a
1729 similar way as the same option in /etc/fstab: it permits configuring
1730 encrypted devices that need to be ordered after the network is up.
1731 Following this logic, two new special targets
1732 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
1733 added that are to cryptsetup.target what remote-fs.target and
1734 remote-fs-pre.target are to local-fs.target.
1736 * Service units gained a new UnsetEnvironment= setting which permits
1737 unsetting specific environment variables for services that are
1738 normally passed to it (for example in order to mask out locale
1739 settings for specific services that can't deal with it).
1741 * Units acquired a new boolean option IPAccounting=. When turned on, IP
1742 traffic accounting (packet count as well as byte count) is done for
1743 the service, and shown as part of "systemctl status" or "systemd-run
1746 * Service units acquired two new options IPAddressAllow= and
1747 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
1748 for configuring a simple IP access control list for all sockets of
1749 the unit. These options are available also on .slice and .socket
1750 units, permitting flexible access list configuration for individual
1751 services as well as groups of services (as defined by a slice unit),
1752 including system-wide. Note that IP ACLs configured this way are
1753 enforced on every single IPv4 and IPv6 socket created by any process
1754 of the service unit, and apply to ingress as well as egress traffic.
1756 * If CPUAccounting= or IPAccounting= is turned on for a unit a new
1757 structured log message is generated each time the unit is stopped,
1758 containing information about the consumed resources of this
1761 * A new setting KeyringMode= has been added to unit files, which may be
1762 used to control how the kernel keyring is set up for executed
1765 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
1766 "systemctl kexec" and "systemctl exit" are now always asynchronous in
1767 behaviour (that is: these commands return immediately after the
1768 operation was enqueued instead of waiting for the operation to
1769 complete). Previously, "systemctl poweroff" and "systemctl reboot"
1770 were asynchronous on systems using systemd-logind (i.e. almost
1771 always, and like they were on sysvinit), and the other three commands
1772 were unconditionally synchronous. With this release this is cleaned
1773 up, and callers will see the same asynchronous behaviour on all
1774 systems for all five operations.
1776 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
1779 * .timer units now accept calendar specifications in other timezones
1780 than UTC or the local timezone.
1782 * The tmpfiles snippet var.conf has been changed to create
1783 /var/log/btmp with access mode 0660 instead of 0600. It was owned by
1784 the "utmp" group already, and it appears to be generally understood
1785 that members of "utmp" can modify/flush the utmp/wtmp/lastlog/btmp
1786 databases. Previously this was implemented correctly for all these
1787 databases excepts btmp, which has been opened up like this now
1788 too. Note that while the other databases are world-readable
1789 (i.e. 0644), btmp is not and remains more restrictive.
1791 * The systemd-resolve tool gained a new --reset-server-features
1792 switch. When invoked like this systemd-resolved will forget
1793 everything it learnt about the features supported by the configured
1794 upstream DNS servers, and restarts the feature probing logic on the
1795 next resolver look-up for them at the highest feature level
1798 * The status dump systemd-resolved sends to the logs upon receiving
1799 SIGUSR1 now also includes information about all DNS servers it is
1800 configured to use, and the features levels it probed for them.
1802 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
1803 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
1804 Burchardt, Beniamino Galvani, Benjamin Berg, Benjamin Robin, Charles
1805 Huber, Christian Hesse, Daniel Berrange, Daniel Kahn Gillmor, Daniel
1806 Mack, Daniel Rusek, Daniel Șerbănescu, Davide Cavalca, Dimitri John
1807 Ledkov, Diogo Pereira, Djalal Harouni, Dmitriy Geels, Dmitry Torokhov,
1808 ettavolt, Evgeny Vereshchagin, Fabio Kung, Felipe Sateler, Franck Bui,
1809 Hans de Goede, Harald Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov,
1810 Jakub Wilk, Jan Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen,
1811 John Lin, jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg
1812 Thalheim, Jouke Witteveen, juga0, Justin Capella, Justin Michaud,
1813 Kai-Heng Feng, Lennart Poettering, Lion Yang, Luca Bruno, Lucas
1814 Werkmeister, Lukáš Nykrýn, Marcel Hollerbach, Marcus Lundblad, Martin
1815 Pitt, Michael Biebl, Michael Grzeschik, Michal Sekletar, Mike Gilbert,
1816 Neil Brown, Nicolas Iooss, Patrik Flykt, pEJipE, Piotr Drąg, Russell
1817 Stuart, S. Fan, Shengyao Xue, Stefan Pietsch, Susant Sahani, Tejun Heo,
1818 Thomas Miller, Thomas Sailer, Tobias Hunger, Tomasz Pala, Tom
1819 Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø, userwithuid,
1820 Vasilis Liaskovitis, Vito Caputo, WaLyong Cho, William Douglas, Xiang
1821 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
1823 — Berlin, 2017-10-06
1827 * Meson is now supported as build system in addition to Automake. It is
1828 our plan to remove Automake in one of our next releases, so that
1829 Meson becomes our exclusive build system. Hence, please start using
1830 the Meson build system in your downstream packaging. There's plenty
1831 of documentation around how to use Meson, the extremely brief
1834 ./autogen.sh && ./configure && make && sudo make install
1838 meson build && ninja -C build && sudo ninja -C build install
1840 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
1841 which permits configuring a timeout on the time a job is
1842 running. This is particularly useful for setting timeouts on jobs for
1845 * Unit files gained two new options ConditionUser= and ConditionGroup=
1846 for conditionalizing units based on the identity of the user/group
1847 running a systemd user instance.
1849 * systemd-networkd now understands a new FlowLabel= setting in the
1850 [VXLAN] section of .network files, as well as a Priority= in
1851 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
1852 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
1853 gained support for configuration of GENEVE links, and IPv6 address
1854 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
1856 * .link files now understand a new Port= setting.
1858 * systemd-networkd's DHCP support gained support for DHCP option 119
1859 (domain search list).
1861 * systemd-networkd gained support for serving IPv6 address ranges using
1862 the Router Advertisement protocol. The new .network configuration
1863 section [IPv6Prefix] may be used to configure the ranges to
1864 serve. This is implemented based on a new, minimal, native server
1865 implementation of RA.
1867 * journalctl's --output= switch gained support for a new parameter
1868 "short-iso-precise" for a mode where timestamps are shown as precise
1871 * systemd-udevd's "net_id" builtin may now generate stable network
1872 interface names from IBM PowerVM VIO devices as well as ACPI platform
1875 * MulticastDNS support in systemd-resolved may now be explicitly
1876 enabled/disabled using the new MulticastDNS= configuration file
1879 * systemd-resolved may now optionally use libidn2 instead of the libidn
1880 for processing internationalized domain names. Support for libidn2
1881 should be considered experimental and should not be enabled by
1884 * "machinectl pull-tar" and related call may now do verification of
1885 downloaded images using SUSE-style .sha256 checksum files in addition
1886 to the already existing support for validating using Ubuntu-style
1889 * sd-bus gained support for a new sd_bus_message_appendv() call which
1890 is va_list equivalent of sd_bus_message_append().
1892 * sd-boot gained support for validating images using SHIM/MOK.
1894 * The SMACK code learnt support for "onlycap".
1896 * systemd-mount --umount is now much smarter in figuring out how to
1897 properly unmount a device given its mount or device path.
1899 * The code to call libnss_dns as a fallback from libnss_resolve when
1900 the communication with systemd-resolved fails was removed. This
1901 fallback was redundant and interfered with the [!UNAVAIL=return]
1902 suffix. See nss-resolve(8) for the recommended configuration.
1904 * systemd-logind may now be restarted without losing state. It stores
1905 the file descriptors for devices it manages in the system manager
1906 using the FDSTORE= mechanism. Please note that further changes in
1907 other components may be required to make use of this (for example
1908 Xorg has code to listen for stops of systemd-logind and terminate
1909 itself when logind is stopped or restarted, in order to avoid using
1910 stale file descriptors for graphical devices, which is now
1911 counterproductive and must be reverted in order for restarts of
1912 systemd-logind to be safe. See
1913 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
1915 * All kernel install plugins are called with the environment variable
1916 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
1917 /etc/machine-id. If the file is missing or empty, the variable is
1918 empty and BOOT_DIR_ABS is the path of a temporary directory which is
1919 removed after all the plugins exit. So, if KERNEL_INSTALL_MACHINE_ID
1920 is empty, all plugins should not put anything in BOOT_DIR_ABS.
1922 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
1923 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
1924 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
1925 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
1926 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
1927 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
1928 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
1929 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
1930 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
1931 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
1932 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
1933 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
1934 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
1935 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
1936 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
1937 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
1938 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
1939 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
1940 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
1941 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
1942 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
1943 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
1944 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
1945 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
1946 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
1947 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
1948 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
1949 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
1950 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
1953 — Berlin, 2017-07-12
1957 * The "hybrid" control group mode has been modified to improve
1958 compatibility with "legacy" cgroups-v1 setups. Specifically, the
1959 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
1960 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
1961 cgroups-v1 hierarchy), the only externally visible change being that
1962 the cgroups-v2 hierarchy is also mounted, to
1963 /sys/fs/cgroup/unified. This should provide a large degree of
1964 compatibility with "legacy" cgroups-v1, while taking benefit of the
1965 better management capabilities of cgroups-v2.
1967 * The default control group setup mode may be selected both a boot-time
1968 via a set of kernel command line parameters (specifically:
1969 systemd.unified_cgroup_hierarchy= and
1970 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
1971 default selected on the configure command line
1972 (--with-default-hierarchy=). The upstream default is "hybrid"
1973 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
1974 this will change in a future systemd version to be "unified" (pure
1975 cgroups-v2 mode). The third option for the compile time option is
1976 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
1977 distributions to default to "hybrid" mode for release distributions,
1978 starting with v233. We recommend "unified" for development
1979 distributions (specifically: distributions such as Fedora's rawhide)
1980 as that's where things are headed in the long run. Use "legacy" for
1981 greatest stability and compatibility only.
1983 * Note one current limitation of "unified" and "hybrid" control group
1984 setup modes: the kernel currently does not permit the systemd --user
1985 instance (i.e. unprivileged code) to migrate processes between two
1986 disconnected cgroup subtrees, even if both are managed and owned by
1987 the user. This effectively means "systemd-run --user --scope" doesn't
1988 work when invoked from outside of any "systemd --user" service or
1989 scope. Specifically, it is not supported from session scopes. We are
1990 working on fixing this in a future systemd version. (See #3388 for
1991 further details about this.)
1993 * DBus policy files are now installed into /usr rather than /etc. Make
1994 sure your system has dbus >= 1.9.18 running before upgrading to this
1995 version, or override the install path with --with-dbuspolicydir= .
1997 * All python scripts shipped with systemd (specifically: the various
1998 tests written in Python) now require Python 3.
2000 * systemd unit tests can now run standalone (without the source or
2001 build directories), and can be installed into /usr/lib/systemd/tests/
2002 with 'make install-tests'.
2004 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
2005 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
2008 * Support for the %c, %r, %R specifiers in unit files has been
2009 removed. Specifiers are not supposed to be dependent on configuration
2010 in the unit file itself (so that they resolve the same regardless
2011 where used in the unit files), but these specifiers were influenced
2012 by the Slice= option.
2014 * The shell invoked by debug-shell.service now defaults to /bin/sh in
2015 all cases. If distributions want to use a different shell for this
2016 purpose (for example Fedora's /sbin/sushell) they need to specify
2017 this explicitly at configure time using --with-debug-shell=.
2019 * The confirmation spawn prompt has been reworked to offer the
2022 (c)ontinue, proceed without asking anymore
2023 (D)ump, show the state of the unit
2024 (f)ail, don't execute the command and pretend it failed
2026 (i)nfo, show a short summary of the unit
2027 (j)obs, show jobs that are in progress
2028 (s)kip, don't execute the command and pretend it succeeded
2029 (y)es, execute the command
2031 The 'n' choice for the confirmation spawn prompt has been removed,
2032 because its meaning was confusing.
2034 The prompt may now also be redirected to an alternative console by
2035 specifying the console as parameter to systemd.confirm_spawn=.
2037 * Services of Type=notify require a READY=1 notification to be sent
2038 during startup. If no such message is sent, the service now fails,
2039 even if the main process exited with a successful exit code.
2041 * Services that fail to start up correctly now always have their
2042 ExecStopPost= commands executed. Previously, they'd enter "failed"
2043 state directly, without executing these commands.
2045 * The option MulticastDNS= of network configuration files has acquired
2046 an actual implementation. With MulticastDNS=yes a host can resolve
2047 names of remote hosts and reply to mDNS A and AAAA requests.
2049 * When units are about to be started an additional check is now done to
2050 ensure that all dependencies of type BindsTo= (when used in
2051 combination with After=) have been started.
2053 * systemd-analyze gained a new verb "syscall-filter" which shows which
2054 system call groups are defined for the SystemCallFilter= unit file
2055 setting, and which system calls they contain.
2057 * A new system call filter group "@filesystem" has been added,
2058 consisting of various file system related system calls. Group
2059 "@reboot" has been added, covering reboot, kexec and shutdown related
2060 calls. Finally, group "@swap" has been added covering swap
2061 configuration related calls.
2063 * A new unit file option RestrictNamespaces= has been added that may be
2064 used to restrict access to the various process namespace types the
2065 Linux kernel provides. Specifically, it may be used to take away the
2066 right for a service unit to create additional file system, network,
2067 user, and other namespaces. This sandboxing option is particularly
2068 relevant due to the high amount of recently discovered namespacing
2069 related vulnerabilities in the kernel.
2071 * systemd-udev's .link files gained support for a new AutoNegotiation=
2072 setting for configuring Ethernet auto-negotiation.
2074 * systemd-networkd's .network files gained support for a new
2075 ListenPort= setting in the [DHCP] section to explicitly configure the
2076 UDP client port the DHCP client shall listen on.
2078 * .network files gained a new Unmanaged= boolean setting for explicitly
2079 excluding one or more interfaces from management by systemd-networkd.
2081 * The systemd-networkd ProxyARP= option has been renamed to
2082 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
2083 renamed to ReduceARPProxy=. The old names continue to be available
2086 * systemd-networkd gained support for configuring IPv6 Proxy NDP
2087 addresses via the new IPv6ProxyNDPAddress= .network file setting.
2089 * systemd-networkd's bonding device support gained support for two new
2090 configuration options ActiveSlave= and PrimarySlave=.
2092 * The various options in the [Match] section of .network files gained
2093 support for negative matching.
2095 * New systemd-specific mount options are now understood in /etc/fstab:
2097 x-systemd.mount-timeout= may be used to configure the maximum
2098 permitted runtime of the mount command.
2100 x-systemd.device-bound may be set to bind a mount point to its
2101 backing device unit, in order to automatically remove a mount point
2102 if its backing device is unplugged. This option may also be
2103 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
2104 on the block device, which is now automatically set for all CDROM
2105 drives, so that mounted CDs are automatically unmounted when they are
2106 removed from the drive.
2108 x-systemd.after= and x-systemd.before= may be used to explicitly
2109 order a mount after or before another unit or mount point.
2111 * Enqueued start jobs for device units are now automatically garbage
2112 collected if there are no jobs waiting for them anymore.
2114 * systemctl list-jobs gained two new switches: with --after, for every
2115 queued job the jobs it's waiting for are shown; with --before the
2116 jobs which it's blocking are shown.
2118 * systemd-nspawn gained support for ephemeral boots from disk images
2119 (or in other words: --ephemeral and --image= may now be
2120 combined). Moreover, ephemeral boots are now supported for normal
2121 directories, even if the backing file system is not btrfs. Of course,
2122 if the file system does not support file system snapshots or
2123 reflinks, the initial copy operation will be relatively expensive, but
2124 this should still be suitable for many use cases.
2126 * Calendar time specifications in .timer units now support
2127 specifications relative to the end of a month by using "~" instead of
2128 "-" as separator between month and day. For example, "*-02~03" means
2129 "the third last day in February". In addition a new syntax for
2130 repeated events has been added using the "/" character. For example,
2131 "9..17/2:00" means "every two hours from 9am to 5pm".
2133 * systemd-socket-proxyd gained a new parameter --connections-max= for
2134 configuring the maximum number of concurrent connections.
2136 * sd-id128 gained a new API for generating unique IDs for the host in a
2137 way that does not leak the machine ID. Specifically,
2138 sd_id128_get_machine_app_specific() derives an ID based on the
2139 machine ID a in well-defined, non-reversible, stable way. This is
2140 useful whenever an identifier for the host is needed but where the
2141 identifier shall not be useful to identify the system beyond the
2142 scope of the application itself. (Internally this uses HMAC-SHA256 as
2143 keyed hash function using the machine ID as input.)
2145 * NotifyAccess= gained a new supported value "exec". When set
2146 notifications are accepted from all processes systemd itself invoked,
2147 including all control processes.
2149 * .nspawn files gained support for defining overlay mounts using the
2150 Overlay= and OverlayReadOnly= options. Previously this functionality
2151 was only available on the systemd-nspawn command line.
2153 * systemd-nspawn's --bind= and --overlay= options gained support for
2154 bind/overlay mounts whose source lies within the container tree by
2155 prefixing the source path with "+".
2157 * systemd-nspawn's --bind= and --overlay= options gained support for
2158 automatically allocating a temporary source directory in /var/tmp
2159 that is removed when the container dies. Specifically, if the source
2160 directory is specified as empty string this mechanism is selected. An
2161 example usage is --overlay=+/var::/var, which creates an overlay
2162 mount based on the original /var contained in the image, overlayed
2163 with a temporary directory in the host's /var/tmp. This way changes
2164 to /var are automatically flushed when the container shuts down.
2166 * systemd-nspawn --image= option does now permit raw file system block
2167 devices (in addition to images containing partition tables, as
2170 * The disk image dissection logic in systemd-nspawn gained support for
2171 automatically setting up LUKS encrypted as well as Verity protected
2172 partitions. When a container is booted from an encrypted image the
2173 passphrase is queried at start-up time. When a container with Verity
2174 data is started, the root hash is search in a ".roothash" file
2175 accompanying the disk image (alternatively, pass the root hash via
2176 the new --root-hash= command line option).
2178 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
2179 be used to dissect disk images the same way as systemd-nspawn does
2180 it, following the Bootable Partition Specification. It may even be
2181 used to mount disk images with complex partition setups (including
2182 LUKS and Verity partitions) to a local host directory, in order to
2183 inspect them. This tool is not considered public API (yet), and is
2184 thus not installed into /usr/bin. Please do not rely on its
2185 existence, since it might go away or be changed in later systemd
2188 * A new generator "systemd-verity-generator" has been added, similar in
2189 style to "systemd-cryptsetup-generator", permitting automatic setup of
2190 Verity root partitions when systemd boots up. In order to make use of
2191 this your partition setup should follow the Discoverable Partitions
2192 Specification, and the GPT partition ID of the root file system
2193 partition should be identical to the upper 128bit of the Verity root
2194 hash. The GPT partition ID of the Verity partition protecting it
2195 should be the lower 128bit of the Verity root hash. If the partition
2196 image follows this model it is sufficient to specify a single
2197 "roothash=" kernel command line argument to both configure which root
2198 image and verity partition to use as well as the root hash for
2199 it. Note that systemd-nspawn's Verity support follows the same
2200 semantics, meaning that disk images with proper Verity data in place
2201 may be booted in containers with systemd-nspawn as well as on
2202 physical systems via the verity generator. Also note that the "mkosi"
2203 tool available at https://github.com/systemd/mkosi has been updated
2204 to generate Verity protected disk images following this scheme. In
2205 fact, it has been updated to generate disk images that optionally
2206 implement a complete UEFI SecureBoot trust chain, involving a signed
2207 kernel and initrd image that incorporates such a root hash as well as
2208 a Verity-enabled root partition.
2210 * The hardware database (hwdb) udev supports has been updated to carry
2211 accelerometer quirks.
2213 * All system services are now run with a fresh kernel keyring set up
2214 for them. The invocation ID is stored by default in it, thus
2215 providing a safe, non-overridable way to determine the invocation
2218 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
2219 options for bind mounting arbitrary paths in a service-specific
2220 way. When these options are used, arbitrary host or service files and
2221 directories may be mounted to arbitrary locations in the service's
2224 * Documentation has been added that lists all of systemd's low-level
2225 environment variables:
2227 https://github.com/systemd/systemd/blob/master/docs/ENVIRONMENT.md
2229 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
2230 whether a specific socket file descriptor matches a specified socket
2233 * systemd-firstboot has been updated to check for the
2234 systemd.firstboot= kernel command line option. It accepts a boolean
2235 and when set to false the first boot questions are skipped.
2237 * systemd-fstab-generator has been updated to check for the
2238 systemd.volatile= kernel command line option, which either takes an
2239 optional boolean parameter or the special value "state". If used the
2240 system may be booted in a "volatile" boot mode. Specifically,
2241 "systemd.volatile" is used, the root directory will be mounted as
2242 tmpfs, and only /usr is mounted from the actual root file system. If
2243 "systemd.volatile=state" is used, the root directory will be mounted
2244 as usual, but /var is mounted as tmpfs. This concept provides similar
2245 functionality as systemd-nspawn's --volatile= option, but provides it
2246 on physical boots. Use this option for implementing stateless
2247 systems, or testing systems with all state and/or configuration reset
2248 to the defaults. (Note though that many distributions are not
2249 prepared to boot up without a populated /etc or /var, though.)
2251 * systemd-gpt-auto-generator gained support for LUKS encrypted root
2252 partitions. Previously it only supported LUKS encrypted partitions
2253 for all other uses, except for the root partition itself.
2255 * Socket units gained support for listening on AF_VSOCK sockets for
2256 communication in virtualized QEMU environments.
2258 * The "configure" script gained a new option --with-fallback-hostname=
2259 for specifying the fallback hostname to use if none is configured in
2260 /etc/hostname. For example, by specifying
2261 --with-fallback-hostname=fedora it is possible to default to a
2262 hostname of "fedora" on pristine installations.
2264 * systemd-cgls gained support for a new --unit= switch for listing only
2265 the control groups of a specific unit. Similar --user-unit= has been
2266 added for listing only the control groups of a specific user unit.
2268 * systemd-mount gained a new --umount switch for unmounting a mount or
2269 automount point (and all mount/automount points below it).
2271 * systemd will now refuse full configuration reloads (via systemctl
2272 daemon-reload and related calls) unless at least 16MiB of free space
2273 are available in /run. This is a safety precaution in order to ensure
2274 that generators can safely operate after the reload completed.
2276 * A new unit file option RootImage= has been added, which has a similar
2277 effect as RootDirectory= but mounts the service's root directory from
2278 a disk image instead of plain directory. This logic reuses the same
2279 image dissection and mount logic that systemd-nspawn already uses,
2280 and hence supports any disk images systemd-nspawn supports, including
2281 those following the Discoverable Partition Specification, as well as
2282 Verity enabled images. This option enables systemd to run system
2283 services directly off disk images acting as resource bundles,
2284 possibly even including full integrity data.
2286 * A new MountAPIVFS= unit file option has been added, taking a boolean
2287 argument. If enabled /proc, /sys and /dev (collectively called the
2288 "API VFS") will be mounted for the service. This is only relevant if
2289 RootDirectory= or RootImage= is used for the service, as these mounts
2290 are of course in place in the host mount namespace anyway.
2292 * systemd-nspawn gained support for a new --pivot-root= switch. If
2293 specified the root directory within the container image is pivoted to
2294 the specified mount point, while the original root disk is moved to a
2295 different place. This option enables booting of ostree images
2296 directly with systemd-nspawn.
2298 * The systemd build scripts will no longer complain if the NTP server
2299 addresses are not changed from the defaults. Google now supports
2300 these NTP servers officially. We still recommend downstreams to
2301 properly register an NTP pool with the NTP pool project though.
2303 * coredumpctl gained a new "--reverse" option for printing the list
2304 of coredumps in reverse order.
2306 * coredumpctl will now show additional information about truncated and
2307 inaccessible coredumps, as well as coredumps that are still being
2308 processed. It also gained a new --quiet switch for suppressing
2309 additional informational message in its output.
2311 * coredumpctl gained support for only showing coredumps newer and/or
2312 older than specific timestamps, using the new --since= and --until=
2313 options, reminiscent of journalctl's options by the same name.
2315 * The systemd-coredump logic has been improved so that it may be reused
2316 to collect backtraces in non-compiled languages, for example in
2317 scripting languages such as Python.
2319 * machinectl will now show the UID shift of local containers, if user
2320 namespacing is enabled for them.
2322 * systemd will now optionally run "environment generator" binaries at
2323 configuration load time. They may be used to add environment
2324 variables to the environment block passed to services invoked. One
2325 user environment generator is shipped by default that sets up
2326 environment variables based on files dropped into /etc/environment.d
2327 and ~/.config/environment.d/.
2329 * systemd-resolved now includes the new, recently published 2017 DNSSEC
2332 * hostnamed has been updated to report a new chassis type of
2333 "convertible" to cover "foldable" laptops that can both act as a
2334 tablet and as a laptop, such as various Lenovo Yoga devices.
2336 Contributions from: Adrián López, Alexander Galanin, Alexander
2337 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
2338 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
2339 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
2340 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
2341 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
2342 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
2343 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
2344 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
2345 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
2346 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
2347 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
2348 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
2349 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
2350 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
2351 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
2352 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
2353 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
2354 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
2355 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
2356 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
2357 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
2358 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
2359 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
2360 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
2361 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
2362 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
2363 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
2366 — Berlin, 2017-03-01
2370 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
2371 RestrictAddressFamilies= enabled. These sandboxing options should
2372 generally be compatible with the various external udev call-out
2373 binaries we are aware of, however there may be exceptions, in
2374 particular when exotic languages for these call-outs are used. In
2375 this case, consider turning off these settings locally.
2377 * The new RemoveIPC= option can be used to remove IPC objects owned by
2378 the user or group of a service when that service exits.
2380 * The new ProtectKernelModules= option can be used to disable explicit
2381 load and unload operations of kernel modules by a service. In
2382 addition access to /usr/lib/modules is removed if this option is set.
2384 * ProtectSystem= option gained a new value "strict", which causes the
2385 whole file system tree with the exception of /dev, /proc, and /sys,
2386 to be remounted read-only for a service.
2388 * The new ProtectKernelTunables= option can be used to disable
2389 modification of configuration files in /sys and /proc by a service.
2390 Various directories and files are remounted read-only, so access is
2391 restricted even if the file permissions would allow it.
2393 * The new ProtectControlGroups= option can be used to disable write
2394 access by a service to /sys/fs/cgroup.
2396 * Various systemd services have been hardened with
2397 ProtectKernelTunables=yes, ProtectControlGroups=yes,
2398 RestrictAddressFamilies=.
2400 * Support for dynamically creating users for the lifetime of a service
2401 has been added. If DynamicUser=yes is specified, user and group IDs
2402 will be allocated from the range 61184..65519 for the lifetime of the
2403 service. They can be resolved using the new nss-systemd.so NSS
2404 module. The module must be enabled in /etc/nsswitch.conf. Services
2405 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
2406 any resources allocated by the service will be cleaned up when the
2407 service exits. They also have ProtectHome=read-only and
2408 ProtectSystem=strict enabled, so they are not able to make any
2409 permanent modifications to the system.
2411 * The nss-systemd module also always resolves root and nobody, making
2412 it possible to have no /etc/passwd or /etc/group files in minimal
2413 container or chroot environments.
2415 * Services may be started with their own user namespace using the new
2416 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
2417 under which the service is running are mapped. All other users are
2420 * Support for the cgroup namespace has been added to systemd-nspawn. If
2421 supported by kernel, the container system started by systemd-nspawn
2422 will have its own view of the cgroup hierarchy. This new behaviour
2423 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
2425 * The new MemorySwapMax= option can be used to limit the maximum swap
2426 usage under the unified cgroup hierarchy.
2428 * Support for the CPU controller in the unified cgroup hierarchy has
2429 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
2430 options. This controller requires out-of-tree patches for the kernel
2431 and the support is provisional.
2433 * Mount and automount units may now be created transiently
2434 (i.e. dynamically at runtime via the bus API, instead of requiring
2435 unit files in the file system).
2437 * systemd-mount is a new tool which may mount file systems – much like
2438 mount(8), optionally pulling in additional dependencies through
2439 transient .mount and .automount units. For example, this tool
2440 automatically runs fsck on a backing block device before mounting,
2441 and allows the automount logic to be used dynamically from the
2442 command line for establishing mount points. This tool is particularly
2443 useful when dealing with removable media, as it will ensure fsck is
2444 run – if necessary – before the first access and that the file system
2445 is quickly unmounted after each access by utilizing the automount
2446 logic. This maximizes the chance that the file system on the
2447 removable media stays in a clean state, and if it isn't in a clean
2448 state is fixed automatically.
2450 * LazyUnmount=yes option for mount units has been added to expose the
2451 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
2454 * /efi will be used as the mount point of the EFI boot partition, if
2455 the directory is present, and the mount point was not configured
2456 through other means (e.g. fstab). If /efi directory does not exist,
2457 /boot will be used as before. This makes it easier to automatically
2458 mount the EFI partition on systems where /boot is used for something
2461 * When operating on GPT disk images for containers, systemd-nspawn will
2462 now mount the ESP to /boot or /efi according to the same rules as PID
2463 1 running on a host. This allows tools like "bootctl" to operate
2464 correctly within such containers, in order to make container images
2465 bootable on physical systems.
2467 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
2469 * Two new user session targets have been added to support running
2470 graphical sessions under the systemd --user instance:
2471 graphical-session.target and graphical-session-pre.target. See
2472 systemd.special(7) for a description of how those targets should be
2475 * The vconsole initialization code has been significantly reworked to
2476 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
2477 support unicode keymaps. Font and keymap configuration will now be
2478 copied to all allocated virtual consoles.
2480 * FreeBSD's bhyve virtualization is now detected.
2482 * Information recorded in the journal for core dumps now includes the
2483 contents of /proc/mountinfo and the command line of the process at
2484 the top of the process hierarchy (which is usually the init process
2487 * systemd-journal-gatewayd learned the --directory= option to serve
2488 files from the specified location.
2490 * journalctl --root=… can be used to peruse the journal in the
2491 /var/log/ directories inside of a container tree. This is similar to
2492 the existing --machine= option, but does not require the container to
2495 * The hardware database has been extended to support
2496 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
2499 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
2500 specify the click rate for mice which include a horizontal wheel with
2501 a click rate that is different than the one for the vertical wheel.
2503 * systemd-run gained a new --wait option that makes service execution
2504 synchronous. (Specifically, the command will not return until the
2505 specified service binary exited.)
2507 * systemctl gained a new --wait option that causes the start command to
2508 wait until the units being started have terminated again.
2510 * A new journal output mode "short-full" has been added which displays
2511 timestamps with abbreviated English day names and adds a timezone
2512 suffix. Those timestamps include more information than the default
2513 "short" output mode, and can be passed directly to journalctl's
2514 --since= and --until= options.
2516 * /etc/resolv.conf will be bind-mounted into containers started by
2517 systemd-nspawn, if possible, so any changes to resolv.conf contents
2518 are automatically propagated to the container.
2520 * The number of instances for socket-activated services originating
2521 from a single IP address can be limited with
2522 MaxConnectionsPerSource=, extending the existing setting of
2525 * systemd-networkd gained support for vcan ("Virtual CAN") interface
2528 * .netdev and .network configuration can now be extended through
2531 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
2532 Segmentation Offload, Generic Receive Offload, Large Receive Offload
2533 can be enabled and disabled using the new UDPSegmentationOffload=,
2534 TCPSegmentationOffload=, GenericSegmentationOffload=,
2535 GenericReceiveOffload=, LargeReceiveOffload= options in the
2536 [Link] section of .link files.
2538 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
2539 Port VLAN ID can be configured for bridge devices using the new STP=,
2540 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
2541 section of .netdev files.
2543 * The route table to which routes received over DHCP or RA should be
2544 added can be configured with the new RouteTable= option in the [DHCP]
2545 and [IPv6AcceptRA] sections of .network files.
2547 * The Address Resolution Protocol can be disabled on links managed by
2548 systemd-networkd using the ARP=no setting in the [Link] section of
2551 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
2552 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
2553 encode information about the result and exit codes of the current
2554 service runtime cycle.
2556 * systemd-sysctl will now configure kernel parameters in the order
2557 they occur in the configuration files. This matches what sysctl
2558 has been traditionally doing.
2560 * kernel-install "plugins" that are executed to perform various
2561 tasks after a new kernel is added and before an old one is removed
2562 can now return a special value to terminate the procedure and
2563 prevent any later plugins from running.
2565 * Journald's SplitMode=login setting has been deprecated. It has been
2566 removed from documentation, and its use is discouraged. In a future
2567 release it will be completely removed, and made equivalent to current
2568 default of SplitMode=uid.
2570 * Storage=both option setting in /etc/systemd/coredump.conf has been
2571 removed. With fast LZ4 compression storing the core dump twice is not
2574 * The --share-system systemd-nspawn option has been replaced with an
2575 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
2576 this functionality is discouraged. In addition the variables
2577 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
2578 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
2579 individual namespaces.
2581 * "machinectl list" now shows the IP address of running containers in
2582 the output, as well as OS release information.
2584 * "loginctl list" now shows the TTY of each session in the output.
2586 * sd-bus gained new API calls sd_bus_track_set_recursive(),
2587 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
2588 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
2589 tracking objects in a "recursive" mode, where a single client can be
2590 counted multiple times, if it takes multiple references.
2592 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
2593 sd_bus_get_exit_on_disconnect(). They may be used to make a
2594 process using sd-bus automatically exit if the bus connection is
2597 * Bus clients of the service manager may now "pin" loaded units into
2598 memory, by taking an explicit reference on them. This is useful to
2599 ensure the client can retrieve runtime data about the service even
2600 after the service completed execution. Taking such a reference is
2601 available only for privileged clients and should be helpful to watch
2602 running services in a race-free manner, and in particular collect
2603 information about exit statuses and results.
2605 * The nss-resolve module has been changed to strictly return UNAVAIL
2606 when communication via D-Bus with resolved failed, and NOTFOUND when
2607 a lookup completed but was negative. This means it is now possible to
2608 neatly configure fallbacks using nsswitch.conf result checking
2609 expressions. Taking benefit of this, the new recommended
2610 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
2612 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
2614 * A new setting CtrlAltDelBurstAction= has been added to
2615 /etc/systemd/system.conf which may be used to configure the precise
2616 behaviour if the user on the console presses Ctrl-Alt-Del more often
2617 than 7 times in 2s. Previously this would unconditionally result in
2618 an expedited, immediate reboot. With this new setting the precise
2619 operation may be configured in more detail, and also turned off
2622 * In .netdev files two new settings RemoteChecksumTx= and
2623 RemoteChecksumRx= are now understood that permit configuring the
2624 remote checksumming logic for VXLAN networks.
2626 * The service manager learnt a new "invocation ID" concept for invoked
2627 services. Each runtime cycle of a service will get a new invocation
2628 ID (a 128bit random UUID) assigned that identifies the current
2629 run of the service uniquely and globally. A new invocation ID
2630 is generated each time a service starts up. The journal will store
2631 the invocation ID of a service along with any logged messages, thus
2632 making the invocation ID useful for matching the online runtime of a
2633 service with the offline log data it generated in a safe way without
2634 relying on synchronized timestamps. In many ways this new service
2635 invocation ID concept is similar to the kernel's boot ID concept that
2636 uniquely and globally identifies the runtime of each boot. The
2637 invocation ID of a service is passed to the service itself via an
2638 environment variable ($INVOCATION_ID). A new bus call
2639 GetUnitByInvocationID() has been added that is similar to GetUnit()
2640 but instead of retrieving the bus path for a unit by its name
2641 retrieves it by its invocation ID. The returned path is valid only as
2642 long as the passed invocation ID is current.
2644 * systemd-resolved gained a new "DNSStubListener" setting in
2645 resolved.conf. It either takes a boolean value or the special values
2646 "udp" and "tcp", and configures whether to enable the stub DNS
2647 listener on 127.0.0.53:53.
2649 * IP addresses configured via networkd may now carry additional
2650 configuration settings supported by the kernel. New options include:
2651 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
2652 PrefixRoute=, AutoJoin=.
2654 * The PAM configuration fragment file for "user@.service" shipped with
2655 systemd (i.e. the --user instance of systemd) has been stripped to
2656 the minimum necessary to make the system boot. Previously, it
2657 contained Fedora-specific stanzas that did not apply to other
2658 distributions. It is expected that downstream distributions add
2659 additional configuration lines, matching their needs to this file,
2660 using it only as rough template of what systemd itself needs. Note
2661 that this reduced fragment does not even include an invocation of
2662 pam_limits which most distributions probably want to add, even though
2663 systemd itself does not need it. (There's also the new build time
2664 option --with-pamconfdir=no to disable installation of the PAM
2667 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
2668 capability is now also dropped from its set (in addition to
2669 CAP_SYS_MKNOD as before).
2671 * In service unit files it is now possible to connect a specific named
2672 file descriptor with stdin/stdout/stdout of an executed service. The
2673 name may be specified in matching .socket units using the
2674 FileDescriptorName= setting.
2676 * A number of journal settings may now be configured on the kernel
2677 command line. Specifically, the following options are now understood:
2678 systemd.journald.max_level_console=,
2679 systemd.journald.max_level_store=,
2680 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
2681 systemd.journald.max_level_wall=.
2683 * "systemctl is-enabled --full" will now show by which symlinks a unit
2684 file is enabled in the unit dependency tree.
2686 * Support for VeraCrypt encrypted partitions has been added to the
2687 "cryptsetup" logic and /etc/crypttab.
2689 * systemd-detect-virt gained support for a new --private-users switch
2690 that checks whether the invoking processes are running inside a user
2691 namespace. Similar, a new special value "private-users" for the
2692 existing ConditionVirtualization= setting has been added, permitting
2693 skipping of specific units in user namespace environments.
2695 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
2696 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
2697 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
2698 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
2699 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
2700 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
2701 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
2702 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
2703 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
2704 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
2705 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
2706 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
2707 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
2708 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
2709 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
2710 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
2711 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
2712 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
2713 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
2714 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
2715 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
2716 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
2717 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
2718 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
2719 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
2720 Jędrzejewski-Szmek, Zeal Jagannatha
2722 — Santa Fe, 2016-11-03
2726 * In service units the various ExecXYZ= settings have been extended
2727 with an additional special character as first argument of the
2728 assigned value: if the character '+' is used the specified command
2729 line it will be run with full privileges, regardless of User=,
2730 Group=, CapabilityBoundingSet= and similar options. The effect is
2731 similar to the existing PermissionsStartOnly= option, but allows
2732 configuration of this concept for each executed command line
2735 * Services may now alter the service watchdog timeout at runtime by
2736 sending a WATCHDOG_USEC= message via sd_notify().
2738 * MemoryLimit= and related unit settings now optionally take percentage
2739 specifications. The percentage is taken relative to the amount of
2740 physical memory in the system (or in case of containers, the assigned
2741 amount of memory). This allows scaling service resources neatly with
2742 the amount of RAM available on the system. Similarly, systemd-logind's
2743 RuntimeDirectorySize= option now also optionally takes percentage
2746 * In similar fashion TasksMax= takes percentage values now, too. The
2747 value is taken relative to the configured maximum number of processes
2748 on the system. The per-service task maximum has been changed to 15%
2749 using this functionality. (Effectively this is an increase of 512 →
2750 4915 for service units, given the kernel's default pid_max setting.)
2752 * Calendar time specifications in .timer units now understand a ".."
2753 syntax for time ranges. Example: "4..7:10" may now be used for
2754 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
2757 * The InaccessableDirectories=, ReadOnlyDirectories= and
2758 ReadWriteDirectories= unit file settings have been renamed to
2759 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
2760 applied to all kinds of file nodes, and not just directories, with
2761 the exception of symlinks. Specifically these settings may now be
2762 used on block and character device nodes, UNIX sockets and FIFOS as
2763 well as regular files. The old names of these settings remain
2764 available for compatibility.
2766 * systemd will now log about all service processes it kills forcibly
2767 (using SIGKILL) because they remained after the clean shutdown phase
2768 of the service completed. This should help identifying services that
2769 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
2770 systemd-logind's configuration a similar log message is generated for
2771 processes killed at the end of each session due to this setting.
2773 * systemd will now set the $JOURNAL_STREAM environment variable for all
2774 services whose stdout/stderr are connected to the Journal (which
2775 effectively means by default: all services). The variable contains
2776 the device and inode number of the file descriptor used for
2777 stdout/stderr. This may be used by invoked programs to detect whether
2778 their stdout/stderr is connected to the Journal, in which case they
2779 can switch over to direct Journal communication, thus being able to
2780 pass extended, structured metadata along with their log messages. As
2781 one example, this is now used by glib's logging primitives.
2783 * When using systemd's default tmp.mount unit for /tmp, the mount point
2784 will now be established with the "nosuid" and "nodev" options. This
2785 avoids privilege escalation attacks that put traps and exploits into
2786 /tmp. However, this might cause problems if you e. g. put container
2787 images or overlays into /tmp; if you need this, override tmp.mount's
2788 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
2791 * systemd now supports the "memory" cgroup controller also on
2794 * The systemd-cgtop tool now optionally takes a control group path as
2795 command line argument. If specified, the control group list shown is
2796 limited to subgroups of that group.
2798 * The SystemCallFilter= unit file setting gained support for
2799 pre-defined, named system call filter sets. For example
2800 SystemCallFilter=@clock is now an effective way to make all clock
2801 changing-related system calls unavailable to a service. A number of
2802 similar pre-defined groups are defined. Writing system call filters
2803 for system services is simplified substantially with this new
2804 concept. Accordingly, all of systemd's own, long-running services now
2805 enable system call filtering based on this, by default.
2807 * A new service setting MemoryDenyWriteExecute= has been added, taking
2808 a boolean value. If turned on, a service may no longer create memory
2809 mappings that are writable and executable at the same time. This
2810 enhances security for services where this is enabled as it becomes
2811 harder to dynamically write and then execute memory in exploited
2812 service processes. This option has been enabled for all of systemd's
2813 own long-running services.
2815 * A new RestrictRealtime= service setting has been added, taking a
2816 boolean argument. If set the service's processes may no longer
2817 acquire realtime scheduling. This improves security as realtime
2818 scheduling may otherwise be used to easily freeze the system.
2820 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
2821 value. This may be used for requesting that the system manager inside
2822 of the container reports start-up completion to nspawn which then
2823 propagates this notification further to the service manager
2824 supervising nspawn itself. A related option NotifyReady= in .nspawn
2825 files has been added too. This functionality allows ordering of the
2826 start-up of multiple containers using the usual systemd ordering
2829 * machinectl gained a new command "stop" that is an alias for
2832 * systemd-resolved gained support for contacting DNS servers on
2833 link-local IPv6 addresses.
2835 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
2836 its caches. A method call for requesting the same operation has been
2837 added to the bus API too, and is made available via "systemd-resolve
2840 * systemd-resolve gained a new --status switch. If passed a brief
2841 summary of the used DNS configuration with per-interface information
2844 * resolved.conf gained a new Cache= boolean option, defaulting to
2845 on. If turned off local DNS caching is disabled. This comes with a
2846 performance penalty in particular when DNSSEC is enabled. Note that
2847 resolved disables its internal caching implicitly anyway, when the
2848 configured DNS server is on a host-local IP address such as ::1 or
2849 127.0.0.1, thus automatically avoiding double local caching.
2851 * systemd-resolved now listens on the local IP address 127.0.0.53:53
2852 for DNS requests. This improves compatibility with local programs
2853 that do not use the libc NSS or systemd-resolved's bus APIs for name
2854 resolution. This minimal DNS service is only available to local
2855 programs and does not implement the full DNS protocol, but enough to
2856 cover local DNS clients. A new, static resolv.conf file, listing just
2857 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
2858 now recommended to make /etc/resolv.conf a symlink to this file in
2859 order to route all DNS lookups to systemd-resolved, regardless if
2860 done via NSS, the bus API or raw DNS packets. Note that this local
2861 DNS service is not as fully featured as the libc NSS or
2862 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
2863 used to deliver link-local address information (as this implies
2864 sending a local interface index along), LLMNR/mDNS support via this
2865 interface is severely restricted. It is thus strongly recommended for
2866 all applications to use the libc NSS API or native systemd-resolved
2869 * systemd-networkd's bridge support learned a new setting
2870 VLANFiltering= for controlling VLAN filtering. Moreover a new section
2871 in .network files has been added for configuring VLAN bridging in
2872 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
2874 * systemd-networkd's IPv6 Router Advertisement code now makes use of
2875 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
2876 now be acquired without relying on DHCPv6. Two new options
2877 UseDomains= and UseDNS= have been added to configure this behaviour.
2879 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
2880 renamed IPv6AcceptRA=, without altering its behaviour. The old
2881 setting name remains available for compatibility reasons.
2883 * The systemd-networkd VTI/VTI6 tunneling support gained new options
2884 Key=, InputKey= and OutputKey=.
2886 * systemd-networkd gained support for VRF ("Virtual Routing Function")
2887 interface configuration.
2889 * "systemctl edit" may now be used to create new unit files by
2890 specifying the --force switch.
2892 * sd-event gained a new function sd_event_get_iteration() for
2893 requesting the current iteration counter of the event loop. It starts
2894 at zero and is increased by one with each event loop iteration.
2896 * A new rpm macro %systemd_ordering is provided by the macros.systemd
2897 file. It can be used in lieu of %systemd_requires in packages which
2898 don't use any systemd functionality and are intended to be installed
2899 in minimal containers without systemd present. This macro provides
2900 ordering dependencies to ensure that if the package is installed in
2901 the same rpm transaction as systemd, systemd will be installed before
2902 the scriptlets for the package are executed, allowing unit presets
2905 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
2906 been added to simplify packaging of generators.
2908 * The os-release file gained VERSION_CODENAME field for the
2909 distribution nickname (e.g. VERSION_CODENAME=woody).
2911 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
2912 can be set to disable parsing of metadata and the creation
2913 of persistent symlinks for that device.
2915 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
2916 to make them available to logged-in users has been reverted.
2918 * Much of the common code of the various systemd components is now
2919 built into an internal shared library libsystemd-shared-231.so
2920 (incorporating the systemd version number in the name, to be updated
2921 with future releases) that the components link to. This should
2922 decrease systemd footprint both in memory during runtime and on
2923 disk. Note that the shared library is not for public use, and is
2924 neither API nor ABI stable, but is likely to change with every new
2925 released update. Packagers need to make sure that binaries
2926 linking to libsystemd-shared.so are updated in step with the
2929 * Configuration for "mkosi" is now part of the systemd
2930 repository. mkosi is a tool to easily build legacy-free OS images,
2931 and is available on github: https://github.com/systemd/mkosi. If
2932 "mkosi" is invoked in the build tree a new raw OS image is generated
2933 incorporating the systemd sources currently being worked on and a
2934 clean, fresh distribution installation. The generated OS image may be
2935 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
2936 UEFI PC. This functionality is particularly useful to easily test
2937 local changes made to systemd in a pristine, defined environment. See
2938 doc/HACKING for details.
2940 * configure learned the --with-support-url= option to specify the
2941 distribution's bugtracker.
2943 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
2944 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
2945 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
2946 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
2947 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
2948 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
2949 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
2950 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
2951 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
2952 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
2953 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
2954 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
2955 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
2956 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
2957 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
2958 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
2959 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
2960 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
2961 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
2963 — Berlin, 2016-07-25
2967 * DNSSEC is now turned on by default in systemd-resolved (in
2968 "allow-downgrade" mode), but may be turned off during compile time by
2969 passing "--with-default-dnssec=no" to "configure" (and of course,
2970 during runtime with DNSSEC= in resolved.conf). We recommend
2971 downstreams to leave this on at least during development cycles and
2972 report any issues with the DNSSEC logic upstream. We are very
2973 interested in collecting feedback about the DNSSEC validator and its
2974 limitations in the wild. Note however, that DNSSEC support is
2975 probably nothing downstreams should turn on in stable distros just
2976 yet, as it might create incompatibilities with a few DNS servers and
2977 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
2978 automatically whenever we detect such incompatible setups, but there
2979 might be systems we do not cover yet. Hence: please help us testing
2980 the DNSSEC code, leave this on where you can, report back, but then
2981 again don't consider turning this on in your stable, LTS or
2982 production release just yet. (Note that you have to enable
2983 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
2984 and its DNSSEC mode for host name resolution from local
2987 * systemd-resolve conveniently resolves DANE records with the --tlsa
2988 option and OPENPGPKEY records with the --openpgp option. It also
2989 supports dumping raw DNS record data via the new --raw= switch.
2991 * systemd-logind will now by default terminate user processes that are
2992 part of the user session scope unit (session-XX.scope) when the user
2993 logs out. This behavior is controlled by the KillUserProcesses=
2994 setting in logind.conf, and the previous default of "no" is now
2995 changed to "yes". This means that user sessions will be properly
2996 cleaned up after, but additional steps are necessary to allow
2997 intentionally long-running processes to survive logout.
2999 While the user is logged in at least once, user@.service is running,
3000 and any service that should survive the end of any individual login
3001 session can be started at a user service or scope using systemd-run.
3002 systemd-run(1) man page has been extended with an example which shows
3003 how to run screen in a scope unit underneath user@.service. The same
3004 command works for tmux.
3006 After the user logs out of all sessions, user@.service will be
3007 terminated too, by default, unless the user has "lingering" enabled.
3008 To effectively allow users to run long-term tasks even if they are
3009 logged out, lingering must be enabled for them. See loginctl(1) for
3010 details. The default polkit policy was modified to allow users to
3011 set lingering for themselves without authentication.
3013 Previous defaults can be restored at compile time by the
3014 --without-kill-user-processes option to "configure".
3016 * systemd-logind gained new configuration settings SessionsMax= and
3017 InhibitorsMax=, both with a default of 8192. It will not register new
3018 user sessions or inhibitors above this limit.
3020 * systemd-logind will now reload configuration on SIGHUP.
3022 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
3023 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
3024 enable. Also, support for the "io" cgroup controller in the unified
3025 hierarchy has been added, so that the "memory", "pids" and "io" are
3026 now the controllers that are supported on the unified hierarchy.
3028 WARNING: it is not possible to use previous systemd versions with
3029 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
3030 is necessary to also update systemd in the initramfs if using the
3031 unified hierarchy. An updated SELinux policy is also required.
3033 * LLDP support has been extended, and both passive (receive-only) and
3034 active (sender) modes are supported. Passive mode ("routers-only") is
3035 enabled by default in systemd-networkd. Active LLDP mode is enabled
3036 by default for containers on the internal network. The "networkctl
3037 lldp" command may be used to list information gathered. "networkctl
3038 status" will also show basic LLDP information on connected peers now.
3040 * The IAID and DUID unique identifier sent in DHCP requests may now be
3041 configured for the system and each .network file managed by
3042 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
3044 * systemd-networkd gained support for configuring proxy ARP support for
3045 each interface, via the ProxyArp= setting in .network files. It also
3046 gained support for configuring the multicast querier feature of
3047 bridge devices, via the new MulticastQuerier= setting in .netdev
3048 files. Similarly, snooping on the IGMP traffic can be controlled
3049 via the new setting MulticastSnooping=.
3051 A new setting PreferredLifetime= has been added for addresses
3052 configured in .network file to configure the lifetime intended for an
3055 The systemd-networkd DHCP server gained the option EmitRouter=, which
3056 defaults to yes, to configure whether the DHCP Option 3 (Router)
3059 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
3060 systemd-socket-activate and installed into /usr/bin. It is now fully
3063 * systemd-journald now uses separate threads to flush changes to disk
3064 when closing journal files, thus reducing impact of slow disk I/O on
3065 logging performance.
3067 * The sd-journal API gained two new calls
3068 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
3069 can be used to open journal files using file descriptors instead of
3070 file or directory paths. sd_journal_open_container() has been
3071 deprecated, sd_journal_open_directory_fd() should be used instead
3072 with the flag SD_JOURNAL_OS_ROOT.
3074 * journalctl learned a new output mode "-o short-unix" that outputs log
3075 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
3076 UTC). It also gained support for a new --no-hostname setting to
3077 suppress the hostname column in the family of "short" output modes.
3079 * systemd-ask-password now optionally skips printing of the password to
3080 stdout with --no-output which can be useful in scripts.
3082 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
3083 (devices tagged with ID_MAKER_TOOL) are now tagged with
3084 "uaccess" and are available to logged in users.
3086 * The DeviceAllow= unit setting now supports specifiers (with "%").
3088 * "systemctl show" gained a new --value switch, which allows print a
3089 only the contents of a specific unit property, without also printing
3090 the property's name. Similar support was added to "show*" verbs
3091 of loginctl and machinectl that output "key=value" lists.
3093 * A new unit type "generated" was added for files dynamically generated
3094 by generator tools. Similarly, a new unit type "transient" is used
3095 for unit files created using the runtime API. "systemctl enable" will
3096 refuse to operate on such files.
3098 * A new command "systemctl revert" has been added that may be used to
3099 revert to the vendor version of a unit file, in case local changes
3100 have been made by adding drop-ins or overriding the unit file.
3102 * "machinectl clean" gained a new verb to automatically remove all or
3103 just hidden container images.
3105 * systemd-tmpfiles gained support for a new line type "e" for emptying
3106 directories, if they exist, without creating them if they don't.
3108 * systemd-nspawn gained support for automatically patching the UID/GIDs
3109 of the owners and the ACLs of all files and directories in a
3110 container tree to match the UID/GID user namespacing range selected
3111 for the container invocation. This mode is enabled via the new
3112 --private-users-chown switch. It also gained support for
3113 automatically choosing a free, previously unused UID/GID range when
3114 starting a container, via the new --private-users=pick setting (which
3115 implies --private-users-chown). Together, these options for the first
3116 time make user namespacing for nspawn containers fully automatic and
3117 thus deployable. The systemd-nspawn@.service template unit file has
3118 been changed to use this functionality by default.
3120 * systemd-nspawn gained a new --network-zone= switch, that allows
3121 creating ad-hoc virtual Ethernet links between multiple containers,
3122 that only exist as long as at least one container referencing them is
3123 running. This allows easy connecting of multiple containers with a
3124 common link that implements an Ethernet broadcast domain. Each of
3125 these network "zones" may be named relatively freely by the user, and
3126 may be referenced by any number of containers, but each container may
3127 only reference one of these "zones". On the lower level, this is
3128 implemented by an automatically managed bridge network interface for
3129 each zone, that is created when the first container referencing its
3130 zone is created and removed when the last one referencing its zone
3133 * The default start timeout may now be configured on the kernel command
3134 line via systemd.default_timeout_start_sec=. It was already
3135 configurable via the DefaultTimeoutStartSec= option in
3136 /etc/systemd/system.conf.
3138 * Socket units gained a new TriggerLimitIntervalSec= and
3139 TriggerLimitBurst= setting to configure a limit on the activation
3140 rate of the socket unit.
3142 * The LimitNICE= setting now optionally takes normal UNIX nice values
3143 in addition to the raw integer limit value. If the specified
3144 parameter is prefixed with "+" or "-" and is in the range -20..19 the
3145 value is understood as UNIX nice value. If not prefixed like this it
3146 is understood as raw RLIMIT_NICE limit.
3148 * Note that the effect of the PrivateDevices= unit file setting changed
3149 slightly with this release: the per-device /dev file system will be
3150 mounted read-only from this version on, and will have "noexec"
3151 set. This (minor) change of behavior might cause some (exceptional)
3152 legacy software to break, when PrivateDevices=yes is set for its
3153 service. Please leave PrivateDevices= off if you run into problems
3156 * systemd-bootchart has been split out to a separate repository:
3157 https://github.com/systemd/systemd-bootchart
3159 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
3160 merged into the kernel in its current form.
3162 * The compatibility libraries libsystemd-daemon.so,
3163 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
3164 which have been deprecated since systemd-209 have been removed along
3165 with the corresponding pkg-config files. All symbols provided by
3166 those libraries are provided by libsystemd.so.
3168 * The Capabilities= unit file setting has been removed (it is ignored
3169 for backwards compatibility). AmbientCapabilities= and
3170 CapabilityBoundingSet= should be used instead.
3172 * A new special target has been added, initrd-root-device.target,
3173 which creates a synchronization point for dependencies of the root
3174 device in early userspace. Initramfs builders must ensure that this
3175 target is now included in early userspace.
3177 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
3178 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
3179 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
3180 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
3181 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
3182 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
3183 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
3184 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
3185 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
3186 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
3187 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
3188 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
3189 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
3190 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
3191 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
3192 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
3193 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
3194 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
3195 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
3196 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
3197 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
3198 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
3199 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
3200 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
3201 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
3204 — Fairfax, 2016-05-21
3208 * The systemd-resolved DNS resolver service has gained a substantial
3209 set of new features, most prominently it may now act as a DNSSEC
3210 validating stub resolver. DNSSEC mode is currently turned off by
3211 default, but is expected to be turned on by default in one of the
3212 next releases. For now, we invite everybody to test the DNSSEC logic
3213 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
3214 service also gained a full set of D-Bus interfaces, including calls
3215 to configure DNS and DNSSEC settings per link (for use by external
3216 network management software). systemd-resolved and systemd-networkd
3217 now distinguish between "search" and "routing" domains. The former
3218 are used to qualify single-label names, the latter are used purely
3219 for routing lookups within certain domains to specific links.
3220 resolved now also synthesizes RRs for all entries from /etc/hosts.
3222 * The systemd-resolve tool (which is a client utility for
3223 systemd-resolved) has been improved considerably and is now fully
3224 supported and documented. Hence it has moved from /usr/lib/systemd to
3227 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
3230 * The coredump collection logic has been reworked: when a coredump is
3231 collected it is now written to disk, compressed and processed
3232 (including stacktrace extraction) from a new instantiated service
3233 systemd-coredump@.service, instead of directly from the
3234 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
3235 processing large coredumps can take up a substantial amount of
3236 resources and time, and this previously happened entirely outside of
3237 systemd's service supervision. With the new logic the core_pattern
3238 hook only does minimal metadata collection before passing off control
3239 to the new instantiated service, which is configured with a time
3240 limit, a nice level and other settings to minimize negative impact on
3241 the rest of the system. Also note that the new logic will honour the
3242 RLIMIT_CORE setting of the crashed process, which now allows users
3243 and processes to turn off coredumping for their processes by setting
3246 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
3247 and all forked processes by default. Previously, PID 1 would leave
3248 the setting at "0" for all processes, as set by the kernel. Note that
3249 the resource limit traditionally has no effect on the generated
3250 coredumps on the system if the /proc/sys/kernel/core_pattern hook
3251 logic is used. Since the limit is now honoured (see above) its
3252 default has been changed so that the coredumping logic is enabled by
3253 default for all processes, while allowing specific opt-out.
3255 * When the stacktrace is extracted from processes of system users, this
3256 is now done as "systemd-coredump" user, in order to sandbox this
3257 potentially security sensitive parsing operation. (Note that when
3258 processing coredumps of normal users this is done under the user ID
3259 of process that crashed, as before.) Packagers should take notice
3260 that it is now necessary to create the "systemd-coredump" system user
3261 and group at package installation time.
3263 * The systemd-activate socket activation testing tool gained support
3264 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
3265 and --seqpacket switches. It also has been extended to support both
3266 new-style and inetd-style file descriptor passing. Use the new
3267 --inetd switch to request inetd-style file descriptor passing.
3269 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
3270 variable, which takes a boolean value. If set to false, ANSI color
3271 output is disabled in the tools even when run on a terminal that
3274 * The VXLAN support in networkd now supports two new settings
3275 DestinationPort= and PortRange=.
3277 * A new systemd.machine_id= kernel command line switch has been added,
3278 that may be used to set the machine ID in /etc/machine-id if it is
3279 not initialized yet. This command line option has no effect if the
3280 file is already initialized.
3282 * systemd-nspawn gained a new --as-pid2 switch that invokes any
3283 specified command line as PID 2 rather than PID 1 in the
3284 container. In this mode PID 1 is a minimal stub init process that
3285 implements the special POSIX and Linux semantics of PID 1 regarding
3286 signal and child process management. Note that this stub init process
3287 is implemented in nspawn itself and requires no support from the
3288 container image. This new logic is useful to support running
3289 arbitrary commands in the container, as normal processes are
3290 generally not prepared to run as PID 1.
3292 * systemd-nspawn gained a new --chdir= switch for setting the current
3293 working directory for the process started in the container.
3295 * "journalctl /dev/sda" will now output all kernel log messages for
3296 specified device from the current boot, in addition to all devices
3297 that are parents of it. This should make log output about devices
3298 pretty useful, as long as kernel drivers attach enough metadata to
3299 the log messages. (The usual SATA drivers do.)
3301 * The sd-journal API gained two new calls
3302 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
3303 that report whether log data from /run or /var has been found.
3305 * journalctl gained a new switch "--fields" that prints all journal
3306 record field names currently in use in the journal. This is backed
3307 by two new sd-journal API calls sd_journal_enumerate_fields() and
3308 sd_journal_restart_fields().
3310 * Most configurable timeouts in systemd now expect an argument of
3311 "infinity" to turn them off, instead of "0" as before. The semantics
3312 from now on is that a timeout of "0" means "now", and "infinity"
3313 means "never". To maintain backwards compatibility, "0" continues to
3314 turn off previously existing timeout settings.
3316 * "systemctl reload-or-try-restart" has been renamed to "systemctl
3317 try-reload-or-restart" to clarify what it actually does: the "try"
3318 logic applies to both reloading and restarting, not just restarting.
3319 The old name continues to be accepted for compatibility.
3321 * On boot-up, when PID 1 detects that the system clock is behind the
3322 release date of the systemd version in use, the clock is now set
3323 to the latter. Previously, this was already done in timesyncd, in order
3324 to avoid running with clocks set to the various clock epochs such as
3325 1902, 1938 or 1970. With this change the logic is now done in PID 1
3326 in addition to timesyncd during early boot-up, so that it is enforced
3327 before the first process is spawned by systemd. Note that the logic
3328 in timesyncd remains, as it is more comprehensive and ensures
3329 clock monotonicity by maintaining a persistent timestamp file in
3330 /var. Since /var is generally not available in earliest boot or the
3331 initrd, this part of the logic remains in timesyncd, and is not done
3334 * Support for tweaking details in net_cls.class_id through the
3335 NetClass= configuration directive has been removed, as the kernel
3336 people have decided to deprecate that controller in cgroup v2.
3337 Userspace tools such as nftables are moving over to setting rules
3338 that are specific to the full cgroup path of a task, which obsoletes
3339 these controllers anyway. The NetClass= directive is kept around for
3340 legacy compatibility reasons. For a more in-depth description of the
3341 kernel change, please refer to the respective upstream commit:
3343 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
3345 * A new service setting RuntimeMaxSec= has been added that may be used
3346 to specify a maximum runtime for a service. If the timeout is hit, the
3347 service is terminated and put into a failure state.
3349 * A new service setting AmbientCapabilities= has been added. It allows
3350 configuration of additional Linux process capabilities that are
3351 passed to the activated processes. This is only available on very
3354 * The process resource limit settings in service units may now be used
3355 to configure hard and soft limits individually.
3357 * The various libsystemd APIs such as sd-bus or sd-event now publicly
3358 expose support for gcc's __attribute__((cleanup())) C extension.
3359 Specifically, for many object destructor functions alternative
3360 versions have been added that have names suffixed with "p" and take a
3361 pointer to a pointer to the object to destroy, instead of just a
3362 pointer to the object itself. This is useful because these destructor
3363 functions may be used directly as parameters to the cleanup
3364 construct. Internally, systemd has been a heavy user of this GCC
3365 extension for a long time, and with this change similar support is
3366 now available to consumers of the library outside of systemd. Note
3367 that by using this extension in your sources compatibility with old
3368 and strictly ANSI compatible C compilers is lost. However, all gcc or
3369 LLVM versions of recent years support this extension.
3371 * Timer units gained support for a new setting RandomizedDelaySec= that
3372 allows configuring some additional randomized delay to the configured
3373 time. This is useful to spread out timer events to avoid load peaks in
3374 clusters or larger setups.
3376 * Calendar time specifications now support sub-second accuracy.
3378 * Socket units now support listening on SCTP and UDP-lite protocol
3381 * The sd-event API now comes with a full set of man pages.
3383 * Older versions of systemd contained experimental support for
3384 compressing journal files and coredumps with the LZ4 compressor that
3385 was not compatible with the lz4 binary (due to API limitations of the
3386 lz4 library). This support has been removed; only support for files
3387 compatible with the lz4 binary remains. This LZ4 logic is now
3388 officially supported and no longer considered experimental.
3390 * The dkr image import logic has been removed again from importd. dkr's
3391 micro-services focus doesn't fit into the machine image focus of
3392 importd, and quickly got out of date with the upstream dkr API.
3394 * Creation of the /run/lock/lockdev/ directory was dropped from
3395 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
3396 been available for many years. If you still need this, you need to
3397 create your own tmpfiles.d config file with:
3399 d /run/lock/lockdev 0775 root lock -
3401 * The settings StartLimitBurst=, StartLimitInterval=, StartLimitAction=
3402 and RebootArgument= have been moved from the [Service] section of
3403 unit files to [Unit], and they are now supported on all unit types,
3404 not just service units. Of course, systemd will continue to
3405 understand these settings also at the old location, in order to
3406 maintain compatibility.
3408 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
3409 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
3410 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
3411 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
3412 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
3413 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
3414 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
3415 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
3416 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
3417 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
3418 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
3419 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
3420 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
3421 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
3422 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
3423 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
3424 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
3425 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
3426 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
3428 — Berlin, 2016-02-11
3432 * A number of properties previously only settable in unit
3433 files are now also available as properties to set when
3434 creating transient units programmatically via the bus, as it
3435 is exposed with systemd-run's --property=
3436 setting. Specifically, these are: SyslogIdentifier=,
3437 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
3438 EnvironmentFile=, ReadWriteDirectories=,
3439 ReadOnlyDirectories=, InaccessibleDirectories=,
3440 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
3442 * When creating transient services via the bus API it is now
3443 possible to pass in a set of file descriptors to use as
3444 STDIN/STDOUT/STDERR for the invoked process.
3446 * Slice units may now be created transiently via the bus APIs,
3447 similar to the way service and scope units may already be
3448 created transiently.
3450 * Wherever systemd expects a calendar timestamp specification
3451 (like in journalctl's --since= and --until= switches) UTC
3452 timestamps are now supported. Timestamps suffixed with "UTC"
3453 are now considered to be in Universal Time Coordinated
3454 instead of the local timezone. Also, timestamps may now
3455 optionally be specified with sub-second accuracy. Both of
3456 these additions also apply to recurring calendar event
3457 specification, such as OnCalendar= in timer units.
3459 * journalctl gained a new "--sync" switch that asks the
3460 journal daemon to write all so far unwritten log messages to
3461 disk and sync the files, before returning.
3463 * systemd-tmpfiles learned two new line types "q" and "Q" that
3464 operate like "v", but also set up a basic btrfs quota
3465 hierarchy when used on a btrfs file system with quota
3468 * tmpfiles' "v", "q" and "Q" will now create a plain directory
3469 instead of a subvolume (even on a btrfs file system) if the
3470 root directory is a plain directory, and not a
3471 subvolume. This should simplify things with certain chroot()
3472 environments which are not aware of the concept of btrfs
3475 * systemd-detect-virt gained a new --chroot switch to detect
3476 whether execution takes place in a chroot() environment.
3478 * CPUAffinity= now takes CPU index ranges in addition to
3481 * The various memory-related resource limit settings (such as
3482 LimitAS=) now understand the usual K, M, G, ... suffixes to
3483 the base of 1024 (IEC). Similar, the time-related resource
3484 limit settings understand the usual min, h, day, ...
3487 * There's a new system.conf setting DefaultTasksMax= to
3488 control the default TasksMax= setting for services and
3489 scopes running on the system. (TasksMax= is the primary
3490 setting that exposes the "pids" cgroup controller on systemd
3491 and was introduced in the previous systemd release.) The
3492 setting now defaults to 512, which means services that are
3493 not explicitly configured otherwise will only be able to
3494 create 512 processes or threads at maximum, from this
3495 version on. Note that this means that thread- or
3496 process-heavy services might need to be reconfigured to set
3497 TasksMax= to a higher value. It is sufficient to set
3498 TasksMax= in these specific unit files to a higher value, or
3499 even "infinity". Similar, there's now a logind.conf setting
3500 UserTasksMax= that defaults to 4096 and limits the total
3501 number of processes or tasks each user may own
3502 concurrently. nspawn containers also have the TasksMax=
3503 value set by default now, to 8192. Note that all of this
3504 only has an effect if the "pids" cgroup controller is
3505 enabled in the kernel. The general benefit of these changes
3506 should be a more robust and safer system, that provides a
3507 certain amount of per-service fork() bomb protection.
3509 * systemd-nspawn gained the new --network-veth-extra= switch
3510 to define additional and arbitrarily-named virtual Ethernet
3511 links between the host and the container.
3513 * A new service execution setting PassEnvironment= has been
3514 added that allows importing select environment variables
3515 from PID1's environment block into the environment block of
3518 * Timer units gained support for a new RemainAfterElapse=
3519 setting which takes a boolean argument. It defaults to on,
3520 exposing behaviour unchanged to previous releases. If set to
3521 off, timer units are unloaded after they elapsed if they
3522 cannot elapse again. This is particularly useful for
3523 transient timer units, which shall not stay around longer
3524 than until they first elapse.
3526 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
3527 default now (the kernel default is 16). This is beneficial
3528 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
3529 allows substantially larger numbers of queued
3530 datagrams. This should increase the capability of systemd to
3531 parallelize boot-up, as logging and sd_notify() are unlikely
3532 to stall execution anymore. If you need to change the value
3533 from the new defaults, use the usual sysctl.d/ snippets.
3535 * The compression framing format used by the journal or
3536 coredump processing has changed to be in line with what the
3537 official LZ4 tools generate. LZ4 compression support in
3538 systemd was considered unsupported previously, as the format
3539 was not compatible with the normal tools. With this release
3540 this has changed now, and it is hence safe for downstream
3541 distributions to turn it on. While not compressing as well
3542 as the XZ, LZ4 is substantially faster, which makes
3543 it a good default choice for the compression logic in the
3544 journal and in coredump handling.
3546 * Any reference to /etc/mtab has been dropped from
3547 systemd. The file has been obsolete since a while, but
3548 systemd refused to work on systems where it was incorrectly
3549 set up (it should be a symlink or non-existent). Please make
3550 sure to update to util-linux 2.27.1 or newer in conjunction
3551 with this systemd release, which also drops any reference to
3552 /etc/mtab. If you maintain a distribution make sure that no
3553 software you package still references it, as this is a
3554 likely source of bugs. There's also a glibc bug pending,
3555 asking for removal of any reference to this obsolete file:
3557 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
3559 Note that only util-linux versions built with
3560 --enable-libmount-force-mountinfo are supported.
3562 * Support for the ".snapshot" unit type has been removed. This
3563 feature turned out to be little useful and little used, and
3564 has now been removed from the core and from systemctl.
3566 * The dependency types RequiresOverridable= and
3567 RequisiteOverridable= have been removed from systemd. They
3568 have been used only very sparingly to our knowledge and
3569 other options that provide a similar effect (such as
3570 systemctl --mode=ignore-dependencies) are much more useful
3571 and commonly used. Moreover, they were only half-way
3572 implemented as the option to control behaviour regarding
3573 these dependencies was never added to systemctl. By removing
3574 these dependency types the execution engine becomes a bit
3575 simpler. Unit files that use these dependencies should be
3576 changed to use the non-Overridable dependency types
3577 instead. In fact, when parsing unit files with these
3578 options, that's what systemd will automatically convert them
3579 too, but it will also warn, asking users to fix the unit
3580 files accordingly. Removal of these dependency types should
3581 only affect a negligible number of unit files in the wild.
3583 * Behaviour of networkd's IPForward= option changed
3584 (again). It will no longer maintain a per-interface setting,
3585 but propagate one way from interfaces where this is enabled
3586 to the global kernel setting. The global setting will be
3587 enabled when requested by a network that is set up, but
3588 never be disabled again. This change was made to make sure
3589 IPv4 and IPv6 behaviour regarding packet forwarding is
3590 similar (as the Linux IPv6 stack does not support
3591 per-interface control of this setting) and to minimize
3594 * In unit files the behaviour of %u, %U, %h, %s has
3595 changed. These specifiers will now unconditionally resolve
3596 to the various user database fields of the user that the
3597 systemd instance is running as, instead of the user
3598 configured in the specific unit via User=. Note that this
3599 effectively doesn't change much, as resolving of these
3600 specifiers was already turned off in the --system instance
3601 of systemd, as we cannot do NSS lookups from PID 1. In the
3602 --user instance of systemd these specifiers where correctly
3603 resolved, but hardly made any sense, since the user instance
3604 lacks privileges to do user switches anyway, and User= is
3605 hence useless. Moreover, even in the --user instance of
3606 systemd behaviour was awkward as it would only take settings
3607 from User= assignment placed before the specifier into
3608 account. In order to unify and simplify the logic around
3609 this the specifiers will now always resolve to the
3610 credentials of the user invoking the manager (which in case
3611 of PID 1 is the root user).
3613 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
3614 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
3615 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
3616 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
3617 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
3618 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
3619 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
3620 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
3621 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
3622 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
3623 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
3624 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
3625 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
3626 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
3629 — Berlin, 2015-11-18
3633 * systemd now depends on util-linux v2.27. More specifically,
3634 the newly added mount monitor feature in libmount now
3635 replaces systemd's former own implementation.
3637 * libmount mandates /etc/mtab not to be regular file, and
3638 systemd now enforces this condition at early boot.
3639 /etc/mtab has been deprecated and warned about for a very
3640 long time, so systems running systemd should already have
3641 stopped having this file around as anything else than a
3642 symlink to /proc/self/mounts.
3644 * Support for the "pids" cgroup controller has been added. It
3645 allows accounting the number of tasks in a cgroup and
3646 enforcing limits on it. This adds two new setting
3647 TasksAccounting= and TasksMax= to each unit, as well as a
3648 global option DefaultTasksAccounting=.
3650 * Support for the "net_cls" cgroup controller has been added.
3651 It allows assigning a net class ID to each task in the
3652 cgroup, which can then be used in firewall rules and traffic
3653 shaping configurations. Note that the kernel netfilter net
3654 class code does not currently work reliably for ingress
3655 packets on unestablished sockets.
3657 This adds a new config directive called NetClass= to CGroup
3658 enabled units. Allowed values are positive numbers for fixed
3659 assignments and "auto" for picking a free value
3662 * 'systemctl is-system-running' now returns 'offline' if the
3663 system is not booted with systemd. This command can now be
3664 used as a substitute for 'systemd-notify --booted'.
3666 * Watchdog timeouts have been increased to 3 minutes for all
3667 in-tree service files. Apparently, disk IO issues are more
3668 frequent than we hoped, and user reported >1 minute waiting
3671 * 'machine-id-commit' functionality has been merged into
3672 'machine-id-setup --commit'. The separate binary has been
3675 * The WorkingDirectory= directive in unit files may now be set
3676 to the special value '~'. In this case, the working
3677 directory is set to the home directory of the user
3678 configured in User=.
3680 * "machinectl shell" will now open the shell in the home
3681 directory of the selected user by default.
3683 * The CrashChVT= configuration file setting is renamed to
3684 CrashChangeVT=, following our usual logic of not
3685 abbreviating unnecessarily. The old directive is still
3686 supported for compat reasons. Also, this directive now takes
3687 an integer value between 1 and 63, or a boolean value. The
3688 formerly supported '-1' value for disabling stays around for
3691 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
3692 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
3693 RootDirectory= properties can now be set for transient
3696 * The systemd-analyze tool gained a new "set-log-target" verb
3697 to change the logging target the system manager logs to
3698 dynamically during runtime. This is similar to how
3699 "systemd-analyze set-log-level" already changes the log
3702 * In nspawn /sys is now mounted as tmpfs, with only a selected
3703 set of subdirectories mounted in from the real sysfs. This
3704 enhances security slightly, and is useful for ensuring user
3705 namespaces work correctly.
3707 * Support for USB FunctionFS activation has been added. This
3708 allows implementation of USB gadget services that are
3709 activated as soon as they are requested, so that they don't
3710 have to run continuously, similar to classic socket
3713 * The "systemctl exit" command now optionally takes an
3714 additional parameter that sets the exit code to return from
3715 the systemd manager when exiting. This is only relevant when
3716 running the systemd user instance, or when running the
3717 system instance in a container.
3719 * sd-bus gained the new API calls sd_bus_path_encode_many()
3720 and sd_bus_path_decode_many() that allow easy encoding and
3721 decoding of multiple identifier strings inside a D-Bus
3722 object path. Another new call sd_bus_default_flush_close()
3723 has been added to flush and close per-thread default
3726 * systemd-cgtop gained support for a -M/--machine= switch to
3727 show the control groups within a certain container only.
3729 * "systemctl kill" gained support for an optional --fail
3730 switch. If specified the requested operation will fail of no
3731 processes have been killed, because the unit had no
3732 processes attached, or similar.
3734 * A new systemd.crash_reboot=1 kernel command line option has
3735 been added that triggers a reboot after crashing. This can
3736 also be set through CrashReboot= in systemd.conf.
3738 * The RuntimeDirectory= setting now understands unit
3739 specifiers like %i or %f.
3741 * A new (still internal) library API sd-ipv4acd has been added,
3742 that implements address conflict detection for IPv4. It's
3743 based on code from sd-ipv4ll, and will be useful for
3744 detecting DHCP address conflicts.
3746 * File descriptors passed during socket activation may now be
3747 named. A new API sd_listen_fds_with_names() is added to
3748 access the names. The default names may be overridden,
3749 either in the .socket file using the FileDescriptorName=
3750 parameter, or by passing FDNAME= when storing the file
3751 descriptors using sd_notify().
3753 * systemd-networkd gained support for:
3755 - Setting the IPv6 Router Advertisement settings via
3756 IPv6AcceptRouterAdvertisements= in .network files.
3758 - Configuring the HelloTimeSec=, MaxAgeSec= and
3759 ForwardDelaySec= bridge parameters in .netdev files.
3761 - Configuring PreferredSource= for static routes in
3764 * The "ask-password" framework used to query for LUKS harddisk
3765 passwords or SSL passwords during boot gained support for
3766 caching passwords in the kernel keyring, if it is
3767 available. This makes sure that the user only has to type in
3768 a passphrase once if there are multiple objects to unlock
3769 with the same one. Previously, such password caching was
3770 available only when Plymouth was used; this moves the
3771 caching logic into the systemd codebase itself. The
3772 "systemd-ask-password" utility gained a new --keyname=
3773 switch to control which kernel keyring key to use for
3774 caching a password in. This functionality is also useful for
3775 enabling display managers such as gdm to automatically
3776 unlock the user's GNOME keyring if its passphrase, the
3777 user's password and the harddisk password are the same, if
3778 gdm-autologin is used.
3780 * When downloading tar or raw images using "machinectl
3781 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
3782 file is now also downloaded, if it is available and stored
3783 next to the image file.
3785 * Units of type ".socket" gained a new boolean setting
3786 Writable= which is only useful in conjunction with
3787 ListenSpecial=. If true, enables opening the specified
3788 special file in O_RDWR mode rather than O_RDONLY mode.
3790 * systemd-rfkill has been reworked to become a singleton
3791 service that is activated through /dev/rfkill on each rfkill
3792 state change and saves the settings to disk. This way,
3793 systemd-rfkill is now compatible with devices that exist
3794 only intermittendly, and even restores state if the previous
3795 system shutdown was abrupt rather than clean.
3797 * The journal daemon gained support for vacuuming old journal
3798 files controlled by the number of files that shall remain,
3799 in addition to the already existing control by size and by
3800 date. This is useful as journal interleaving performance
3801 degrades with too many separate journal files, and allows
3802 putting an effective limit on them. The new setting defaults
3803 to 100, but this may be changed by setting SystemMaxFiles=
3804 and RuntimeMaxFiles= in journald.conf. Also, the
3805 "journalctl" tool gained the new --vacuum-files= switch to
3806 manually vacuum journal files to leave only the specified
3807 number of files in place.
3809 * udev will now create /dev/disk/by-path links for ATA devices
3810 on kernels where that is supported.
3812 * Galician, Serbian, Turkish and Korean translations were added.
3814 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
3815 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
3816 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
3817 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
3818 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
3819 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
3820 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
3821 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
3822 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
3823 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
3824 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
3825 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
3826 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
3827 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
3828 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
3829 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
3830 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
3831 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
3833 — Berlin, 2015-10-07
3837 * The DHCP implementation of systemd-networkd gained a set of
3840 - The DHCP server now supports emitting DNS and NTP
3841 information. It may be enabled and configured via
3842 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
3843 and NTP information is enabled, but no servers are
3844 configured, the corresponding uplink information (if there
3845 is any) is propagated.
3847 - Server and client now support transmission and reception
3848 of timezone information. It can be configured via the
3849 newly introduced network options UseTimezone=,
3850 EmitTimezone=, and Timezone=. Transmission of timezone
3851 information is enabled between host and containers by
3852 default now: the container will change its local timezone
3853 to what the host has set.
3855 - Lease timeouts can now be configured via
3856 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
3858 - The DHCP server improved on the stability of
3859 leases. Clients are more likely to get the same lease
3860 information back, even if the server loses state.
3862 - The DHCP server supports two new configuration options to
3863 control the lease address pool metrics, PoolOffset= and
3866 * The encapsulation limit of tunnels in systemd-networkd may
3867 now be configured via 'EncapsulationLimit='. It allows
3868 modifying the maximum additional levels of encapsulation
3869 that are permitted to be prepended to a packet.
3871 * systemd now supports the concept of user buses replacing
3872 session buses, if used with dbus-1.10 (and enabled via dbus
3873 --enable-user-session). It previously only supported this on
3874 kdbus-enabled systems, and this release expands this to
3875 'dbus-daemon' systems.
3877 * systemd-networkd now supports predictable interface names
3880 * systemd now optionally supports the new Linux kernel
3881 "unified" control group hierarchy. If enabled via the kernel
3882 command-line option 'systemd.unified_cgroup_hierarchy=1',
3883 systemd will try to mount the unified cgroup hierarchy
3884 directly on /sys/fs/cgroup. If not enabled, or not
3885 available, systemd will fall back to the legacy cgroup
3886 hierarchy setup, as before. Host system and containers can
3887 mix and match legacy and unified hierarchies as they
3888 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
3889 environment variable to individually select the hierarchy to
3890 use for executed containers. By default, nspawn will use the
3891 unified hierarchy for the containers if the host uses the
3892 unified hierarchy, and the legacy hierarchy otherwise.
3893 Please note that at this point the unified hierarchy is an
3894 experimental kernel feature and is likely to change in one
3895 of the next kernel releases. Therefore, it should not be
3896 enabled by default in downstream distributions yet. The
3897 minimum required kernel version for the unified hierarchy to
3898 work is 4.2. Note that when the unified hierarchy is used
3899 for the first time delegated access to controllers is
3900 safe. Because of this systemd-nspawn containers will get
3901 access to controllers now, as will systemd user
3902 sessions. This means containers and user sessions may now
3903 manage their own resources, partitioning up what the system
3906 * A new special scope unit "init.scope" has been introduced
3907 that encapsulates PID 1 of the system. It may be used to
3908 determine resource usage and enforce resource limits on PID
3909 1 itself. PID 1 hence moved out of the root of the control
3912 * The cgtop tool gained support for filtering out kernel
3913 threads when counting tasks in a control group. Also, the
3914 count of processes is now recursively summed up by
3915 default. Two options -k and --recursive= have been added to
3916 revert to old behaviour. The tool has also been updated to
3917 work correctly in containers now.
3919 * systemd-nspawn's --bind= and --bind-ro= options have been
3920 extended to allow creation of non-recursive bind mounts.
3922 * libsystemd gained two new calls sd_pid_get_cgroup() and
3923 sd_peer_get_cgroup() which return the control group path of
3924 a process or peer of a connected AF_UNIX socket. This
3925 function call is particularly useful when implementing
3926 delegated subtrees support in the control group hierarchy.
3928 * The "sd-event" event loop API of libsystemd now supports
3929 correct dequeuing of real-time signals, without losing
3932 * When systemd requests a polkit decision when managing units it
3933 will now add additional fields to the request, including unit
3934 name and desired operation. This enables more powerful polkit
3935 policies, that make decisions depending on these parameters.
3937 * nspawn learnt support for .nspawn settings files, that may
3938 accompany the image files or directories of containers, and
3939 may contain additional settings for the container. This is
3940 an alternative to configuring container parameters via the
3941 nspawn command line.
3943 Contributions from: Cristian Rodríguez, Daniel Mack, David
3944 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
3945 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
3946 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
3947 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
3948 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
3949 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
3950 Andersen, Tom Gundersen, Torstein Husebø
3952 — Berlin, 2015-09-08
3956 * machinectl gained a new verb 'shell' which opens a fresh
3957 shell on the target container or the host. It is similar to
3958 the existing 'login' command of machinectl, but spawns the
3959 shell directly without prompting for username or
3960 password. The pseudo machine '.host' now refers to the local
3961 host and is used by default. Hence, 'machinectl shell' can
3962 be used as replacement for 'su -' which spawns a session as
3963 a fresh systemd unit in a way that is fully isolated from
3964 the originating session.
3966 * systemd-networkd learned to cope with private-zone DHCP
3967 options and allows other programs to query the values.
3969 * SELinux access control when enabling/disabling units is no
3970 longer enforced with this release. The previous implementation
3971 was incorrect, and a new corrected implementation is not yet
3972 available. As unit file operations are still protected via
3973 polkit and D-Bus policy this is not a security problem. Yet,
3974 distributions which care about optimal SELinux support should
3975 probably not stabilize on this release.
3977 * sd-bus gained support for matches of type "arg0has=", that
3978 test for membership of strings in string arrays sent in bus
3981 * systemd-resolved now dumps the contents of its DNS and LLMNR
3982 caches to the logs on reception of the SIGUSR1 signal. This
3983 is useful to debug DNS behaviour.
3985 * The coredumpctl tool gained a new --directory= option to
3986 operate on journal files in a specific directory.
3988 * "systemctl reboot" and related commands gained a new
3989 "--message=" option which may be used to set a free-text
3990 wall message when shutting down or rebooting the
3991 system. This message is also logged, which is useful for
3992 figuring out the reason for a reboot or shutdown a
3995 * The "systemd-resolve-host" tool's -i switch now takes
3996 network interface numbers as alternative to interface names.
3998 * A new unit file setting for services has been introduced:
3999 UtmpMode= allows configuration of how precisely systemd
4000 handles utmp and wtmp entries for the service if this is
4001 enabled. This allows writing services that appear similar to
4002 user sessions in the output of the "w", "who", "last" and
4005 * systemd-resolved will now locally synthesize DNS resource
4006 records for the "localhost" and "gateway" domains as well as
4007 the local hostname. This should ensure that clients querying
4008 RRs via resolved will get similar results as those going via
4009 NSS, if nss-myhostname is enabled.
4011 Contributions from: Alastair Hughes, Alex Crawford, Daniel
4012 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
4013 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
4014 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
4015 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
4016 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
4017 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
4018 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
4019 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
4020 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
4021 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
4022 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
4024 — Berlin, 2015-08-27
4028 * The systemd-efi-boot-generator functionality was merged into
4029 systemd-gpt-auto-generator.
4031 * systemd-networkd now supports Group Policy for vxlan
4032 devices. It can be enabled via the new boolean configuration
4033 option called 'GroupPolicyExtension='.
4035 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
4036 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
4037 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
4039 — Berlin, 2015-07-31
4043 * The python-systemd code has been removed from the systemd repository.
4044 A new repository has been created which accommodates the code from
4045 now on, and we kindly ask distributions to create a separate package
4046 for this: https://github.com/systemd/python-systemd
4048 * The systemd daemon will now reload its main configuration
4049 (/etc/systemd/system.conf) on daemon-reload.
4051 * sd-dhcp now exposes vendor specific extensions via
4052 sd_dhcp_lease_get_vendor_specific().
4054 * systemd-networkd gained a number of new configuration options.
4056 - A new boolean configuration option for TAP devices called
4057 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
4058 device, thus allowing to send and receive GSO packets.
4060 - A new tunnel configuration option called 'CopyDSCP='.
4061 If enabled, the DSCP field of ip6 tunnels is copied into the
4062 decapsulated packet.
4064 - A set of boolean bridge configuration options were added.
4065 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
4066 and 'UnicastFlood=' are now parsed by networkd and applied to the
4067 respective bridge link device via the respective IFLA_BRPORT_*
4070 - A new string configuration option to override the hostname sent
4071 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
4072 is true, networkd will use the configured hostname instead of the
4073 system hostname when sending DHCP requests.
4075 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
4076 networkd will configure the IPv6 flow-label of the tunnel device
4077 according to RFC2460.
4079 - The 'macvtap' virtual network devices are now supported, similar to
4080 the already supported 'macvlan' devices.
4082 * systemd-resolved now implements RFC5452 to improve resilience against
4083 cache poisoning. Additionally, source port randomization is enabled
4084 by default to further protect against DNS spoofing attacks.
4086 * nss-mymachines now supports translating UIDs and GIDs of running
4087 containers with user-namespaces enabled. If a container 'foo'
4088 translates a host uid 'UID' to the container uid 'TUID', then
4089 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
4090 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
4091 mapped as 'vg-foo-TGID'.
4093 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
4094 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
4095 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
4096 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
4097 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
4098 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
4099 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
4100 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
4101 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
4102 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
4104 — Berlin, 2015-07-29
4108 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
4109 There are no known issues with current sysfs, and udev does not need
4110 or should be used to work around such bugs.
4112 * udev does no longer enable USB HID power management. Several reports
4113 indicate, that some devices cannot handle that setting.
4115 * The udev accelerometer helper was removed. The functionality
4116 is now fully included in iio-sensor-proxy. But this means,
4117 older iio-sensor-proxy versions will no longer provide
4118 accelerometer/orientation data with this systemd version.
4119 Please upgrade iio-sensor-proxy to version 1.0.
4121 * networkd gained a new configuration option IPv6PrivacyExtensions=
4122 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
4123 for Stateless Address") on selected networks.
4125 * For the sake of fewer build-time dependencies and less code in the
4126 main repository, the python bindings are about to be removed in the
4127 next release. A new repository has been created which accommodates
4128 the code from now on, and we kindly ask distributions to create a
4129 separate package for this. The removal will take place in v223.
4131 https://github.com/systemd/python-systemd
4133 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
4134 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
4135 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
4136 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
4137 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
4138 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
4139 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
4140 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
4141 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
4142 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
4144 — Berlin, 2015-07-07
4148 * The sd-bus.h and sd-event.h APIs have now been declared
4149 stable and have been added to the official interface of
4150 libsystemd.so. sd-bus implements an alternative D-Bus client
4151 library, that is relatively easy to use, very efficient and
4152 supports both classic D-Bus as well as kdbus as transport
4153 backend. sd-event is a generic event loop abstraction that
4154 is built around Linux epoll, but adds features such as event
4155 prioritization or efficient timer handling. Both APIs are good
4156 choices for C programs looking for a bus and/or event loop
4157 implementation that is minimal and does not have to be
4158 portable to other kernels.
4160 * kdbus support is no longer compile-time optional. It is now
4161 always built-in. However, it can still be disabled at
4162 runtime using the kdbus=0 kernel command line setting, and
4163 that setting may be changed to default to off, by specifying
4164 --disable-kdbus at build-time. Note though that the kernel
4165 command line setting has no effect if the kdbus.ko kernel
4166 module is not installed, in which case kdbus is (obviously)
4167 also disabled. We encourage all downstream distributions to
4168 begin testing kdbus by adding it to the kernel images in the
4169 development distributions, and leaving kdbus support in
4172 * The minimal required util-linux version has been bumped to
4175 * Support for chkconfig (--enable-chkconfig) was removed in
4176 favor of calling an abstraction tool
4177 /lib/systemd/systemd-sysv-install. This needs to be
4178 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
4179 in README for details.
4181 * If there's a systemd unit and a SysV init script for the
4182 same service name, and the user executes "systemctl enable"
4183 for it (or a related call), then this will now enable both
4184 (or execute the related operation on both), not just the
4187 * The libudev API documentation has been converted from gtkdoc
4190 * gudev has been removed from the systemd tree, it is now an
4193 * The systemd-cgtop tool learnt a new --raw switch to generate
4194 "raw" (machine parsable) output.
4196 * networkd's IPForwarding= .network file setting learnt the
4197 new setting "kernel", which ensures that networkd does not
4198 change the IP forwarding sysctl from the default kernel
4201 * The systemd-logind bus API now exposes a new boolean
4202 property "Docked" that reports whether logind considers the
4203 system "docked", i.e. connected to a docking station or not.
4205 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
4206 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
4207 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
4208 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
4209 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
4210 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
4211 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
4212 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
4213 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
4214 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
4215 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
4216 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
4217 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
4218 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
4219 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
4220 Fink, Zbigniew Jędrzejewski-Szmek
4222 — Berlin, 2015-06-19
4226 * The gudev library has been extracted into a separate repository
4227 available at: https://git.gnome.org/browse/libgudev/
4228 It is now managed as part of the Gnome project. Distributions
4229 are recommended to pass --disable-gudev to systemd and use
4230 gudev from the Gnome project instead. gudev is still included
4231 in systemd, for now. It will be removed soon, though. Please
4232 also see the announcement-thread on systemd-devel:
4233 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
4235 * systemd now exposes a CPUUsageNSec= property for each
4236 service unit on the bus, that contains the overall consumed
4237 CPU time of a service (the sum of what each process of the
4238 service consumed). This value is only available if
4239 CPUAccounting= is turned on for a service, and is then shown
4240 in the "systemctl status" output.
4242 * Support for configuring alternative mappings of the old SysV
4243 runlevels to systemd targets has been removed. They are now
4244 hardcoded in a way that runlevels 2, 3, 4 all map to
4245 multi-user.target and 5 to graphical.target (which
4246 previously was already the default behaviour).
4248 * The auto-mounter logic gained support for mount point
4249 expiry, using a new TimeoutIdleSec= setting in .automount
4250 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
4252 * The EFI System Partition (ESP) as mounted to /boot by
4253 systemd-efi-boot-generator will now be unmounted
4254 automatically after 2 minutes of not being used. This should
4255 minimize the risk of ESP corruptions.
4257 * New /etc/fstab options x-systemd.requires= and
4258 x-systemd.requires-mounts-for= are now supported to express
4259 additional dependencies for mounts. This is useful for
4260 journalling file systems that support external journal
4261 devices or overlay file systems that require underlying file
4262 systems to be mounted.
4264 * systemd does not support direct live-upgrades (via systemctl
4265 daemon-reexec) from versions older than v44 anymore. As no
4266 distribution we are aware of shipped such old versions in a
4267 stable release this should not be problematic.
4269 * When systemd forks off a new per-connection service instance
4270 it will now set the $REMOTE_ADDR environment variable to the
4271 remote IP address, and $REMOTE_PORT environment variable to
4272 the remote IP port. This behaviour is similar to the
4273 corresponding environment variables defined by CGI.
4275 * systemd-networkd gained support for uplink failure
4276 detection. The BindCarrier= option allows binding interface
4277 configuration dynamically to the link sense of other
4278 interfaces. This is useful to achieve behaviour like in
4281 * systemd-networkd gained support for configuring the DHCP
4282 client identifier to use when requesting leases.
4284 * systemd-networkd now has a per-network UseNTP= option to
4285 configure whether NTP server information acquired via DHCP
4286 is passed on to services like systemd-timesyncd.
4288 * systemd-networkd gained support for vti6 tunnels.
4290 * Note that systemd-networkd manages the sysctl variable
4291 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
4292 it is configured for since v219. The variable controls IP
4293 forwarding, and is a per-interface alternative to the global
4294 /proc/sys/net/ipv[46]/ip_forward. This setting is
4295 configurable in the IPForward= option, which defaults to
4296 "no". This means if networkd is used for an interface it is
4297 no longer sufficient to set the global sysctl option to turn
4298 on IP forwarding! Instead, the .network file option
4299 IPForward= needs to be turned on! Note that the
4300 implementation of this behaviour was broken in v219 and has
4303 * Many bonding and vxlan options are now configurable in
4306 * systemd-nspawn gained a new --property= setting to set unit
4307 properties for the container scope. This is useful for
4308 setting resource parameters (e.g. "CPUShares=500") on
4309 containers started from the command line.
4311 * systemd-nspawn gained a new --private-users= switch to make
4312 use of user namespacing available on recent Linux kernels.
4314 * systemd-nspawn may now be called as part of a shell pipeline
4315 in which case the pipes used for stdin and stdout are passed
4316 directly to the process invoked in the container, without
4317 indirection via a pseudo tty.
4319 * systemd-nspawn gained a new switch to control the UNIX
4320 signal to use when killing the init process of the container
4323 * systemd-nspawn gained a new --overlay= switch for mounting
4324 overlay file systems into the container using the new kernel
4327 * When a container image is imported via systemd-importd and
4328 the host file system is not btrfs, a loopback block device
4329 file is created in /var/lib/machines.raw with a btrfs file
4330 system inside. It is then mounted to /var/lib/machines to
4331 enable btrfs features for container management. The loopback
4332 file and btrfs file system is grown as needed when container
4333 images are imported via systemd-importd.
4335 * systemd-machined/systemd-importd gained support for btrfs
4336 quota, to enforce container disk space limits on disk. This
4337 is exposed in "machinectl set-limit".
4339 * systemd-importd now can import containers from local .tar,
4340 .raw and .qcow2 images, and export them to .tar and .raw. It
4341 can also import dkr v2 images now from the network (on top
4344 * systemd-importd gained support for verifying downloaded
4345 images with gpg2 (previously only gpg1 was supported).
4347 * systemd-machined, systemd-logind, systemd: most bus calls are
4348 now accessible to unprivileged processes via polkit. Also,
4349 systemd-logind will now allow users to kill their own sessions
4350 without further privileges or authorization.
4352 * systemd-shutdownd has been removed. This service was
4353 previously responsible for implementing scheduled shutdowns
4354 as exposed in /usr/bin/shutdown's time parameter. This
4355 functionality has now been moved into systemd-logind and is
4356 accessible via a bus interface.
4358 * "systemctl reboot" gained a new switch --firmware-setup that
4359 can be used to reboot into the EFI firmware setup, if that
4360 is available. systemd-logind now exposes an API on the bus
4361 to trigger such reboots, in case graphical desktop UIs want
4362 to cover this functionality.
4364 * "systemctl enable", "systemctl disable" and "systemctl mask"
4365 now support a new "--now" switch. If specified the units
4366 that are enabled will also be started, and the ones
4367 disabled/masked also stopped.
4369 * The Gummiboot EFI boot loader tool has been merged into
4370 systemd, and renamed to "systemd-boot". The bootctl tool has been
4371 updated to support systemd-boot.
4373 * An EFI kernel stub has been added that may be used to create
4374 kernel EFI binaries that contain not only the actual kernel,
4375 but also an initrd, boot splash, command line and OS release
4376 information. This combined binary can then be signed as a
4377 single image, so that the firmware can verify it all in one
4378 step. systemd-boot has special support for EFI binaries created
4379 like this and can extract OS release information from them
4380 and show them in the boot menu. This functionality is useful
4381 to implement cryptographically verified boot schemes.
4383 * Optional support has been added to systemd-fsck to pass
4384 fsck's progress report to an AF_UNIX socket in the file
4387 * udev will no longer create device symlinks for all block
4388 devices by default. A blacklist for excluding special block
4389 devices from this logic has been turned into a whitelist
4390 that requires picking block devices explicitly that require
4393 * A new (currently still internal) API sd-device.h has been
4394 added to libsystemd. This modernized API is supposed to
4395 replace libudev eventually. In fact, already much of libudev
4396 is now just a wrapper around sd-device.h.
4398 * A new hwdb database for storing metadata about pointing
4399 stick devices has been added.
4401 * systemd-tmpfiles gained support for setting file attributes
4402 similar to the "chattr" tool with new 'h' and 'H' lines.
4404 * systemd-journald will no longer unconditionally set the
4405 btrfs NOCOW flag on new journal files. This is instead done
4406 with tmpfiles snippet using the new 'h' line type. This
4407 allows easy disabling of this logic, by masking the
4408 journal-nocow.conf tmpfiles file.
4410 * systemd-journald will now translate audit message types to
4411 human readable identifiers when writing them to the
4412 journal. This should improve readability of audit messages.
4414 * The LUKS logic gained support for the offset= and skip=
4415 options in /etc/crypttab, as previously implemented by
4418 * /usr/lib/os-release gained a new optional field VARIANT= for
4419 distributions that support multiple variants (such as a
4420 desktop edition, a server edition, ...)
4422 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
4423 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
4424 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
4425 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
4426 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
4427 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
4428 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
4429 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
4430 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
4431 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
4432 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
4433 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
4434 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
4435 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
4436 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
4437 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
4438 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
4439 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
4440 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
4441 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
4442 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
4443 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
4444 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
4445 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
4446 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
4447 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
4448 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
4450 — Berlin, 2015-05-22
4454 * Introduce a new API "sd-hwdb.h" for querying the hardware
4455 metadata database. With this minimal interface one can query
4456 and enumerate the udev hwdb, decoupled from the old libudev
4457 library. libudev's interface for this is now only a wrapper
4458 around sd-hwdb. A new tool systemd-hwdb has been added to
4459 interface with and update the database.
4461 * When any of systemd's tools copies files (for example due to
4462 tmpfiles' C lines) a btrfs reflink will attempted first,
4463 before bytewise copying is done.
4465 * systemd-nspawn gained a new --ephemeral switch. When
4466 specified a btrfs snapshot is taken of the container's root
4467 directory, and immediately removed when the container
4468 terminates again. Thus, a container can be started whose
4469 changes never alter the container's root directory, and are
4470 lost on container termination. This switch can also be used
4471 for starting a container off the root file system of the
4472 host without affecting the host OS. This switch is only
4473 available on btrfs file systems.
4475 * systemd-nspawn gained a new --template= switch. It takes the
4476 path to a container tree to use as template for the tree
4477 specified via --directory=, should that directory be
4478 missing. This allows instantiating containers dynamically,
4479 on first run. This switch is only available on btrfs file
4482 * When a .mount unit refers to a mount point on which multiple
4483 mounts are stacked, and the .mount unit is stopped all of
4484 the stacked mount points will now be unmounted until no
4485 mount point remains.
4487 * systemd now has an explicit notion of supported and
4488 unsupported unit types. Jobs enqueued for unsupported unit
4489 types will now fail with an "unsupported" error code. More
4490 specifically .swap, .automount and .device units are not
4491 supported in containers, .busname units are not supported on
4492 non-kdbus systems. .swap and .automount are also not
4493 supported if their respective kernel compile time options
4496 * machinectl gained support for two new "copy-from" and
4497 "copy-to" commands for copying files from a running
4498 container to the host or vice versa.
4500 * machinectl gained support for a new "bind" command to bind
4501 mount host directories into local containers. This is
4502 currently only supported for nspawn containers.
4504 * networkd gained support for configuring bridge forwarding
4505 database entries (fdb) from .network files.
4507 * A new tiny daemon "systemd-importd" has been added that can
4508 download container images in tar, raw, qcow2 or dkr formats,
4509 and make them available locally in /var/lib/machines, so
4510 that they can run as nspawn containers. The daemon can GPG
4511 verify the downloads (not supported for dkr, since it has no
4512 provisions for verifying downloads). It will transparently
4513 decompress bz2, xz, gzip compressed downloads if necessary,
4514 and restore sparse files on disk. The daemon uses privilege
4515 separation to ensure the actual download logic runs with
4516 fewer privileges than the daemon itself. machinectl has
4517 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
4518 make the functionality of importd available to the
4519 user. With this in place the Fedora and Ubuntu "Cloud"
4520 images can be downloaded and booted as containers unmodified
4521 (the Fedora images lack the appropriate GPG signature files
4522 currently, so they cannot be verified, but this will change
4523 soon, hopefully). Note that downloading images is currently
4524 only fully supported on btrfs.
4526 * machinectl is now able to list container images found in
4527 /var/lib/machines, along with some metadata about sizes of
4528 disk and similar. If the directory is located on btrfs and
4529 quota is enabled, this includes quota display. A new command
4530 "image-status" has been added that shows additional
4531 information about images.
4533 * machinectl is now able to clone container images
4534 efficiently, if the underlying file system (btrfs) supports
4535 it, with the new "machinectl clone" command. It also
4536 gained commands for renaming and removing images, as well as
4537 marking them read-only or read-write (supported also on
4538 legacy file systems).
4540 * networkd gained support for collecting LLDP network
4541 announcements, from hardware that supports this. This is
4542 shown in networkctl output.
4544 * systemd-run gained support for a new -t (--pty) switch for
4545 invoking a binary on a pty whose input and output is
4546 connected to the invoking terminal. This allows executing
4547 processes as system services while interactively
4548 communicating with them via the terminal. Most interestingly
4549 this is supported across container boundaries. Invoking
4550 "systemd-run -t /bin/bash" is an alternative to running a
4551 full login session, the difference being that the former
4552 will not register a session, nor go through the PAM session
4555 * tmpfiles gained support for a new "v" line type for creating
4556 btrfs subvolumes. If the underlying file system is a legacy
4557 file system, this automatically degrades to creating a
4558 normal directory. Among others /var/lib/machines is now
4559 created like this at boot, should it be missing.
4561 * The directory /var/lib/containers/ has been deprecated and
4562 been replaced by /var/lib/machines. The term "machines" has
4563 been used in the systemd context as generic term for both
4564 VMs and containers, and hence appears more appropriate for
4565 this, as the directory can also contain raw images bootable
4568 * systemd-nspawn when invoked with -M but without --directory=
4569 or --image= is now capable of searching for the container
4570 root directory, subvolume or disk image automatically, in
4571 /var/lib/machines. systemd-nspawn@.service has been updated
4572 to make use of this, thus allowing it to be used for raw
4575 * A new machines.target unit has been introduced that is
4576 supposed to group all containers/VMs invoked as services on
4577 the system. systemd-nspawn@.service has been updated to
4578 integrate with that.
4580 * machinectl gained a new "start" command, for invoking a
4581 container as a service. "machinectl start foo" is mostly
4582 equivalent to "systemctl start systemd-nspawn@foo.service",
4583 but handles escaping in a nicer way.
4585 * systemd-nspawn will now mount most of the cgroupfs tree
4586 read-only into each container, with the exception of the
4587 container's own subtree in the name=systemd hierarchy.
4589 * journald now sets the special FS_NOCOW file flag for its
4590 journal files. This should improve performance on btrfs, by
4591 avoiding heavy fragmentation when journald's write-pattern
4592 is used on COW file systems. It degrades btrfs' data
4593 integrity guarantees for the files to the same levels as for
4594 ext3/ext4 however. This should be OK though as journald does
4595 its own data integrity checks and all its objects are
4596 checksummed on disk. Also, journald should handle btrfs disk
4597 full events a lot more gracefully now, by processing SIGBUS
4598 errors, and not relying on fallocate() anymore.
4600 * When journald detects that journal files it is writing to
4601 have been deleted it will immediately start new journal
4604 * systemd now provides a way to store file descriptors
4605 per-service in PID 1. This is useful for daemons to ensure
4606 that fds they require are not lost during a daemon
4607 restart. The fds are passed to the daemon on the next
4608 invocation in the same way socket activation fds are
4609 passed. This is now used by journald to ensure that the
4610 various sockets connected to all the system's stdout/stderr
4611 are not lost when journald is restarted. File descriptors
4612 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
4613 an extension to sd_notify(). Note that a limit is enforced
4614 on the number of fds a service can store in PID 1, and it
4615 defaults to 0, so that no fds may be stored, unless this is
4616 explicitly turned on.
4618 * The default TERM variable to use for units connected to a
4619 terminal, when no other value is explicitly is set is now
4620 vt220 rather than vt102. This should be fairly safe still,
4621 but allows PgUp/PgDn work.
4623 * The /etc/crypttab option header= as known from Debian is now
4626 * "loginctl user-status" and "loginctl session-status" will
4627 now show the last 10 lines of log messages of the
4628 user/session following the status output. Similar,
4629 "machinectl status" will show the last 10 log lines
4630 associated with a virtual machine or container
4631 service. (Note that this is usually not the log messages
4632 done in the VM/container itself, but simply what the
4633 container manager logs. For nspawn this includes all console
4636 * "loginctl session-status" without further argument will now
4637 show the status of the session of the caller. Similar,
4638 "lock-session", "unlock-session", "activate",
4639 "enable-linger", "disable-linger" may now be called without
4640 session/user parameter in which case they apply to the
4641 caller's session/user.
4643 * An X11 session scriptlet is now shipped that uploads
4644 $DISPLAY and $XAUTHORITY into the environment of the systemd
4645 --user daemon if a session begins. This should improve
4646 compatibility with X11 enabled applications run as systemd
4649 * Generators are now subject to masking via /etc and /run, the
4650 same way as unit files.
4652 * networkd .network files gained support for configuring
4653 per-link IPv4/IPv6 packet forwarding as well as IPv4
4654 masquerading. This is by default turned on for veth links to
4655 containers, as registered by systemd-nspawn. This means that
4656 nspawn containers run with --network-veth will now get
4657 automatic routed access to the host's networks without any
4658 further configuration or setup, as long as networkd runs on
4661 * systemd-nspawn gained the --port= (-p) switch to expose TCP
4662 or UDP posts of a container on the host. With this in place
4663 it is possible to run containers with private veth links
4664 (--network-veth), and have their functionality exposed on
4665 the host as if their services were running directly on the
4668 * systemd-nspawn's --network-veth switch now gained a short
4669 version "-n", since with the changes above it is now truly
4670 useful out-of-the-box. The systemd-nspawn@.service has been
4671 updated to make use of it too by default.
4673 * systemd-nspawn will now maintain a per-image R/W lock, to
4674 ensure that the same image is not started more than once
4675 writable. (It's OK to run an image multiple times
4676 simultaneously in read-only mode.)
4678 * systemd-nspawn's --image= option is now capable of
4679 dissecting and booting MBR and GPT disk images that contain
4680 only a single active Linux partition. Previously it
4681 supported only GPT disk images with proper GPT type
4682 IDs. This allows running cloud images from major
4683 distributions directly with systemd-nspawn, without
4686 * In addition to collecting mouse dpi data in the udev
4687 hardware database, there's now support for collecting angle
4688 information for mouse scroll wheels. The database is
4689 supposed to guarantee similar scrolling behavior on mice
4690 that it knows about. There's also support for collecting
4691 information about Touchpad types.
4693 * udev's input_id built-in will now also collect touch screen
4694 dimension data and attach it to probed devices.
4696 * /etc/os-release gained support for a Distribution Privacy
4699 * networkd gained support for creating "ipvlan", "gretap",
4700 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
4702 * systemd-tmpfiles gained support for "a" lines for setting
4705 * systemd-nspawn will now mount /tmp in the container to
4706 tmpfs, automatically.
4708 * systemd now exposes the memory.usage_in_bytes cgroup
4709 attribute and shows it for each service in the "systemctl
4710 status" output, if available.
4712 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
4713 immediate reboot is triggered. This useful if shutdown is
4714 hung and is unable to complete, to expedite the
4715 operation. Note that this kind of reboot will still unmount
4716 all file systems, and hence should not result in fsck being
4719 * A .device unit for an optical block device will now be
4720 considered active only when a medium is in the drive. Also,
4721 mount units are now bound to their backing devices thus
4722 triggering automatic unmounting when devices become
4723 unavailable. With this in place systemd will now
4724 automatically unmount left-over mounts when a CD-ROM is
4725 ejected or an USB stick is yanked from the system.
4727 * networkd-wait-online now has support for waiting for
4728 specific interfaces only (with globbing), and for giving up
4729 after a configurable timeout.
4731 * networkd now exits when idle. It will be automatically
4732 restarted as soon as interfaces show up, are removed or
4733 change state. networkd will stay around as long as there is
4734 at least one DHCP state machine or similar around, that keep
4737 * networkd may now configure IPv6 link-local addressing in
4738 addition to IPv4 link-local addressing.
4740 * The IPv6 "token" for use in SLAAC may now be configured for
4741 each .network interface in networkd.
4743 * Routes configured with networkd may now be assigned a scope
4746 * networkd's [Match] sections now support globbing and lists
4747 of multiple space-separated matches per item.
4749 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
4750 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
4751 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
4752 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
4753 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
4754 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
4755 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
4756 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
4757 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
4758 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
4759 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
4760 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
4761 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
4762 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
4763 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
4764 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
4765 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
4766 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
4767 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
4768 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
4769 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
4770 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
4771 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
4772 Hoffmann, Zbigniew Jędrzejewski-Szmek
4774 — Berlin, 2015-02-16
4778 * When querying unit file enablement status (for example via
4779 "systemctl is-enabled"), a new state "indirect" is now known
4780 which indicates that a unit might not be enabled itself, but
4781 another unit listed in its Also= setting might be.
4783 * Similar to the various existing ConditionXYZ= settings for
4784 units, there are now matching AssertXYZ= settings. While
4785 failing conditions cause a unit to be skipped, but its job
4786 to succeed, failing assertions declared like this will cause
4787 a unit start operation and its job to fail.
4789 * hostnamed now knows a new chassis type "embedded".
4791 * systemctl gained a new "edit" command. When used on a unit
4792 file, this allows extending unit files with .d/ drop-in
4793 configuration snippets or editing the full file (after
4794 copying it from /usr/lib to /etc). This will invoke the
4795 user's editor (as configured with $EDITOR), and reload the
4796 modified configuration after editing.
4798 * "systemctl status" now shows the suggested enablement state
4799 for a unit, as declared in the (usually vendor-supplied)
4800 system preset files.
4802 * nss-myhostname will now resolve the single-label host name
4803 "gateway" to the locally configured default IP routing
4804 gateways, ordered by their metrics. This assigns a stable
4805 name to the used gateways, regardless which ones are
4806 currently configured. Note that the name will only be
4807 resolved after all other name sources (if nss-myhostname is
4808 configured properly) and should hence not negatively impact
4809 systems that use the single-label host name "gateway" in
4812 * systemd-inhibit now allows filtering by mode when listing
4815 * Scope and service units gained a new "Delegate" boolean
4816 property, which, when set, allows processes running inside the
4817 unit to further partition resources. This is primarily
4818 useful for systemd user instances as well as container
4821 * journald will now pick up audit messages directly from
4822 the kernel, and log them like any other log message. The
4823 audit fields are split up and fully indexed. This means that
4824 journalctl in many ways is now a (nicer!) alternative to
4825 ausearch, the traditional audit client. Note that this
4826 implements only a minimal audit client. If you want the
4827 special audit modes like reboot-on-log-overflow, please use
4828 the traditional auditd instead, which can be used in
4829 parallel to journald.
4831 * The ConditionSecurity= unit file option now understands the
4832 special string "audit" to check whether auditing is
4835 * journalctl gained two new commands --vacuum-size= and
4836 --vacuum-time= to delete old journal files until the
4837 remaining ones take up no more than the specified size on disk,
4838 or are not older than the specified time.
4840 * A new, native PPPoE library has been added to sd-network,
4841 systemd's library of light-weight networking protocols. This
4842 library will be used in a future version of networkd to
4843 enable PPPoE communication without an external pppd daemon.
4845 * The busctl tool now understands a new "capture" verb that
4846 works similar to "monitor", but writes a packet capture
4847 trace to STDOUT that can be redirected to a file which is
4848 compatible with libcap's capture file format. This can then
4849 be loaded in Wireshark and similar tools to inspect bus
4852 * The busctl tool now understands a new "tree" verb that shows
4853 the object trees of a specific service on the bus, or of all
4856 * The busctl tool now understands a new "introspect" verb that
4857 shows all interfaces and members of objects on the bus,
4858 including their signature and values. This is particularly
4859 useful to get more information about bus objects shown by
4860 the new "busctl tree" command.
4862 * The busctl tool now understands new verbs "call",
4863 "set-property" and "get-property" for invoking bus method
4864 calls, setting and getting bus object properties in a
4867 * busctl gained a new --augment-creds= argument that controls
4868 whether the tool shall augment credential information it
4869 gets from the bus with data from /proc, in a possibly
4872 * nspawn's --link-journal= switch gained two new values
4873 "try-guest" and "try-host" that work like "guest" and
4874 "host", but do not fail if the host has no persistent
4875 journalling enabled. -j is now equivalent to
4876 --link-journal=try-guest.
4878 * macvlan network devices created by nspawn will now have
4879 stable MAC addresses.
4881 * A new SmackProcessLabel= unit setting has been added, which
4882 controls the SMACK security label processes forked off by
4883 the respective unit shall use.
4885 * If compiled with --enable-xkbcommon, systemd-localed will
4886 verify x11 keymap settings by compiling the given keymap. It
4887 will spew out warnings if the compilation fails. This
4888 requires libxkbcommon to be installed.
4890 * When a coredump is collected, a larger number of metadata
4891 fields is now collected and included in the journal records
4892 created for it. More specifically, control group membership,
4893 environment variables, memory maps, working directory,
4894 chroot directory, /proc/$PID/status, and a list of open file
4895 descriptors is now stored in the log entry.
4897 * The udev hwdb now contains DPI information for mice. For
4900 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
4902 * All systemd programs that read standalone configuration
4903 files in /etc now also support a corresponding series of
4904 .conf.d configuration directories in /etc/, /run/,
4905 /usr/local/lib/, /usr/lib/, and (if configured with
4906 --enable-split-usr) /lib/. In particular, the following
4907 configuration files now have corresponding configuration
4908 directories: system.conf user.conf, logind.conf,
4909 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
4910 resolved.conf, timesyncd.conf, journal-remote.conf, and
4911 journal-upload.conf. Note that distributions should use the
4912 configuration directories in /usr/lib/; the directories in
4913 /etc/ are reserved for the system administrator.
4915 * systemd-rfkill will no longer take the rfkill device name
4916 into account when storing rfkill state on disk, as the name
4917 might be dynamically assigned and not stable. Instead, the
4918 ID_PATH udev variable combined with the rfkill type (wlan,
4919 bluetooth, ...) is used.
4921 * A new service systemd-machine-id-commit.service has been
4922 added. When used on systems where /etc is read-only during
4923 boot, and /etc/machine-id is not initialized (but an empty
4924 file), this service will copy the temporary machine ID
4925 created as replacement into /etc after the system is fully
4926 booted up. This is useful for systems that are freshly
4927 installed with a non-initialized machine ID, but should get
4928 a fixed machine ID for subsequent boots.
4930 * networkd's .netdev files now provide a large set of
4931 configuration parameters for VXLAN devices. Similarly, the
4932 bridge port cost parameter is now configurable in .network
4933 files. There's also new support for configuring IP source
4934 routing. networkd .link files gained support for a new
4935 OriginalName= match that is useful to match against the
4936 original interface name the kernel assigned. .network files
4937 may include MTU= and MACAddress= fields for altering the MTU
4938 and MAC address while being connected to a specific network
4941 * The LUKS logic gained supported for configuring
4942 UUID-specific key files. There's also new support for naming
4943 LUKS device from the kernel command line, using the new
4944 luks.name= argument.
4946 * Timer units may now be transiently created via the bus API
4947 (this was previously already available for scope and service
4948 units). In addition it is now possible to create multiple
4949 transient units at the same time with a single bus call. The
4950 "systemd-run" tool has been updated to make use of this for
4951 running commands on a specified time, in at(1)-style.
4953 * tmpfiles gained support for "t" lines, for assigning
4954 extended attributes to files. Among other uses this may be
4955 used to assign SMACK labels to files.
4957 Contributions from: Alin Rauta, Alison Chaiken, Andrej
4958 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
4959 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
4960 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
4961 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
4962 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
4963 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
4964 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
4965 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
4966 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
4967 Michael Biebl, Michael Chapman, Michael Marineau, Michal
4968 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
4969 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
4970 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
4971 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
4972 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
4973 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
4974 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
4976 — Berlin, 2014-12-10
4980 * journalctl gained the new options -t/--identifier= to match
4981 on the syslog identifier (aka "tag"), as well as --utc to
4982 show log timestamps in the UTC timezone. journalctl now also
4983 accepts -n/--lines=all to disable line capping in a pager.
4985 * journalctl gained a new switch, --flush, that synchronously
4986 flushes logs from /run/log/journal to /var/log/journal if
4987 persistent storage is enabled. systemd-journal-flush.service
4988 now waits until the operation is complete.
4990 * Services can notify the manager before they start a reload
4991 (by sending RELOADING=1) or shutdown (by sending
4992 STOPPING=1). This allows the manager to track and show the
4993 internal state of daemons and closes a race condition when
4994 the process is still running but has closed its D-Bus
4997 * Services with Type=oneshot do not have to have any ExecStart
5000 * User units are now loaded also from
5001 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
5002 /run/systemd/user directory that was already previously
5003 supported, but is under the control of the user.
5005 * Job timeouts (i.e. timeouts on the time a job that is
5006 queued stays in the run queue) can now optionally result in
5007 immediate reboot or power-off actions (JobTimeoutAction= and
5008 JobTimeoutRebootArgument=). This is useful on ".target"
5009 units, to limit the maximum time a target remains
5010 undispatched in the run queue, and to trigger an emergency
5011 operation in such a case. This is now used by default to
5012 turn off the system if boot-up (as defined by everything in
5013 basic.target) hangs and does not complete for at least
5014 15min. Also, if power-off or reboot hang for at least 30min
5015 an immediate power-off/reboot operation is triggered. This
5016 functionality is particularly useful to increase reliability
5017 on embedded devices, but also on laptops which might
5018 accidentally get powered on when carried in a backpack and
5019 whose boot stays stuck in a hard disk encryption passphrase
5022 * systemd-logind can be configured to also handle lid switch
5023 events even when the machine is docked or multiple displays
5024 are attached (HandleLidSwitchDocked= option).
5026 * A helper binary and a service have been added which can be
5027 used to resume from hibernation in the initramfs. A
5028 generator will parse the resume= option on the kernel
5029 command line to trigger resume.
5031 * A user console daemon systemd-consoled has been
5032 added. Currently, it is a preview, and will so far open a
5033 single terminal on each session of the user marked as
5034 Desktop=systemd-console.
5036 * Route metrics can be specified for DHCP routes added by
5039 * The SELinux context of socket-activated services can be set
5040 from the information provided by the networking stack
5041 (SELinuxContextFromNet= option).
5043 * Userspace firmware loading support has been removed and
5044 the minimum supported kernel version is thus bumped to 3.7.
5046 * Timeout for udev workers has been increased from 1 to 3
5047 minutes, but a warning will be printed after 1 minute to
5048 help diagnose kernel modules that take a long time to load.
5050 * Udev rules can now remove tags on devices with TAG-="foobar".
5052 * systemd's readahead implementation has been removed. In many
5053 circumstances it didn't give expected benefits even for
5054 rotational disk drives and was becoming less relevant in the
5055 age of SSDs. As none of the developers has been using
5056 rotating media anymore, and nobody stepped up to actively
5057 maintain this component of systemd it has now been removed.
5059 * Swap units can use Options= to specify discard options.
5060 Discard options specified for swaps in /etc/fstab are now
5063 * Docker containers are now detected as a separate type of
5066 * The Password Agent protocol gained support for queries where
5067 the user input is shown, useful e.g. for user names.
5068 systemd-ask-password gained a new --echo option to turn that
5071 * The default sysctl.d/ snippets will now set:
5073 net.core.default_qdisc = fq_codel
5075 This selects Fair Queuing Controlled Delay as the default
5076 queuing discipline for network interfaces. fq_codel helps
5077 fight the network bufferbloat problem. It is believed to be
5078 a good default with no tuning required for most workloads.
5079 Downstream distributions may override this choice. On 10Gbit
5080 servers that do not do forwarding, "fq" may perform better.
5081 Systems without a good clocksource should use "pfifo_fast".
5083 * If kdbus is enabled during build a new option BusPolicy= is
5084 available for service units, that allows locking all service
5085 processes into a stricter bus policy, in order to limit
5086 access to various bus services, or even hide most of them
5087 from the service's view entirely.
5089 * networkctl will now show the .network and .link file
5090 networkd has applied to a specific interface.
5092 * sd-login gained a new API call sd_session_get_desktop() to
5093 query which desktop environment has been selected for a
5096 * UNIX utmp support is now compile-time optional to support
5097 legacy-free systems.
5099 * systemctl gained two new commands "add-wants" and
5100 "add-requires" for pulling in units from specific targets
5103 * If the word "rescue" is specified on the kernel command line
5104 the system will now boot into rescue mode (aka
5105 rescue.target), which was previously available only by
5106 specifying "1" or "systemd.unit=rescue.target" on the kernel
5107 command line. This new kernel command line option nicely
5108 mirrors the already existing "emergency" kernel command line
5111 * New kernel command line options mount.usr=, mount.usrflags=,
5112 mount.usrfstype= have been added that match root=, rootflags=,
5113 rootfstype= but allow mounting a specific file system to
5116 * The $NOTIFY_SOCKET is now also passed to control processes of
5117 services, not only the main process.
5119 * This version reenables support for fsck's -l switch. This
5120 means at least version v2.25 of util-linux is required for
5121 operation, otherwise dead-locks on device nodes may
5122 occur. Again: you need to update util-linux to at least
5123 v2.25 when updating systemd to v217.
5125 * The "multi-seat-x" tool has been removed from systemd, as
5126 its functionality has been integrated into X servers 1.16,
5127 and the tool is hence redundant. It is recommended to update
5128 display managers invoking this tool to simply invoke X
5129 directly from now on, again.
5131 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
5132 message flag has been added for all of systemd's polkit
5133 authenticated method calls has been added. In particular this
5134 now allows optional interactive authorization via polkit for
5135 many of PID1's privileged operations such as unit file
5136 enabling and disabling.
5138 * "udevadm hwdb --update" learnt a new switch "--usr" for
5139 placing the rebuilt hardware database in /usr instead of
5140 /etc. When used only hardware database entries stored in
5141 /usr will be used, and any user database entries in /etc are
5142 ignored. This functionality is useful for vendors to ship a
5143 pre-built database on systems where local configuration is
5144 unnecessary or unlikely.
5146 * Calendar time specifications in .timer units now also
5147 understand the strings "semi-annually", "quarterly" and
5148 "minutely" as shortcuts (in addition to the preexisting
5149 "anually", "hourly", ...).
5151 * systemd-tmpfiles will now correctly create files in /dev
5152 at boot which are marked for creation only at boot. It is
5153 recommended to always create static device nodes with 'c!'
5154 and 'b!', so that they are created only at boot and not
5155 overwritten at runtime.
5157 * When the watchdog logic is used for a service (WatchdogSec=)
5158 and the watchdog timeout is hit the service will now be
5159 terminated with SIGABRT (instead of just SIGTERM), in order
5160 to make sure a proper coredump and backtrace is
5161 generated. This ensures that hanging services will result in
5162 similar coredump/backtrace behaviour as services that hit a
5165 Contributions from: Andreas Henriksson, Andrei Borzenkov,
5166 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
5167 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
5168 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
5169 Herrmann, David Sommerseth, David Strauss, Emil Renner
5170 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
5171 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
5172 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
5173 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
5174 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
5175 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
5176 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
5177 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
5178 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
5179 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
5180 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
5181 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
5182 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
5183 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
5184 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
5185 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
5188 — Berlin, 2014-10-28
5192 * timedated no longer reads NTP implementation unit names from
5193 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
5194 implementations should add a
5196 Conflicts=systemd-timesyncd.service
5198 to their unit files to take over and replace systemd's NTP
5199 default functionality.
5201 * systemd-sysusers gained a new line type "r" for configuring
5202 which UID/GID ranges to allocate system users/groups
5203 from. Lines of type "u" may now add an additional column
5204 that specifies the home directory for the system user to be
5205 created. Also, systemd-sysusers may now optionally read user
5206 information from STDIN instead of a file. This is useful for
5207 invoking it from RPM preinst scriptlets that need to create
5208 users before the first RPM file is installed since these
5209 files might need to be owned by them. A new
5210 %sysusers_create_inline RPM macro has been introduced to do
5211 just that. systemd-sysusers now updates the shadow files as
5212 well as the user/group databases, which should enhance
5213 compatibility with certain tools like grpck.
5215 * A number of bus APIs of PID 1 now optionally consult polkit to
5216 permit access for otherwise unprivileged clients under certain
5217 conditions. Note that this currently doesn't support
5218 interactive authentication yet, but this is expected to be
5219 added eventually, too.
5221 * /etc/machine-info now has new fields for configuring the
5222 deployment environment of the machine, as well as the
5223 location of the machine. hostnamectl has been updated with
5224 new command to update these fields.
5226 * systemd-timesyncd has been updated to automatically acquire
5227 NTP server information from systemd-networkd, which might
5228 have been discovered via DHCP.
5230 * systemd-resolved now includes a caching DNS stub resolver
5231 and a complete LLMNR name resolution implementation. A new
5232 NSS module "nss-resolve" has been added which can be used
5233 instead of glibc's own "nss-dns" to resolve hostnames via
5234 systemd-resolved. Hostnames, addresses and arbitrary RRs may
5235 be resolved via systemd-resolved D-Bus APIs. In contrast to
5236 the glibc internal resolver systemd-resolved is aware of
5237 multi-homed system, and keeps DNS server and caches separate
5238 and per-interface. Queries are sent simultaneously on all
5239 interfaces that have DNS servers configured, in order to
5240 properly handle VPNs and local LANs which might resolve
5241 separate sets of domain names. systemd-resolved may acquire
5242 DNS server information from systemd-networkd automatically,
5243 which in turn might have discovered them via DHCP. A tool
5244 "systemd-resolve-host" has been added that may be used to
5245 query the DNS logic in resolved. systemd-resolved implements
5246 IDNA and automatically uses IDNA or UTF-8 encoding depending
5247 on whether classic DNS or LLMNR is used as transport. In the
5248 next releases we intend to add a DNSSEC and mDNS/DNS-SD
5249 implementation to systemd-resolved.
5251 * A new NSS module nss-mymachines has been added, that
5252 automatically resolves the names of all local registered
5253 containers to their respective IP addresses.
5255 * A new client tool "networkctl" for systemd-networkd has been
5256 added. It currently is entirely passive and will query
5257 networking configuration from udev, rtnetlink and networkd,
5258 and present it to the user in a very friendly
5259 way. Eventually, we hope to extend it to become a full
5260 control utility for networkd.
5262 * .socket units gained a new DeferAcceptSec= setting that
5263 controls the kernels' TCP_DEFER_ACCEPT sockopt for
5264 TCP. Similarly, support for controlling TCP keep-alive
5265 settings has been added (KeepAliveTimeSec=,
5266 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
5267 turning off Nagle's algorithm on TCP has been added
5270 * logind learned a new session type "web", for use in projects
5271 like Cockpit which register web clients as PAM sessions.
5273 * timer units with at least one OnCalendar= setting will now
5274 be started only after time-sync.target has been
5275 reached. This way they will not elapse before the system
5276 clock has been corrected by a local NTP client or
5277 similar. This is particular useful on RTC-less embedded
5278 machines, that come up with an invalid system clock.
5280 * systemd-nspawn's --network-veth= switch should now result in
5281 stable MAC addresses for both the outer and the inner side
5284 * systemd-nspawn gained a new --volatile= switch for running
5285 container instances with /etc or /var unpopulated.
5287 * The kdbus client code has been updated to use the new Linux
5288 3.17 memfd subsystem instead of the old kdbus-specific one.
5290 * systemd-networkd's DHCP client and server now support
5291 FORCERENEW. There are also new configuration options to
5292 configure the vendor client identifier and broadcast mode
5295 * systemd will no longer inform the kernel about the current
5296 timezone, as this is necessarily incorrect and racy as the
5297 kernel has no understanding of DST and similar
5298 concepts. This hence means FAT timestamps will be always
5299 considered UTC, similar to what Android is already
5300 doing. Also, when the RTC is configured to the local time
5301 (rather than UTC) systemd will never synchronize back to it,
5302 as this might confuse Windows at a later boot.
5304 * systemd-analyze gained a new command "verify" for offline
5305 validation of unit files.
5307 * systemd-networkd gained support for a couple of additional
5308 settings for bonding networking setups. Also, the metric for
5309 statically configured routes may now be configured. For
5310 network interfaces where this is appropriate the peer IP
5311 address may now be configured.
5313 * systemd-networkd's DHCP client will no longer request
5314 broadcasting by default, as this tripped up some networks.
5315 For hardware where broadcast is required the feature should
5316 be switched back on using RequestBroadcast=yes.
5318 * systemd-networkd will now set up IPv4LL addresses (when
5319 enabled) even if DHCP is configured successfully.
5321 * udev will now default to respect network device names given
5322 by the kernel when the kernel indicates that these are
5323 predictable. This behavior can be tweaked by changing
5324 NamePolicy= in the relevant .link file.
5326 * A new library systemd-terminal has been added that
5327 implements full TTY stream parsing and rendering. This
5328 library is supposed to be used later on for implementing a
5329 full userspace VT subsystem, replacing the current kernel
5332 * A new tool systemd-journal-upload has been added to push
5333 journal data to a remote system running
5334 systemd-journal-remote.
5336 * journald will no longer forward all local data to another
5337 running syslog daemon. This change has been made because
5338 rsyslog (which appears to be the most commonly used syslog
5339 implementation these days) no longer makes use of this, and
5340 instead pulls the data out of the journal on its own. Since
5341 forwarding the messages to a non-existent syslog server is
5342 more expensive than we assumed we have now turned this
5343 off. If you run a syslog server that is not a recent rsyslog
5344 version, you have to turn this option on again
5345 (ForwardToSyslog= in journald.conf).
5347 * journald now optionally supports the LZ4 compressor for
5348 larger journal fields. This compressor should perform much
5349 better than XZ which was the previous default.
5351 * machinectl now shows the IP addresses of local containers,
5352 if it knows them, plus the interface name of the container.
5354 * A new tool "systemd-escape" has been added that makes it
5355 easy to escape strings to build unit names and similar.
5357 * sd_notify() messages may now include a new ERRNO= field
5358 which is parsed and collected by systemd and shown among the
5359 "systemctl status" output for a service.
5361 * A new component "systemd-firstboot" has been added that
5362 queries the most basic systemd information (timezone,
5363 hostname, root password) interactively on first
5364 boot. Alternatively it may also be used to provision these
5365 things offline on OS images installed into directories.
5367 * The default sysctl.d/ snippets will now set
5369 net.ipv4.conf.default.promote_secondaries=1
5371 This has the benefit of no flushing secondary IP addresses
5372 when primary addresses are removed.
5374 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
5375 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
5376 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
5377 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
5378 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
5379 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
5380 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
5381 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
5382 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
5383 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
5384 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
5385 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
5386 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
5387 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
5388 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
5390 — Berlin, 2014-08-19
5394 * A new tool systemd-sysusers has been added. This tool
5395 creates system users and groups in /etc/passwd and
5396 /etc/group, based on static declarative system user/group
5397 definitions in /usr/lib/sysusers.d/. This is useful to
5398 enable factory resets and volatile systems that boot up with
5399 an empty /etc directory, and thus need system users and
5400 groups created during early boot. systemd now also ships
5401 with two default sysusers.d/ files for the most basic
5402 users and groups systemd and the core operating system
5405 * A new tmpfiles snippet has been added that rebuilds the
5406 essential files in /etc on boot, should they be missing.
5408 * A directive for ensuring automatic clean-up of
5409 /var/cache/man/ has been removed from the default
5410 configuration. This line should now be shipped by the man
5411 implementation. The necessary change has been made to the
5412 man-db implementation. Note that you need to update your man
5413 implementation to one that ships this line, otherwise no
5414 automatic clean-up of /var/cache/man will take place.
5416 * A new condition ConditionNeedsUpdate= has been added that
5417 may conditionalize services to only run when /etc or /var
5418 are "older" than the vendor operating system resources in
5419 /usr. This is useful for reconstructing or updating /etc
5420 after an offline update of /usr or a factory reset, on the
5421 next reboot. Services that want to run once after such an
5422 update or reset should use this condition and order
5423 themselves before the new systemd-update-done.service, which
5424 will mark the two directories as fully updated. A number of
5425 service files have been added making use of this, to rebuild
5426 the udev hardware database, the journald message catalog and
5427 dynamic loader cache (ldconfig). The systemd-sysusers tool
5428 described above also makes use of this now. With this in
5429 place it is now possible to start up a minimal operating
5430 system with /etc empty cleanly. For more information on the
5431 concepts involved see this recent blog story:
5433 http://0pointer.de/blog/projects/stateless.html
5435 * A new system group "input" has been introduced, and all
5436 input device nodes get this group assigned. This is useful
5437 for system-level software to get access to input devices. It
5438 complements what is already done for "audio" and "video".
5440 * systemd-networkd learnt minimal DHCPv4 server support in
5441 addition to the existing DHCPv4 client support. It also
5442 learnt DHCPv6 client and IPv6 Router Solicitation client
5443 support. The DHCPv4 client gained support for static routes
5444 passed in from the server. Note that the [DHCPv4] section
5445 known in older systemd-networkd versions has been renamed to
5446 [DHCP] and is now also used by the DHCPv6 client. Existing
5447 .network files using settings of this section should be
5448 updated, though compatibility is maintained. Optionally, the
5449 client hostname may now be sent to the DHCP server.
5451 * networkd gained support for vxlan virtual networks as well
5452 as tun/tap and dummy devices.
5454 * networkd gained support for automatic allocation of address
5455 ranges for interfaces from a system-wide pool of
5456 addresses. This is useful for dynamically managing a large
5457 number of interfaces with a single network configuration
5458 file. In particular this is useful to easily assign
5459 appropriate IP addresses to the veth links of a large number
5460 of nspawn instances.
5462 * RPM macros for processing sysusers, sysctl and binfmt
5463 drop-in snippets at package installation time have been
5466 * The /etc/os-release file should now be placed in
5467 /usr/lib/os-release. The old location is automatically
5468 created as symlink. /usr/lib is the more appropriate
5469 location of this file, since it shall actually describe the
5470 vendor operating system shipped in /usr, and not the
5471 configuration stored in /etc.
5473 * .mount units gained a new boolean SloppyOptions= setting
5474 that maps to mount(8)'s -s option which enables permissive
5475 parsing of unknown mount options.
5477 * tmpfiles learnt a new "L+" directive which creates a symlink
5478 but (unlike "L") deletes a pre-existing file first, should
5479 it already exist and not already be the correct
5480 symlink. Similarly, "b+", "c+" and "p+" directives have been
5481 added as well, which create block and character devices, as
5482 well as fifos in the filesystem, possibly removing any
5483 pre-existing files of different types.
5485 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
5486 'argument' field (which so far specified the source to
5487 symlink/copy the files from) is now optional. If omitted the
5488 same file os copied from /usr/share/factory/ suffixed by the
5489 full destination path. This is useful for populating /etc
5490 with essential files, by copying them from vendor defaults
5491 shipped in /usr/share/factory/etc.
5493 * A new command "systemctl preset-all" has been added that
5494 applies the service preset settings to all installed unit
5495 files. A new switch --preset-mode= has been added that
5496 controls whether only enable or only disable operations
5499 * A new command "systemctl is-system-running" has been added
5500 that allows checking the overall state of the system, for
5501 example whether it is fully up and running.
5503 * When the system boots up with an empty /etc, the equivalent
5504 to "systemctl preset-all" is executed during early boot, to
5505 make sure all default services are enabled after a factory
5508 * systemd now contains a minimal preset file that enables the
5509 most basic services systemd ships by default.
5511 * Unit files' [Install] section gained a new DefaultInstance=
5512 field for defining the default instance to create if a
5513 template unit is enabled with no instance specified.
5515 * A new passive target cryptsetup-pre.target has been added
5516 that may be used by services that need to make they run and
5517 finish before the first LUKS cryptographic device is set up.
5519 * The /dev/loop-control and /dev/btrfs-control device nodes
5520 are now owned by the "disk" group by default, opening up
5521 access to this group.
5523 * systemd-coredump will now automatically generate a
5524 stack trace of all core dumps taking place on the system,
5525 based on elfutils' libdw library. This stack trace is logged
5528 * systemd-coredump may now optionally store coredumps directly
5529 on disk (in /var/lib/systemd/coredump, possibly compressed),
5530 instead of storing them unconditionally in the journal. This
5531 mode is the new default. A new configuration file
5532 /etc/systemd/coredump.conf has been added to configure this
5533 and other parameters of systemd-coredump.
5535 * coredumpctl gained a new "info" verb to show details about a
5536 specific coredump. A new switch "-1" has also been added
5537 that makes sure to only show information about the most
5538 recent entry instead of all entries. Also, as the tool is
5539 generally useful now the "systemd-" prefix of the binary
5540 name has been removed. Distributions that want to maintain
5541 compatibility with the old name should add a symlink from
5542 the old name to the new name.
5544 * journald's SplitMode= now defaults to "uid". This makes sure
5545 that unprivileged users can access their own coredumps with
5546 coredumpctl without restrictions.
5548 * New kernel command line options "systemd.wants=" (for
5549 pulling an additional unit during boot), "systemd.mask="
5550 (for masking a specific unit for the boot), and
5551 "systemd.debug-shell" (for enabling the debug shell on tty9)
5552 have been added. This is implemented in the new generator
5553 "systemd-debug-generator".
5555 * systemd-nspawn will now by default filter a couple of
5556 syscalls for containers, among them those required for
5557 kernel module loading, direct x86 IO port access, swap
5558 management, and kexec. Most importantly though
5559 open_by_handle_at() is now prohibited for containers,
5560 closing a hole similar to a recently discussed vulnerability
5561 in docker regarding access to files on file hierarchies the
5562 container should normally not have access to. Note that, for
5563 nspawn, we generally make no security claims anyway (and
5564 this is explicitly documented in the man page), so this is
5565 just a fix for one of the most obvious problems.
5567 * A new man page file-hierarchy(7) has been added that
5568 contains a minimized, modernized version of the file system
5569 layout systemd expects, similar in style to the FHS
5570 specification or hier(5). A new tool systemd-path(1) has
5571 been added to query many of these paths for the local
5574 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
5575 longer done. Since the directory now has a per-user size
5576 limit, and is cleaned on logout this appears unnecessary,
5577 in particular since this now brings the lifecycle of this
5578 directory closer in line with how IPC objects are handled.
5580 * systemd.pc now exports a number of additional directories,
5581 including $libdir (which is useful to identify the library
5582 path for the primary architecture of the system), and a
5583 couple of drop-in directories.
5585 * udev's predictable network interface names now use the dev_port
5586 sysfs attribute, introduced in linux 3.15 instead of dev_id to
5587 distinguish between ports of the same PCI function. dev_id should
5588 only be used for ports using the same HW address, hence the need
5591 * machined has been updated to export the OS version of a
5592 container (read from /etc/os-release and
5593 /usr/lib/os-release) on the bus. This is now shown in
5594 "machinectl status" for a machine.
5596 * A new service setting RestartForceExitStatus= has been
5597 added. If configured to a set of exit signals or process
5598 return values, the service will be restarted when the main
5599 daemon process exits with any of them, regardless of the
5602 * systemctl's -H switch for connecting to remote systemd
5603 machines has been extended so that it may be used to
5604 directly connect to a specific container on the
5605 host. "systemctl -H root@foobar:waldi" will now connect as
5606 user "root" to host "foobar", and then proceed directly to
5607 the container named "waldi". Note that currently you have to
5608 authenticate as user "root" for this to work, as entering
5609 containers is a privileged operation.
5611 Contributions from: Andreas Henriksson, Benjamin Steinwender,
5612 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
5613 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
5614 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
5615 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
5616 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
5617 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
5618 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
5619 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
5620 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
5621 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
5622 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
5624 — Berlin, 2014-07-03
5628 * As an experimental feature, udev now tries to lock the
5629 disk device node (flock(LOCK_SH|LOCK_NB)) while it
5630 executes events for the disk or any of its partitions.
5631 Applications like partitioning programs can lock the
5632 disk device node (flock(LOCK_EX)) and claim temporary
5633 device ownership that way; udev will entirely skip all event
5634 handling for this disk and its partitions. If the disk
5635 was opened for writing, the close will trigger a partition
5636 table rescan in udev's "watch" facility, and if needed
5637 synthesize "change" events for the disk and all its partitions.
5638 This is now unconditionally enabled, and if it turns out to
5639 cause major problems, we might turn it on only for specific
5640 devices, or might need to disable it entirely. Device Mapper
5641 devices are excluded from this logic.
5643 * We temporarily dropped the "-l" switch for fsck invocations,
5644 since they collide with the flock() logic above. util-linux
5645 upstream has been changed already to avoid this conflict,
5646 and we will readd "-l" as soon as util-linux with this
5647 change has been released.
5649 * The dependency on libattr has been removed. Since a long
5650 time, the extended attribute calls have moved to glibc, and
5651 libattr is thus unnecessary.
5653 * Virtualization detection works without privileges now. This
5654 means the systemd-detect-virt binary no longer requires
5655 CAP_SYS_PTRACE file capabilities, and our daemons can run
5656 with fewer privileges.
5658 * systemd-networkd now runs under its own "systemd-network"
5659 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
5660 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
5661 loses the ability to write to files owned by root this way.
5663 * Similarly, systemd-resolved now runs under its own
5664 "systemd-resolve" user with no capabilities remaining.
5666 * Similarly, systemd-bus-proxyd now runs under its own
5667 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
5669 * systemd-networkd gained support for setting up "veth"
5670 virtual Ethernet devices for container connectivity, as well
5671 as GRE and VTI tunnels.
5673 * systemd-networkd will no longer automatically attempt to
5674 manually load kernel modules necessary for certain tunnel
5675 transports. Instead, it is assumed the kernel loads them
5676 automatically when required. This only works correctly on
5677 very new kernels. On older kernels, please consider adding
5678 the kernel modules to /etc/modules-load.d/ as a work-around.
5680 * The resolv.conf file systemd-resolved generates has been
5681 moved to /run/systemd/resolve/. If you have a symlink from
5682 /etc/resolv.conf, it might be necessary to correct it.
5684 * Two new service settings, ProtectHome= and ProtectSystem=,
5685 have been added. When enabled, they will make the user data
5686 (such as /home) inaccessible or read-only and the system
5687 (such as /usr) read-only, for specific services. This allows
5688 very light-weight per-service sandboxing to avoid
5689 modifications of user data or system files from
5690 services. These two new switches have been enabled for all
5691 of systemd's long-running services, where appropriate.
5693 * Socket units gained new SocketUser= and SocketGroup=
5694 settings to set the owner user and group of AF_UNIX sockets
5695 and FIFOs in the file system.
5697 * Socket units gained a new RemoveOnStop= setting. If enabled,
5698 all FIFOS and sockets in the file system will be removed
5699 when the specific socket unit is stopped.
5701 * Socket units gained a new Symlinks= setting. It takes a list
5702 of symlinks to create to file system sockets or FIFOs
5703 created by the specific Unix sockets. This is useful to
5704 manage symlinks to socket nodes with the same life-cycle as
5707 * The /dev/log socket and /dev/initctl FIFO have been moved to
5708 /run, and have been replaced by symlinks. This allows
5709 connecting to these facilities even if PrivateDevices=yes is
5710 used for a service (which makes /dev/log itself unavailable,
5711 but /run is left). This also has the benefit of ensuring
5712 that /dev only contains device nodes, directories and
5713 symlinks, and nothing else.
5715 * sd-daemon gained two new calls sd_pid_notify() and
5716 sd_pid_notifyf(). They are similar to sd_notify() and
5717 sd_notifyf(), but allow overriding of the source PID of
5718 notification messages if permissions permit this. This is
5719 useful to send notify messages on behalf of a different
5720 process (for example, the parent process). The
5721 systemd-notify tool has been updated to make use of this
5722 when sending messages (so that notification messages now
5723 originate from the shell script invoking systemd-notify and
5724 not the systemd-notify process itself. This should minimize
5725 a race where systemd fails to associate notification
5726 messages to services when the originating process already
5729 * A new "on-abnormal" setting for Restart= has been added. If
5730 set, it will result in automatic restarts on all "abnormal"
5731 reasons for a process to exit, which includes unclean
5732 signals, core dumps, timeouts and watchdog timeouts, but
5733 does not include clean and unclean exit codes or clean
5734 signals. Restart=on-abnormal is an alternative for
5735 Restart=on-failure for services that shall be able to
5736 terminate and avoid restarts on certain errors, by
5737 indicating so with an unclean exit code. Restart=on-failure
5738 or Restart=on-abnormal is now the recommended setting for
5739 all long-running services.
5741 * If the InaccessibleDirectories= service setting points to a
5742 mount point (or if there are any submounts contained within
5743 it), it is now attempted to completely unmount it, to make
5744 the file systems truly unavailable for the respective
5747 * The ReadOnlyDirectories= service setting and
5748 systemd-nspawn's --read-only parameter are now recursively
5749 applied to all submounts, too.
5751 * Mount units may now be created transiently via the bus APIs.
5753 * The support for SysV and LSB init scripts has been removed
5754 from the systemd daemon itself. Instead, it is now
5755 implemented as a generator that creates native systemd units
5756 from these scripts when needed. This enables us to remove a
5757 substantial amount of legacy code from PID 1, following the
5758 fact that many distributions only ship a very small number
5759 of LSB/SysV init scripts nowadays.
5761 * Privileged Xen (dom0) domains are not considered
5762 virtualization anymore by the virtualization detection
5763 logic. After all, they generally have unrestricted access to
5764 the hardware and usually are used to manage the unprivileged
5767 * systemd-tmpfiles gained a new "C" line type, for copying
5768 files or entire directories.
5770 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
5771 lines. So far, they have been non-globbing versions of the
5772 latter, and have thus been redundant. In future, it is
5773 recommended to only use "z". "m" has hence been removed
5774 from the documentation, even though it stays supported.
5776 * A tmpfiles snippet to recreate the most basic structure in
5777 /var has been added. This is enough to create the /var/run →
5778 /run symlink and create a couple of structural
5779 directories. This allows systems to boot up with an empty or
5780 volatile /var. Of course, while with this change, the core OS
5781 now is capable with dealing with a volatile /var, not all
5782 user services are ready for it. However, we hope that sooner
5783 or later, many service daemons will be changed upstream so
5784 that they are able to automatically create their necessary
5785 directories in /var at boot, should they be missing. This is
5786 the first step to allow state-less systems that only require
5787 the vendor image for /usr to boot.
5789 * systemd-nspawn has gained a new --tmpfs= switch to mount an
5790 empty tmpfs instance to a specific directory. This is
5791 particularly useful for making use of the automatic
5792 reconstruction of /var (see above), by passing --tmpfs=/var.
5794 * Access modes specified in tmpfiles snippets may now be
5795 prefixed with "~", which indicates that they shall be masked
5796 by whether the existing file or directory is currently
5797 writable, readable or executable at all. Also, if specified,
5798 the sgid/suid/sticky bits will be masked for all
5801 * A new passive target unit "network-pre.target" has been
5802 added which is useful for services that shall run before any
5803 network is configured, for example firewall scripts.
5805 * The "floppy" group that previously owned the /dev/fd*
5806 devices is no longer used. The "disk" group is now used
5807 instead. Distributions should probably deprecate usage of
5810 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
5811 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
5812 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
5813 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
5814 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
5815 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
5818 — Berlin, 2014-06-11
5822 * A new "systemd-timesyncd" daemon has been added for
5823 synchronizing the system clock across the network. It
5824 implements an SNTP client. In contrast to NTP
5825 implementations such as chrony or the NTP reference server,
5826 this only implements a client side, and does not bother with
5827 the full NTP complexity, focusing only on querying time from
5828 one remote server and synchronizing the local clock to
5829 it. Unless you intend to serve NTP to networked clients or
5830 want to connect to local hardware clocks, this simple NTP
5831 client should be more than appropriate for most
5832 installations. The daemon runs with minimal privileges, and
5833 has been hooked up with networkd to only operate when
5834 network connectivity is available. The daemon saves the
5835 current clock to disk every time a new NTP sync has been
5836 acquired, and uses this to possibly correct the system clock
5837 early at bootup, in order to accommodate for systems that
5838 lack an RTC such as the Raspberry Pi and embedded devices,
5839 and to make sure that time monotonically progresses on these
5840 systems, even if it is not always correct. To make use of
5841 this daemon, a new system user and group "systemd-timesync"
5842 needs to be created on installation of systemd.
5844 * The queue "seqnum" interface of libudev has been disabled, as
5845 it was generally incompatible with device namespacing as
5846 sequence numbers of devices go "missing" if the devices are
5847 part of a different namespace.
5849 * "systemctl list-timers" and "systemctl list-sockets" gained
5850 a --recursive switch for showing units of these types also
5851 for all local containers, similar in style to the already
5852 supported --recursive switch for "systemctl list-units".
5854 * A new RebootArgument= setting has been added for service
5855 units, which may be used to specify a kernel reboot argument
5856 to use when triggering reboots with StartLimitAction=.
5858 * A new FailureAction= setting has been added for service
5859 units which may be used to specify an operation to trigger
5860 when a service fails. This works similarly to
5861 StartLimitAction=, but unlike it, controls what is done
5862 immediately rather than only after several attempts to
5863 restart the service in question.
5865 * hostnamed got updated to also expose the kernel name,
5866 release, and version on the bus. This is useful for
5867 executing commands like hostnamectl with the -H switch.
5868 systemd-analyze makes use of this to properly display
5869 details when running non-locally.
5871 * The bootchart tool can now show cgroup information in the
5872 graphs it generates.
5874 * The CFS CPU quota cgroup attribute is now exposed for
5875 services. The new CPUQuota= switch has been added for this
5876 which takes a percentage value. Setting this will have the
5877 result that a service may never get more CPU time than the
5878 specified percentage, even if the machine is otherwise idle.
5880 * systemd-networkd learned IPIP and SIT tunnel support.
5882 * LSB init scripts exposing a dependency on $network will now
5883 get a dependency on network-online.target rather than simply
5884 network.target. This should bring LSB handling closer to
5885 what it was on SysV systems.
5887 * A new fsck.repair= kernel option has been added to control
5888 how fsck shall deal with unclean file systems at boot.
5890 * The (.ini) configuration file parser will now silently
5891 ignore sections whose name begins with "X-". This may be
5892 used to maintain application-specific extension sections in unit
5895 * machined gained a new API to query the IP addresses of
5896 registered containers. "machinectl status" has been updated
5897 to show these addresses in its output.
5899 * A new call sd_uid_get_display() has been added to the
5900 sd-login APIs for querying the "primary" session of a
5901 user. The "primary" session of the user is elected from the
5902 user's sessions and generally a graphical session is
5903 preferred over a text one.
5905 * A minimal systemd-resolved daemon has been added. It
5906 currently simply acts as a companion to systemd-networkd and
5907 manages resolv.conf based on per-interface DNS
5908 configuration, possibly supplied via DHCP. In the long run
5909 we hope to extend this into a local DNSSEC enabled DNS and
5912 * The systemd-networkd-wait-online tool is now enabled by
5913 default. It will delay network-online.target until a network
5914 connection has been configured. The tool primarily integrates
5915 with networkd, but will also make a best effort to make sense
5916 of network configuration performed in some other way.
5918 * Two new service options StartupCPUShares= and
5919 StartupBlockIOWeight= have been added that work similarly to
5920 CPUShares= and BlockIOWeight= however only apply during
5921 system startup. This is useful to prioritize certain services
5922 differently during bootup than during normal runtime.
5924 * hostnamed has been changed to prefer the statically
5925 configured hostname in /etc/hostname (unless set to
5926 'localhost' or empty) over any dynamic one supplied by
5927 dhcp. With this change, the rules for picking the hostname
5928 match more closely the rules of other configuration settings
5929 where the local administrator's configuration in /etc always
5930 overrides any other settings.
5932 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
5933 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
5934 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
5935 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
5936 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
5937 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
5938 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
5939 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
5940 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
5941 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
5942 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
5943 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
5944 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
5945 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
5946 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
5947 Lindskog, WaLyong Cho, Will Woods, Zbigniew
5950 — Beijing, 2014-05-28
5954 * When restoring the screen brightness at boot, stay away from
5955 the darkest setting or from the lowest 5% of the available
5956 range, depending on which is the larger value of both. This
5957 should effectively protect the user from rebooting into a
5958 black screen, should the brightness have been set to minimum
5961 * sd-login gained a new sd_machine_get_class() call to
5962 determine the class ("vm" or "container") of a machine
5963 registered with machined.
5965 * sd-login gained new calls
5966 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
5967 to query the identity of the peer of a local AF_UNIX
5968 connection. They operate similarly to their sd_pid_get_xyz()
5971 * PID 1 will now maintain a system-wide system state engine
5972 with the states "starting", "running", "degraded",
5973 "maintenance", "stopping". These states are bound to system
5974 startup, normal runtime, runtime with at least one failed
5975 service, rescue/emergency mode and system shutdown. This
5976 state is shown in the "systemctl status" output when no unit
5977 name is passed. It is useful to determine system state, in
5978 particularly when doing so for many systems or containers at
5981 * A new command "list-machines" has been added to "systemctl"
5982 that lists all local OS containers and shows their system
5983 state (see above), if systemd runs inside of them.
5985 * systemctl gained a new "-r" switch to recursively enumerate
5986 units on all local containers, when used with the
5987 "list-unit" command (which is the default one that is
5988 executed when no parameters are specified).
5990 * The GPT automatic partition discovery logic will now honour
5991 two GPT partition flags: one may be set on a partition to
5992 cause it to be mounted read-only, and the other may be set
5993 on a partition to ignore it during automatic discovery.
5995 * Two new GPT type UUIDs have been added for automatic root
5996 partition discovery, for 32-bit and 64-bit ARM. This is not
5997 particularly useful for discovering the root directory on
5998 these architectures during bare-metal boots (since UEFI is
5999 not common there), but still very useful to allow booting of
6000 ARM disk images in nspawn with the -i option.
6002 * MAC addresses of interfaces created with nspawn's
6003 --network-interface= switch will now be generated from the
6004 machine name, and thus be stable between multiple invocations
6007 * logind will now automatically remove all IPC objects owned
6008 by a user if she or he fully logs out. This makes sure that
6009 users who are logged out cannot continue to consume IPC
6010 resources. This covers SysV memory, semaphores and message
6011 queues as well as POSIX shared memory and message
6012 queues. Traditionally, SysV and POSIX IPC had no life-cycle
6013 limits. With this functionality, that is corrected. This may
6014 be turned off by using the RemoveIPC= switch of logind.conf.
6016 * The systemd-machine-id-setup and tmpfiles tools gained a
6017 --root= switch to operate on a specific root directory,
6020 * journald can now forward logged messages to the TTYs of all
6021 logged in users ("wall"). This is the default for all
6022 emergency messages now.
6024 * A new tool systemd-journal-remote has been added to stream
6025 journal log messages across the network.
6027 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
6028 controller trees are mounted into it. Note that the
6029 directories mounted beneath it are not read-only. This is a
6030 security measure and is particularly useful because glibc
6031 actually includes a search logic to pick any tmpfs it can
6032 find to implement shm_open() if /dev/shm is not available
6033 (which it might very well be in namespaced setups).
6035 * machinectl gained a new "poweroff" command to cleanly power
6036 down a local OS container.
6038 * The PrivateDevices= unit file setting will now also drop the
6039 CAP_MKNOD capability from the capability bound set, and
6040 imply DevicePolicy=closed.
6042 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
6043 comprehensively on all long-running systemd services where
6044 this is appropriate.
6046 * systemd-udevd will now run in a disassociated mount
6047 namespace. To mount directories from udev rules, make sure to
6048 pull in mount units via SYSTEMD_WANTS properties.
6050 * The kdbus support gained support for uploading policy into
6051 the kernel. sd-bus gained support for creating "monitoring"
6052 connections that can eavesdrop into all bus communication
6053 for debugging purposes.
6055 * Timestamps may now be specified in seconds since the UNIX
6056 epoch Jan 1st, 1970 by specifying "@" followed by the value
6059 * Native tcpwrap support in systemd has been removed. tcpwrap
6060 is old code, not really maintained anymore and has serious
6061 shortcomings, and better options such as firewalls
6062 exist. For setups that require tcpwrap usage, please
6063 consider invoking your socket-activated service via tcpd,
6064 like on traditional inetd.
6066 * A new system.conf configuration option
6067 DefaultTimerAccuracySec= has been added that controls the
6068 default AccuracySec= setting of .timer units.
6070 * Timer units gained a new WakeSystem= switch. If enabled,
6071 timers configured this way will cause the system to resume
6072 from system suspend (if the system supports that, which most
6075 * Timer units gained a new Persistent= switch. If enabled,
6076 timers configured this way will save to disk when they have
6077 been last triggered. This information is then used on next
6078 reboot to possible execute overdue timer events, that
6079 could not take place because the system was powered off.
6080 This enables simple anacron-like behaviour for timer units.
6082 * systemctl's "list-timers" will now also list the time a
6083 timer unit was last triggered in addition to the next time
6084 it will be triggered.
6086 * systemd-networkd will now assign predictable IPv4LL
6087 addresses to its local interfaces.
6089 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
6090 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
6091 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
6092 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
6093 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
6094 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
6095 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
6096 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
6099 — Berlin, 2014-03-25
6103 * A new unit file setting RestrictAddressFamilies= has been
6104 added to restrict which socket address families unit
6105 processes gain access to. This takes address family names
6106 like "AF_INET" or "AF_UNIX", and is useful to minimize the
6107 attack surface of services via exotic protocol stacks. This
6108 is built on seccomp system call filters.
6110 * Two new unit file settings RuntimeDirectory= and
6111 RuntimeDirectoryMode= have been added that may be used to
6112 manage a per-daemon runtime directories below /run. This is
6113 an alternative for setting up directory permissions with
6114 tmpfiles snippets, and has the advantage that the runtime
6115 directory's lifetime is bound to the daemon runtime and that
6116 the daemon starts up with an empty directory each time. This
6117 is particularly useful when writing services that drop
6118 privileges using the User= or Group= setting.
6120 * The DeviceAllow= unit setting now supports globbing for
6121 matching against device group names.
6123 * The systemd configuration file system.conf gained new
6124 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
6125 DefaultMemoryAccounting= to globally turn on/off accounting
6126 for specific resources (cgroups) for all units. These
6127 settings may still be overridden individually in each unit
6130 * systemd-gpt-auto-generator is now able to discover /srv and
6131 root partitions in addition to /home and swap partitions. It
6132 also supports LUKS-encrypted partitions now. With this in
6133 place, automatic discovery of partitions to mount following
6134 the Discoverable Partitions Specification
6135 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
6136 is now a lot more complete. This allows booting without
6137 /etc/fstab and without root= on the kernel command line on
6138 systems prepared appropriately.
6140 * systemd-nspawn gained a new --image= switch which allows
6141 booting up disk images and Linux installations on any block
6142 device that follow the Discoverable Partitions Specification
6143 (see above). This means that installations made with
6144 appropriately updated installers may now be started and
6145 deployed using container managers, completely
6146 unmodified. (We hope that libvirt-lxc will add support for
6147 this feature soon, too.)
6149 * systemd-nspawn gained a new --network-macvlan= setting to
6150 set up a private macvlan interface for the
6151 container. Similarly, systemd-networkd gained a new
6152 Kind=macvlan setting in .netdev files.
6154 * systemd-networkd now supports configuring local addresses
6157 * A new tool systemd-network-wait-online has been added to
6158 synchronously wait for network connectivity using
6161 * The sd-bus.h bus API gained a new sd_bus_track object for
6162 tracking the life-cycle of bus peers. Note that sd-bus.h is
6163 still not a public API though (unless you specify
6164 --enable-kdbus on the configure command line, which however
6165 voids your warranty and you get no API stability guarantee).
6167 * The $XDG_RUNTIME_DIR runtime directories for each user are
6168 now individual tmpfs instances, which has the benefit of
6169 introducing separate pools for each user, with individual
6170 size limits, and thus making sure that unprivileged clients
6171 can no longer negatively impact the system or other users by
6172 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
6173 RuntimeDirectorySize= has been introduced that allows
6174 controlling the default size limit for all users. It
6175 defaults to 10% of the available physical memory. This is no
6176 replacement for quotas on tmpfs though (which the kernel
6177 still does not support), as /dev/shm and /tmp are still
6178 shared resources used by both the system and unprivileged
6181 * logind will now automatically turn off automatic suspending
6182 on laptop lid close when more than one display is
6183 connected. This was previously expected to be implemented
6184 individually in desktop environments (such as GNOME),
6185 however has been added to logind now, in order to fix a
6186 boot-time race where a desktop environment might not have
6187 been started yet and thus not been able to take an inhibitor
6188 lock at the time where logind already suspends the system
6189 due to a closed lid.
6191 * logind will now wait at least 30s after each system
6192 suspend/resume cycle, and 3min after system boot before
6193 suspending the system due to a closed laptop lid. This
6194 should give USB docking stations and similar enough time to
6195 be probed and configured after system resume and boot in
6196 order to then act as suspend blocker.
6198 * systemd-run gained a new --property= setting which allows
6199 initialization of resource control properties (and others)
6200 for the created scope or service unit. Example: "systemd-run
6201 --property=BlockIOWeight=10 updatedb" may be used to run
6202 updatedb at a low block IO scheduling weight.
6204 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
6205 now also work in --scope mode.
6207 * When systemd is compiled with kdbus support, basic support
6208 for enforced policies is now in place. (Note that enabling
6209 kdbus still voids your warranty and no API compatibility
6212 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
6213 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
6214 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
6215 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
6216 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
6217 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
6218 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
6219 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
6220 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
6221 Zbigniew Jędrzejewski-Szmek
6223 — Berlin, 2014-03-12
6227 * systemd will now relabel /dev after loading the SMACK policy
6228 according to SMACK rules.
6230 * A new unit file option AppArmorProfile= has been added to
6231 set the AppArmor profile for the processes of a unit.
6233 * A new condition check ConditionArchitecture= has been added
6234 to conditionalize units based on the system architecture, as
6235 reported by uname()'s "machine" field.
6237 * systemd-networkd now supports matching on the system
6238 virtualization, architecture, kernel command line, host name
6241 * logind is now a lot more aggressive when suspending the
6242 machine due to a closed laptop lid. Instead of acting only
6243 on the lid close action, it will continuously watch the lid
6244 status and act on it. This is useful for laptops where the
6245 power button is on the outside of the chassis so that it can
6246 be reached without opening the lid (such as the Lenovo
6247 Yoga). On those machines, logind will now immediately
6248 re-suspend the machine if the power button has been
6249 accidentally pressed while the laptop was suspended and in a
6250 backpack or similar.
6252 * logind will now watch SW_DOCK switches and inhibit reaction
6253 to the lid switch if it is pressed. This means that logind
6254 will not suspend the machine anymore if the lid is closed
6255 and the system is docked, if the laptop supports SW_DOCK
6256 notifications via the input layer. Note that ACPI docking
6257 stations do not generate this currently. Also note that this
6258 logic is usually not fully sufficient and Desktop
6259 Environments should take a lid switch inhibitor lock when an
6260 external display is connected, as systemd will not watch
6263 * nspawn will now make use of the devices cgroup controller by
6264 default, and only permit creation of and access to the usual
6265 API device nodes like /dev/null or /dev/random, as well as
6266 access to (but not creation of) the pty devices.
6268 * We will now ship a default .network file for
6269 systemd-networkd that automatically configures DHCP for
6270 network interfaces created by nspawn's --network-veth or
6271 --network-bridge= switches.
6273 * systemd will now understand the usual M, K, G, T suffixes
6274 according to SI conventions (i.e. to the base 1000) when
6275 referring to throughput and hardware metrics. It will stay
6276 with IEC conventions (i.e. to the base 1024) for software
6277 metrics, according to what is customary according to
6278 Wikipedia. We explicitly document which base applies for
6279 each configuration option.
6281 * The DeviceAllow= setting in unit files now supports a syntax
6282 to whitelist an entire group of devices node majors at once,
6283 based on the /proc/devices listing. For example, with the
6284 string "char-pts", it is now possible to whitelist all
6285 current and future pseudo-TTYs at once.
6287 * sd-event learned a new "post" event source. Event sources of
6288 this type are triggered by the dispatching of any event
6289 source of a type that is not "post". This is useful for
6290 implementing clean-up and check event sources that are
6291 triggered by other work being done in the program.
6293 * systemd-networkd is no longer statically enabled, but uses
6294 the usual [Install] sections so that it can be
6295 enabled/disabled using systemctl. It still is enabled by
6298 * When creating a veth interface pair with systemd-nspawn, the
6299 host side will now be prefixed with "vb-" if
6300 --network-bridge= is used, and with "ve-" if --network-veth
6301 is used. This way, it is easy to distinguish these cases on
6302 the host, for example to apply different configuration to
6303 them with systemd-networkd.
6305 * The compatibility libraries for libsystemd-journal.so,
6306 libsystem-id128.so, libsystemd-login.so and
6307 libsystemd-daemon.so do not make use of IFUNC
6308 anymore. Instead, we now build libsystemd.so multiple times
6309 under these alternative names. This means that the footprint
6310 is drastically increased, but given that these are
6311 transitional compatibility libraries, this should not matter
6312 much. This change has been made necessary to support the ARM
6313 platform for these compatibility libraries, as the ARM
6314 toolchain is not really at the same level as the toolchain
6315 for other architectures like x86 and does not support
6316 IFUNC. Please make sure to use --enable-compat-libs only
6317 during a transitional period!
6319 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
6320 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
6321 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
6322 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
6323 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
6324 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
6325 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
6326 Zbigniew Jędrzejewski-Szmek
6328 — Berlin, 2014-02-24
6332 * A new component "systemd-networkd" has been added that can
6333 be used to configure local network interfaces statically or
6334 via DHCP. It is capable of bringing up bridges, VLANs, and
6335 bonding. Currently, no hook-ups for interactive network
6336 configuration are provided. Use this for your initrd,
6337 container, embedded, or server setup if you need a simple,
6338 yet powerful, network configuration solution. This
6339 configuration subsystem is quite nifty, as it allows wildcard
6340 hotplug matching in interfaces. For example, with a single
6341 configuration snippet, you can configure that all Ethernet
6342 interfaces showing up are automatically added to a bridge,
6343 or similar. It supports link-sensing and more.
6345 * A new tool "systemd-socket-proxyd" has been added which can
6346 act as a bidirectional proxy for TCP sockets. This is
6347 useful for adding socket activation support to services that
6348 do not actually support socket activation, including virtual
6349 machines and the like.
6351 * Add a new tool to save/restore rfkill state on
6354 * Save/restore state of keyboard backlights in addition to
6355 display backlights on shutdown/boot.
6357 * udev learned a new SECLABEL{} construct to label device
6358 nodes with a specific security label when they appear. For
6359 now, only SECLABEL{selinux} is supported, but the syntax is
6360 prepared for additional security frameworks.
6362 * udev gained a new scheme to configure link-level attributes
6363 from files in /etc/systemd/network/*.link. These files can
6364 match against MAC address, device path, driver name and type,
6365 and will apply attributes like the naming policy, link speed,
6366 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
6367 address assignment policy (randomized, ...).
6369 * The configuration of network interface naming rules for
6370 "permanent interface names" has changed: a new NamePolicy=
6371 setting in the [Link] section of .link files determines the
6372 priority of possible naming schemes (onboard, slot, MAC,
6373 path). The default value of this setting is determined by
6374 /usr/lib/net/links/99-default.link. Old
6375 80-net-name-slot.rules udev configuration file has been
6376 removed, so local configuration overriding this file should
6377 be adapted to override 99-default.link instead.
6379 * When the User= switch is used in a unit file, also
6380 initialize $SHELL= based on the user database entry.
6382 * systemd no longer depends on libdbus. All communication is
6383 now done with sd-bus, systemd's low-level bus library
6386 * kdbus support has been added to PID 1 itself. When kdbus is
6387 enabled, this causes PID 1 to set up the system bus and
6388 enable support for a new ".busname" unit type that
6389 encapsulates bus name activation on kdbus. It works a little
6390 bit like ".socket" units, except for bus names. A new
6391 generator has been added that converts classic dbus1 service
6392 activation files automatically into native systemd .busname
6395 * sd-bus: add a light-weight vtable implementation that allows
6396 defining objects on the bus with a simple static const
6397 vtable array of its methods, signals and properties.
6399 * systemd will not generate or install static dbus
6400 introspection data anymore to /usr/share/dbus-1/interfaces,
6401 as the precise format of these files is unclear, and
6402 nothing makes use of it.
6404 * A proxy daemon is now provided to proxy clients connecting
6405 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
6406 compatibility with classic D-Bus.
6408 * A bus driver implementation has been added that supports the
6409 classic D-Bus bus driver calls on kdbus, also for
6410 compatibility purposes.
6412 * A new API "sd-event.h" has been added that implements a
6413 minimal event loop API built around epoll. It provides a
6414 couple of features that direct epoll usage is lacking:
6415 prioritization of events, scales to large numbers of timer
6416 events, per-event timer slack (accuracy), system-wide
6417 coalescing of timer events, exit handlers, watchdog
6418 supervision support using systemd's sd_notify() API, child
6421 * A new API "sd-rntl.h" has been added that provides an API
6422 around the route netlink interface of the kernel, similar in
6423 style to "sd-bus.h".
6425 * A new API "sd-dhcp-client.h" has been added that provides a
6426 small DHCPv4 client-side implementation. This is used by
6429 * There is a new kernel command line option
6430 "systemd.restore_state=0|1". When set to "0", none of the
6431 systemd tools will restore saved runtime state to hardware
6432 devices. More specifically, the rfkill and backlight states
6435 * The FsckPassNo= compatibility option in mount/service units
6436 has been removed. The fstab generator will now add the
6437 necessary dependencies automatically, and does not require
6438 PID1's support for that anymore.
6440 * journalctl gained a new switch, --list-boots, that lists
6441 recent boots with their times and boot IDs.
6443 * The various tools like systemctl, loginctl, timedatectl,
6444 busctl, systemd-run, ... have gained a new switch "-M" to
6445 connect to a specific, local OS container (as direct
6446 connection, without requiring SSH). This works on any
6447 container that is registered with machined, such as those
6448 created by libvirt-lxc or nspawn.
6450 * systemd-run and systemd-analyze also gained support for "-H"
6451 to connect to remote hosts via SSH. This is particularly
6452 useful for systemd-run because it enables queuing of jobs
6453 onto remote systems.
6455 * machinectl gained a new command "login" to open a getty
6456 login in any local container. This works with any container
6457 that is registered with machined (such as those created by
6458 libvirt-lxc or nspawn), and which runs systemd inside.
6460 * machinectl gained a new "reboot" command that may be used to
6461 trigger a reboot on a specific container that is registered
6462 with machined. This works on any container that runs an init
6463 system of some kind.
6465 * systemctl gained a new "list-timers" command to print a nice
6466 listing of installed timer units with the times they elapse
6469 * Alternative reboot() parameters may now be specified on the
6470 "systemctl reboot" command line and are passed to the
6471 reboot() system call.
6473 * systemctl gained a new --job-mode= switch to configure the
6474 mode to queue a job with. This is a more generic version of
6475 --fail, --irreversible, and --ignore-dependencies, which are
6476 still available but not advertised anymore.
6478 * /etc/systemd/system.conf gained new settings to configure
6479 various default timeouts of units, as well as the default
6480 start limit interval and burst. These may still be overridden
6483 * PID1 will now export on the bus profile data of the security
6484 policy upload process (such as the SELinux policy upload to
6487 * journald: when forwarding logs to the console, include
6488 timestamps (following the setting in
6489 /sys/module/printk/parameters/time).
6491 * OnCalendar= in timer units now understands the special
6492 strings "yearly" and "annually". (Both are equivalent)
6494 * The accuracy of timer units is now configurable with the new
6495 AccuracySec= setting. It defaults to 1min.
6497 * A new dependency type JoinsNamespaceOf= has been added that
6498 allows running two services within the same /tmp and network
6499 namespace, if PrivateNetwork= or PrivateTmp= are used.
6501 * A new command "cat" has been added to systemctl. It outputs
6502 the original unit file of a unit, and concatenates the
6503 contents of additional "drop-in" unit file snippets, so that
6504 the full configuration is shown.
6506 * systemctl now supports globbing on the various "list-xyz"
6507 commands, like "list-units" or "list-sockets", as well as on
6508 those commands which take multiple unit names.
6510 * journalctl's --unit= switch gained support for globbing.
6512 * All systemd daemons now make use of the watchdog logic so
6513 that systemd automatically notices when they hang.
6515 * If the $container_ttys environment variable is set,
6516 getty-generator will automatically spawn a getty for each
6517 listed tty. This is useful for container managers to request
6518 login gettys to be spawned on as many ttys as needed.
6520 * %h, %s, %U specifier support is not available anymore when
6521 used in unit files for PID 1. This is because NSS calls are
6522 not safe from PID 1. They stay available for --user
6523 instances of systemd, and as special case for the root user.
6525 * loginctl gained a new "--no-legend" switch to turn off output
6528 * The "sd-login.h" API gained three new calls:
6529 sd_session_is_remote(), sd_session_get_remote_user(),
6530 sd_session_get_remote_host() to query information about
6533 * The udev hardware database now also carries vendor/product
6534 information of SDIO devices.
6536 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
6537 determine whether watchdog notifications are requested by
6540 * Socket-activated per-connection services now include a
6541 short description of the connection parameters in the
6544 * tmpfiles gained a new "--boot" option. When this is not used,
6545 only lines where the command character is not suffixed with
6546 "!" are executed. When this option is specified, those
6547 options are executed too. This partitions tmpfiles
6548 directives into those that can be safely executed at any
6549 time, and those which should be run only at boot (for
6550 example, a line that creates /run/nologin).
6552 * A new API "sd-resolve.h" has been added which provides a simple
6553 asynchronous wrapper around glibc NSS host name resolution
6554 calls, such as getaddrinfo(). In contrast to glibc's
6555 getaddrinfo_a(), it does not use signals. In contrast to most
6556 other asynchronous name resolution libraries, this one does
6557 not reimplement DNS, but reuses NSS, so that alternate
6558 host name resolution systems continue to work, such as mDNS,
6559 LDAP, etc. This API is based on libasyncns, but it has been
6560 cleaned up for inclusion in systemd.
6562 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
6563 "sd-daemon.h" are no longer found in individual libraries
6564 libsystemd-journal.so, libsystemd-login.so,
6565 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
6566 merged them into a single library, libsystemd.so, which
6567 provides all symbols. The reason for this is cyclic
6568 dependencies, as these libraries tend to use each other's
6569 symbols. So far, we have managed to workaround that by linking
6570 a copy of a good part of our code into each of these
6571 libraries again and again, which, however, makes certain
6572 things hard to do, like sharing static variables. Also, it
6573 substantially increases footprint. With this change, there
6574 is only one library for the basic APIs systemd
6575 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
6576 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
6577 library as well, however are subject to the --enable-kdbus
6578 switch (see below). Note that "sd-dhcp-client.h" is not part
6579 of this library (this is because it only consumes, never
6580 provides, services of/to other APIs). To make the transition
6581 easy from the separate libraries to the unified one, we
6582 provide the --enable-compat-libs compile-time switch which
6583 will generate stub libraries that are compatible with the
6584 old ones but redirect all calls to the new one.
6586 * All of the kdbus logic and the new APIs "sd-bus.h",
6587 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
6588 and "sd-utf8.h" are compile-time optional via the
6589 "--enable-kdbus" switch, and they are not compiled in by
6590 default. To make use of kdbus, you have to explicitly enable
6591 the switch. Note however, that neither the kernel nor the
6592 userspace API for all of this is considered stable yet. We
6593 want to maintain the freedom to still change the APIs for
6594 now. By specifying this build-time switch, you acknowledge
6595 that you are aware of the instability of the current
6598 * Also, note that while kdbus is pretty much complete,
6599 it lacks one thing: proper policy support. This means you
6600 can build a fully working system with all features; however,
6601 it will be highly insecure. Policy support will be added in
6602 one of the next releases, at the same time that we will
6603 declare the APIs stable.
6605 * When the kernel command line argument "kdbus" is specified,
6606 systemd will automatically load the kdbus.ko kernel module. At
6607 this stage of development, it is only useful for testing kdbus
6608 and should not be used in production. Note: if "--enable-kdbus"
6609 is specified, and the kdbus.ko kernel module is available, and
6610 "kdbus" is added to the kernel command line, the entire system
6611 runs with kdbus instead of dbus-daemon, with the above mentioned
6612 problem of missing the system policy enforcement. Also a future
6613 version of kdbus.ko or a newer systemd will not be compatible with
6614 each other, and will unlikely be able to boot the machine if only
6615 one of them is updated.
6617 * systemctl gained a new "import-environment" command which
6618 uploads the caller's environment (or parts thereof) into the
6619 service manager so that it is inherited by services started
6620 by the manager. This is useful to upload variables like
6621 $DISPLAY into the user service manager.
6623 * A new PrivateDevices= switch has been added to service units
6624 which allows running a service with a namespaced /dev
6625 directory that does not contain any device nodes for
6626 physical devices. More specifically, it only includes devices
6627 such as /dev/null, /dev/urandom, and /dev/zero which are API
6630 * logind has been extended to support behaviour like VT
6631 switching on seats that do not support a VT. This makes
6632 multi-session available on seats that are not the first seat
6633 (seat0), and on systems where kernel support for VTs has
6634 been disabled at compile-time.
6636 * If a process holds a delay lock for system sleep or shutdown
6637 and fails to release it in time, we will now log its
6638 identity. This makes it easier to identify processes that
6639 cause slow suspends or power-offs.
6641 * When parsing /etc/crypttab, support for a new key-slot=
6642 option as supported by Debian is added. It allows indicating
6643 which LUKS slot to use on disk, speeding up key loading.
6645 * The sd_journal_sendv() API call has been checked and
6646 officially declared to be async-signal-safe so that it may
6647 be invoked from signal handlers for logging purposes.
6649 * Boot-time status output is now enabled automatically after a
6650 short timeout if boot does not progress, in order to give
6651 the user an indication what she or he is waiting for.
6653 * The boot-time output has been improved to show how much time
6654 remains until jobs expire.
6656 * The KillMode= switch in service units gained a new possible
6657 value "mixed". If set, and the unit is shut down, then the
6658 initial SIGTERM signal is sent only to the main daemon
6659 process, while the following SIGKILL signal is sent to
6660 all remaining processes of the service.
6662 * When a scope unit is registered, a new property "Controller"
6663 may be set. If set to a valid bus name, systemd will send a
6664 RequestStop() signal to this name when it would like to shut
6665 down the scope. This may be used to hook manager logic into
6666 the shutdown logic of scope units. Also, scope units may now
6667 be put in a special "abandoned" state, in which case the
6668 manager process which created them takes no further
6669 responsibilities for it.
6671 * When reading unit files, systemd will now verify
6672 the access mode of these files, and warn about certain
6673 suspicious combinations. This has been added to make it
6674 easier to track down packaging bugs where unit files are
6675 marked executable or world-writable.
6677 * systemd-nspawn gained a new "--setenv=" switch to set
6678 container-wide environment variables. The similar option in
6679 systemd-activate was renamed from "--environment=" to
6680 "--setenv=" for consistency.
6682 * systemd-nspawn has been updated to create a new kdbus domain
6683 for each container that is invoked, thus allowing each
6684 container to have its own set of system and user buses,
6685 independent of the host.
6687 * systemd-nspawn gained a new --drop-capability= switch to run
6688 the container with less capabilities than the default. Both
6689 --drop-capability= and --capability= now take the special
6690 string "all" for dropping or keeping all capabilities.
6692 * systemd-nspawn gained new switches for executing containers
6693 with specific SELinux labels set.
6695 * systemd-nspawn gained a new --quiet switch to not generate
6696 any additional output but the container's own console
6699 * systemd-nspawn gained a new --share-system switch to run a
6700 container without PID namespacing enabled.
6702 * systemd-nspawn gained a new --register= switch to control
6703 whether the container is registered with systemd-machined or
6704 not. This is useful for containers that do not run full
6705 OS images, but only specific apps.
6707 * systemd-nspawn gained a new --keep-unit which may be used
6708 when invoked as the only program from a service unit, and
6709 results in registration of the unit service itself in
6710 systemd-machined, instead of a newly opened scope unit.
6712 * systemd-nspawn gained a new --network-interface= switch for
6713 moving arbitrary interfaces to the container. The new
6714 --network-veth switch creates a virtual Ethernet connection
6715 between host and container. The new --network-bridge=
6716 switch then allows assigning the host side of this virtual
6717 Ethernet connection to a bridge device.
6719 * systemd-nspawn gained a new --personality= switch for
6720 setting the kernel personality for the container. This is
6721 useful when running a 32-bit container on a 64-bit host. A
6722 similar option Personality= is now also available for service
6725 * logind will now also track a "Desktop" identifier for each
6726 session which encodes the desktop environment of it. This is
6727 useful for desktop environments that want to identify
6728 multiple running sessions of itself easily.
6730 * A new SELinuxContext= setting for service units has been
6731 added that allows setting a specific SELinux execution
6732 context for a service.
6734 * Most systemd client tools will now honour $SYSTEMD_LESS for
6735 settings of the "less" pager. By default, these tools will
6736 override $LESS to allow certain operations to work, such as
6737 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
6738 influence this logic.
6740 * systemd's "seccomp" hook-up has been changed to make use of
6741 the libseccomp library instead of using its own
6742 implementation. This has benefits for portability among
6745 * For usage together with SystemCallFilter=, a new
6746 SystemCallErrorNumber= setting has been introduced that
6747 allows configuration of a system error number to be returned
6748 on filtered system calls, instead of immediately killing the
6749 process. Also, SystemCallArchitectures= has been added to
6750 limit access to system calls of a particular architecture
6751 (in order to turn off support for unused secondary
6752 architectures). There is also a global
6753 SystemCallArchitectures= setting in system.conf now to turn
6754 off support for non-native system calls system-wide.
6756 * systemd requires a kernel with a working name_to_handle_at(),
6757 please see the kernel config requirements in the README file.
6759 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
6760 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
6761 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
6762 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
6763 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
6764 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
6765 Elia Pinto, Florian Weimer, George McCollister, Goffredo
6766 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
6767 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
6768 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
6769 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
6770 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
6771 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
6772 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
6773 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
6774 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
6775 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
6776 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
6777 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
6778 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
6779 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
6780 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
6781 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
6782 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
6784 — Berlin, 2014-02-20
6788 * logind has gained support for facilitating privileged input
6789 and drm device access for unprivileged clients. This work is
6790 useful to allow Wayland display servers (and similar
6791 programs, such as kmscon) to run under the user's ID and
6792 access input and drm devices which are normally
6793 protected. When this is used (and the kernel is new enough)
6794 logind will "mute" IO on the file descriptors passed to
6795 Wayland as long as it is in the background and "unmute" it
6796 if it returns into the foreground. This allows secure
6797 session switching without allowing background sessions to
6798 eavesdrop on input and display data. This also introduces
6799 session switching support if VT support is turned off in the
6800 kernel, and on seats that are not seat0.
6802 * A new kernel command line option luks.options= is understood
6803 now which allows specifying LUKS options for usage for LUKS
6804 encrypted partitions specified with luks.uuid=.
6806 * tmpfiles.d(5) snippets may now use specifier expansion in
6807 path names. More specifically %m, %b, %H, %v, are now
6808 replaced by the local machine id, boot id, hostname, and
6809 kernel version number.
6811 * A new tmpfiles.d(5) command "m" has been introduced which
6812 may be used to change the owner/group/access mode of a file
6813 or directory if it exists, but do nothing if it does not.
6815 * This release removes high-level support for the
6816 MemorySoftLimit= cgroup setting. The underlying kernel
6817 cgroup attribute memory.soft_limit= is currently badly
6818 designed and likely to be removed from the kernel API in its
6819 current form, hence we should not expose it for now.
6821 * The memory.use_hierarchy cgroup attribute is now enabled for
6822 all cgroups systemd creates in the memory cgroup
6823 hierarchy. This option is likely to be come the built-in
6824 default in the kernel anyway, and the non-hierarchical mode
6825 never made much sense in the intrinsically hierarchical
6828 * A new field _SYSTEMD_SLICE= is logged along with all journal
6829 messages containing the slice a message was generated
6830 from. This is useful to allow easy per-customer filtering of
6831 logs among other things.
6833 * systemd-journald will no longer adjust the group of journal
6834 files it creates to the "systemd-journal" group. Instead we
6835 rely on the journal directory to be owned by the
6836 "systemd-journal" group, and its setgid bit set, so that the
6837 kernel file system layer will automatically enforce that
6838 journal files inherit this group assignment. The reason for
6839 this change is that we cannot allow NSS look-ups from
6840 journald which would be necessary to resolve
6841 "systemd-journal" to a numeric GID, because this might
6842 create deadlocks if NSS involves synchronous queries to
6843 other daemons (such as nscd, or sssd) which in turn are
6844 logging clients of journald and might block on it, which
6845 would then dead lock. A tmpfiles.d(5) snippet included in
6846 systemd will make sure the setgid bit and group are
6847 properly set on the journal directory if it exists on every
6848 boot. However, we recommend adjusting it manually after
6849 upgrades too (or from RPM scriptlets), so that the change is
6850 not delayed until next reboot.
6852 * Backlight and random seed files in /var/lib/ have moved into
6853 the /var/lib/systemd/ directory, in order to centralize all
6854 systemd generated files in one directory.
6856 * Boot time performance measurements (as displayed by
6857 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
6858 performance information if that's available to determine how
6859 much time BIOS and boot loader initialization required. With
6860 a sufficiently new BIOS you hence no longer need to boot
6861 with Gummiboot to get access to such information.
6863 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
6864 Cristian Rodríguez, Dave Reisner, David Herrmann, David
6865 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
6866 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
6867 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
6868 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
6869 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6871 — Berlin, 2013-10-02
6875 * The Restart= option for services now understands a new
6876 on-watchdog setting, which will restart the service
6877 automatically if the service stops sending out watchdog keep
6878 alive messages (as configured with WatchdogSec=).
6880 * The getty generator (which is responsible for bringing up a
6881 getty on configured serial consoles) will no longer only
6882 start a getty on the primary kernel console but on all
6883 others, too. This makes the order in which console= is
6884 specified on the kernel command line less important.
6886 * libsystemd-logind gained a new sd_session_get_vt() call to
6887 retrieve the VT number of a session.
6889 * If the option "tries=0" is set for an entry of /etc/crypttab
6890 its passphrase is queried indefinitely instead of any
6891 maximum number of tries.
6893 * If a service with a configure PID file terminates its PID
6894 file will now be removed automatically if it still exists
6895 afterwards. This should put an end to stale PID files.
6897 * systemd-run will now also take relative binary path names
6898 for execution and no longer insists on absolute paths.
6900 * InaccessibleDirectories= and ReadOnlyDirectories= now take
6901 paths that are optionally prefixed with "-" to indicate that
6902 it should not be considered a failure if they do not exist.
6904 * journalctl -o (and similar commands) now understands a new
6905 output mode "short-precise", it is similar to "short" but
6906 shows timestamps with usec accuracy.
6908 * The option "discard" (as known from Debian) is now
6909 synonymous to "allow-discards" in /etc/crypttab. In fact,
6910 "discard" is preferred now (since it is easier to remember
6913 * Some licensing clean-ups were made, so that more code is now
6914 LGPL-2.1 licensed than before.
6916 * A minimal tool to save/restore the display backlight
6917 brightness across reboots has been added. It will store the
6918 backlight setting as late as possible at shutdown, and
6919 restore it as early as possible during reboot.
6921 * A logic to automatically discover and enable home and swap
6922 partitions on GPT disks has been added. With this in place
6923 /etc/fstab becomes optional for many setups as systemd can
6924 discover certain partitions located on the root disk
6925 automatically. Home partitions are recognized under their
6926 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
6927 partitions are recognized under their GPT type ID
6928 0657fd6da4ab43c484e50933c84b4f4f.
6930 * systemd will no longer pass any environment from the kernel
6931 or initrd to system services. If you want to set an
6932 environment for all services, do so via the kernel command
6933 line systemd.setenv= assignment.
6935 * The systemd-sysctl tool no longer natively reads the file
6936 /etc/sysctl.conf. If desired, the file should be symlinked
6937 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
6938 legacy support by a symlink rather than built-in code, it
6939 also makes the otherwise hidden order of application of the
6940 different files visible. (Note that this partly reverts to a
6941 pre-198 application order of sysctl knobs!)
6943 * The "systemctl set-log-level" and "systemctl dump" commands
6944 have been moved to systemd-analyze.
6946 * systemd-run learned the new --remain-after-exit switch,
6947 which causes the scope unit not to be cleaned up
6948 automatically after the process terminated.
6950 * tmpfiles learned a new --exclude-prefix= switch to exclude
6951 certain paths from operation.
6953 * journald will now automatically flush all messages to disk
6954 as soon as a message at the log level CRIT, ALERT or EMERG
6957 Contributions from: Andrew Cook, Brandon Philips, Christian
6958 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
6959 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
6960 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
6961 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
6962 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
6963 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
6964 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
6965 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
6966 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
6967 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
6968 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
6969 William Giokas, Zbigniew Jędrzejewski-Szmek
6971 — Berlin, 2013-09-13
6975 * The documentation has been updated to cover the various new
6976 concepts introduced with 205.
6978 * Unit files now understand the new %v specifier which
6979 resolves to the kernel version string as returned by "uname
6982 * systemctl now supports filtering the unit list output by
6983 load state, active state and sub state, using the new
6986 * "systemctl status" will now show the results of the
6987 condition checks (like ConditionPathExists= and similar) of
6988 the last start attempts of the unit. They are also logged to
6991 * "journalctl -b" may now be used to look for boot output of a
6992 specific boot. Try "journalctl -b -1" for the previous boot,
6993 but the syntax is substantially more powerful.
6995 * "journalctl --show-cursor" has been added which prints the
6996 cursor string the last shown log line. This may then be used
6997 with the new "journalctl --after-cursor=" switch to continue
6998 browsing logs from that point on.
7000 * "journalctl --force" may now be used to force regeneration
7003 * Creation of "dead" device nodes has been moved from udev
7004 into kmod and tmpfiles. Previously, udev would read the kmod
7005 databases to pre-generate dead device nodes based on meta
7006 information contained in kernel modules, so that these would
7007 be auto-loaded on access rather then at boot. As this
7008 does not really have much to do with the exposing actual
7009 kernel devices to userspace this has always been slightly
7010 alien in the udev codebase. Following the new scheme kmod
7011 will now generate a runtime snippet for tmpfiles from the
7012 module meta information and it now is tmpfiles' job to the
7013 create the nodes. This also allows overriding access and
7014 other parameters for the nodes using the usual tmpfiles
7015 facilities. As side effect this allows us to remove the
7016 CAP_SYS_MKNOD capability bit from udevd entirely.
7018 * logind's device ACLs may now be applied to these "dead"
7019 devices nodes too, thus finally allowing managed access to
7020 devices such as /dev/snd/sequencer without loading the
7021 backing module right-away.
7023 * A new RPM macro has been added that may be used to apply
7024 tmpfiles configuration during package installation.
7026 * systemd-detect-virt and ConditionVirtualization= now can
7027 detect User-Mode-Linux machines (UML).
7029 * journald will now implicitly log the effective capabilities
7030 set of processes in the message metadata.
7032 * systemd-cryptsetup has gained support for TrueCrypt volumes.
7034 * The initrd interface has been simplified (more specifically,
7035 support for passing performance data via environment
7036 variables and fsck results via files in /run has been
7037 removed). These features were non-essential, and are
7038 nowadays available in a much nicer way by having systemd in
7039 the initrd serialize its state and have the hosts systemd
7040 deserialize it again.
7042 * The udev "keymap" data files and tools to apply keyboard
7043 specific mappings of scan to key codes, and force-release
7044 scan code lists have been entirely replaced by a udev
7045 "keyboard" builtin and a hwdb data file.
7047 * systemd will now honour the kernel's "quiet" command line
7048 argument also during late shutdown, resulting in a
7049 completely silent shutdown when used.
7051 * There's now an option to control the SO_REUSEPORT socket
7052 option in .socket units.
7054 * Instance units will now automatically get a per-template
7055 subslice of system.slice unless something else is explicitly
7056 configured. For example, instances of sshd@.service will now
7057 implicitly be placed in system-sshd.slice rather than
7058 system.slice as before.
7060 * Test coverage support may now be enabled at build time.
7062 Contributions from: Dave Reisner, Frederic Crozat, Harald
7063 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
7064 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
7065 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
7066 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
7067 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
7068 Giokas, Zbigniew Jędrzejewski-Szmek
7070 — Berlin, 2013-07-23
7074 * Two new unit types have been introduced:
7076 Scope units are very similar to service units, however, are
7077 created out of pre-existing processes — instead of PID 1
7078 forking off the processes. By using scope units it is
7079 possible for system services and applications to group their
7080 own child processes (worker processes) in a powerful way
7081 which then maybe used to organize them, or kill them
7082 together, or apply resource limits on them.
7084 Slice units may be used to partition system resources in an
7085 hierarchical fashion and then assign other units to them. By
7086 default there are now three slices: system.slice (for all
7087 system services), user.slice (for all user sessions),
7088 machine.slice (for VMs and containers).
7090 Slices and scopes have been introduced primarily in
7091 context of the work to move cgroup handling to a
7092 single-writer scheme, where only PID 1
7093 creates/removes/manages cgroups.
7095 * There's a new concept of "transient" units. In contrast to
7096 normal units these units are created via an API at runtime,
7097 not from configuration from disk. More specifically this
7098 means it is now possible to run arbitrary programs as
7099 independent services, with all execution parameters passed
7100 in via bus APIs rather than read from disk. Transient units
7101 make systemd substantially more dynamic then it ever was,
7102 and useful as a general batch manager.
7104 * logind has been updated to make use of scope and slice units
7105 for managing user sessions. As a user logs in he will get
7106 his own private slice unit, to which all sessions are added
7107 as scope units. We also added support for automatically
7108 adding an instance of user@.service for the user into the
7109 slice. Effectively logind will no longer create cgroup
7110 hierarchies on its own now, it will defer entirely to PID 1
7111 for this by means of scope, service and slice units. Since
7112 user sessions this way become entities managed by PID 1
7113 the output of "systemctl" is now a lot more comprehensive.
7115 * A new mini-daemon "systemd-machined" has been added which
7116 may be used by virtualization managers to register local
7117 VMs/containers. nspawn has been updated accordingly, and
7118 libvirt will be updated shortly. machined will collect a bit
7119 of meta information about the VMs/containers, and assign
7120 them their own scope unit (see above). The collected
7121 meta-data is then made available via the "machinectl" tool,
7122 and exposed in "ps" and similar tools. machined/machinectl
7123 is compile-time optional.
7125 * As discussed earlier, the low-level cgroup configuration
7126 options ControlGroup=, ControlGroupModify=,
7127 ControlGroupPersistent=, ControlGroupAttribute= have been
7128 removed. Please use high-level attribute settings instead as
7129 well as slice units.
7131 * A new bus call SetUnitProperties() has been added to alter
7132 various runtime parameters of a unit. This is primarily
7133 useful to alter cgroup parameters dynamically in a nice way,
7134 but will be extended later on to make more properties
7135 modifiable at runtime. systemctl gained a new set-properties
7136 command that wraps this call.
7138 * A new tool "systemd-run" has been added which can be used to
7139 run arbitrary command lines as transient services or scopes,
7140 while configuring a number of settings via the command
7141 line. This tool is currently very basic, however already
7142 very useful. We plan to extend this tool to even allow
7143 queuing of execution jobs with time triggers from the
7144 command line, similar in fashion to "at".
7146 * nspawn will now inform the user explicitly that kernels with
7147 audit enabled break containers, and suggest the user to turn
7150 * Support for detecting the IMA and AppArmor security
7151 frameworks with ConditionSecurity= has been added.
7153 * journalctl gained a new "-k" switch for showing only kernel
7154 messages, mimicking dmesg output; in addition to "--user"
7155 and "--system" switches for showing only user's own logs
7158 * systemd-delta can now show information about drop-in
7159 snippets extending unit files.
7161 * libsystemd-bus has been substantially updated but is still
7162 not available as public API.
7164 * systemd will now look for the "debug" argument on the kernel
7165 command line and enable debug logging, similar to what
7166 "systemd.log_level=debug" already did before.
7168 * "systemctl set-default", "systemctl get-default" has been
7169 added to configure the default.target symlink, which
7170 controls what to boot into by default.
7172 * "systemctl set-log-level" has been added as a convenient
7173 way to raise and lower systemd logging threshold.
7175 * "systemd-analyze plot" will now show the time the various
7176 generators needed for execution, as well as information
7177 about the unit file loading.
7179 * libsystemd-journal gained a new sd_journal_open_files() call
7180 for opening specific journal files. journactl also gained a
7181 new switch to expose this new functionality. Previously we
7182 only supported opening all files from a directory, or all
7183 files from the system, as opening individual files only is
7184 racy due to journal file rotation.
7186 * systemd gained the new DefaultEnvironment= setting in
7187 /etc/systemd/system.conf to set environment variables for
7190 * If a privileged process logs a journal message with the
7191 OBJECT_PID= field set, then journald will automatically
7192 augment this with additional OBJECT_UID=, OBJECT_GID=,
7193 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
7194 system services want to log events about specific client
7195 processes. journactl/systemctl has been updated to make use
7196 of this information if all log messages regarding a specific
7199 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
7200 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
7201 Reisner, David Coppa, David King, David Strauss, Eelco
7202 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
7203 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
7204 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
7205 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
7206 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
7207 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
7208 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
7209 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
7210 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
7211 Łukasz Stelmach, 장동준
7215 * The Python bindings gained some minimal support for the APIs
7216 exposed by libsystemd-logind.
7218 * ConditionSecurity= gained support for detecting SMACK. Since
7219 this condition already supports SELinux and AppArmor we only
7220 miss IMA for this. Patches welcome!
7222 Contributions from: Karol Lewandowski, Lennart Poettering,
7223 Zbigniew Jędrzejewski-Szmek
7227 * systemd-nspawn will now create /etc/resolv.conf if
7228 necessary, before bind-mounting the host's file onto it.
7230 * systemd-nspawn will now store meta information about a
7231 container on the container's cgroup as extended attribute
7232 fields, including the root directory.
7234 * The cgroup hierarchy has been reworked in many ways. All
7235 objects any of the components systemd creates in the cgroup
7236 tree are now suffixed. More specifically, user sessions are
7237 now placed in cgroups suffixed with ".session", users in
7238 cgroups suffixed with ".user", and nspawn containers in
7239 cgroups suffixed with ".nspawn". Furthermore, all cgroup
7240 names are now escaped in a simple scheme to avoid collision
7241 of userspace object names with kernel filenames. This work
7242 is preparation for making these objects relocatable in the
7243 cgroup tree, in order to allow easy resource partitioning of
7244 these objects without causing naming conflicts.
7246 * systemctl list-dependencies gained the new switches
7247 --plain, --reverse, --after and --before.
7249 * systemd-inhibit now shows the process name of processes that
7250 have taken an inhibitor lock.
7252 * nss-myhostname will now also resolve "localhost"
7253 implicitly. This makes /etc/hosts an optional file and
7254 nicely handles that on IPv6 ::1 maps to both "localhost" and
7257 * libsystemd-logind.so gained a new call
7258 sd_get_machine_names() to enumerate running containers and
7259 VMs (currently only supported by very new libvirt and
7260 nspawn). sd_login_monitor can now be used to watch
7261 VMs/containers coming and going.
7263 * .include is not allowed recursively anymore, and only in
7264 unit files. Usually it is better to use drop-in snippets in
7265 .d/*.conf anyway, as introduced with systemd 198.
7267 * systemd-analyze gained a new "critical-chain" command that
7268 determines the slowest chain of units run during system
7269 boot-up. It is very useful for tracking down where
7270 optimizing boot time is the most beneficial.
7272 * systemd will no longer allow manipulating service paths in
7273 the name=systemd:/system cgroup tree using ControlGroup= in
7274 units. (But is still fine with it in all other dirs.)
7276 * There's a new systemd-nspawn@.service service file that may
7277 be used to easily run nspawn containers as system
7278 services. With the container's root directory in
7279 /var/lib/container/foobar it is now sufficient to run
7280 "systemctl start systemd-nspawn@foobar.service" to boot it.
7282 * systemd-cgls gained a new parameter "--machine" to list only
7283 the processes within a certain container.
7285 * ConditionSecurity= now can check for "apparmor". We still
7286 are lacking checks for SMACK and IMA for this condition
7287 check though. Patches welcome!
7289 * A new configuration file /etc/systemd/sleep.conf has been
7290 added that may be used to configure which kernel operation
7291 systemd is supposed to execute when "suspend", "hibernate"
7292 or "hybrid-sleep" is requested. This makes the new kernel
7293 "freeze" state accessible to the user.
7295 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
7296 the passed argument if applicable.
7298 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
7299 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
7300 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
7301 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
7302 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
7303 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
7304 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
7309 * The output of 'systemctl list-jobs' got some polishing. The
7310 '--type=' argument may now be passed more than once. A new
7311 command 'systemctl list-sockets' has been added which shows
7312 a list of kernel sockets systemd is listening on with the
7313 socket units they belong to, plus the units these socket
7316 * The experimental libsystemd-bus library got substantial
7317 updates to work in conjunction with the (also experimental)
7318 kdbus kernel project. It works well enough to exchange
7319 messages with some sophistication. Note that kdbus is not
7320 ready yet, and the library is mostly an elaborate test case
7321 for now, and not installable.
7323 * systemd gained a new unit 'systemd-static-nodes.service'
7324 that generates static device nodes earlier during boot, and
7325 can run in conjunction with udev.
7327 * libsystemd-login gained a new call sd_pid_get_user_unit()
7328 to retrieve the user systemd unit a process is running
7329 in. This is useful for systems where systemd is used as
7332 * systemd-nspawn now places all containers in the new /machine
7333 top-level cgroup directory in the name=systemd
7334 hierarchy. libvirt will soon do the same, so that we get a
7335 uniform separation of /system, /user and /machine for system
7336 services, user processes and containers/virtual
7337 machines. This new cgroup hierarchy is also useful to stick
7338 stable names to specific container instances, which can be
7339 recognized later this way (this name may be controlled
7340 via systemd-nspawn's new -M switch). libsystemd-login also
7341 gained a new call sd_pid_get_machine_name() to retrieve the
7342 name of the container/VM a specific process belongs to.
7344 * bootchart can now store its data in the journal.
7346 * libsystemd-journal gained a new call
7347 sd_journal_add_conjunction() for AND expressions to the
7348 matching logic. This can be used to express more complex
7349 logical expressions.
7351 * journactl can now take multiple --unit= and --user-unit=
7354 * The cryptsetup logic now understands the "luks.key=" kernel
7355 command line switch for specifying a file to read the
7356 decryption key from. Also, if a configured key file is not
7357 found the tool will now automatically fall back to prompting
7360 * Python systemd.journal module was updated to wrap recently
7361 added functions from libsystemd-journal. The interface was
7362 changed to bring the low level interface in s.j._Reader
7363 closer to the C API, and the high level interface in
7364 s.j.Reader was updated to wrap and convert all data about
7367 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
7368 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
7369 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
7370 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
7371 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
7372 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
7376 * journalctl --update-catalog now understands a new --root=
7377 option to operate on catalogs found in a different root
7380 * During shutdown after systemd has terminated all running
7381 services a final killing loop kills all remaining left-over
7382 processes. We will now print the name of these processes
7383 when we send SIGKILL to them, since this usually indicates a
7386 * If /etc/crypttab refers to password files stored on
7387 configured mount points automatic dependencies will now be
7388 generated to ensure the specific mount is established first
7389 before the key file is attempted to be read.
7391 * 'systemctl status' will now show information about the
7392 network sockets a socket unit is listening on.
7394 * 'systemctl status' will also shown information about any
7395 drop-in configuration file for units. (Drop-In configuration
7396 files in this context are files such as
7397 /etc/systemd/systemd/foobar.service.d/*.conf)
7399 * systemd-cgtop now optionally shows summed up CPU times of
7400 cgroups. Press '%' while running cgtop to switch between
7401 percentage and absolute mode. This is useful to determine
7402 which cgroups use up the most CPU time over the entire
7403 runtime of the system. systemd-cgtop has also been updated
7404 to be 'pipeable' for processing with further shell tools.
7406 * 'hostnamectl set-hostname' will now allow setting of FQDN
7409 * The formatting and parsing of time span values has been
7410 changed. The parser now understands fractional expressions
7411 such as "5.5h". The formatter will now output fractional
7412 expressions for all time spans under 1min, i.e. "5.123456s"
7413 rather than "5s 123ms 456us". For time spans under 1s
7414 millisecond values are shown, for those under 1ms
7415 microsecond values are shown. This should greatly improve
7416 all time-related output of systemd.
7418 * libsystemd-login and libsystemd-journal gained new
7419 functions for querying the poll() events mask and poll()
7420 timeout value for integration into arbitrary event
7423 * localectl gained the ability to list available X11 keymaps
7424 (models, layouts, variants, options).
7426 * 'systemd-analyze dot' gained the ability to filter for
7427 specific units via shell-style globs, to create smaller,
7428 more useful graphs. I.e. it is now possible to create simple
7429 graphs of all the dependencies between only target units, or
7430 of all units that Avahi has dependencies with.
7432 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
7433 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
7434 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
7435 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
7436 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
7437 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
7438 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
7442 * The boot-time readahead implementation for rotating media
7443 will now read the read-ahead data in multiple passes which
7444 consist of all read requests made in equidistant time
7445 intervals. This means instead of strictly reading read-ahead
7446 data in its physical order on disk we now try to find a
7447 middle ground between physical and access time order.
7449 * /etc/os-release files gained a new BUILD_ID= field for usage
7450 on operating systems that provide continuous builds of OS
7453 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
7454 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
7455 William Douglas, Zbigniew Jędrzejewski-Szmek
7459 * systemd-python gained an API exposing libsystemd-daemon.
7461 * The SMACK setup logic gained support for uploading CIPSO
7464 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
7465 ReadOnlyDirectories= and InaccessibleDirectories= has
7466 changed. The private /tmp and /var/tmp directories are now
7467 shared by all processes of a service (which means
7468 ExecStartPre= may now leave data in /tmp that ExecStart= of
7469 the same service can still access). When a service is
7470 stopped its temporary directories are immediately deleted
7471 (normal clean-up with tmpfiles is still done in addition to
7474 * By default, systemd will now set a couple of sysctl
7475 variables in the kernel: the safe sysrq options are turned
7476 on, IP route verification is turned on, and source routing
7477 disabled. The recently added hardlink and softlink
7478 protection of the kernel is turned on. These settings should
7479 be reasonably safe, and good defaults for all new systems.
7481 * The predictable network naming logic may now be turned off
7482 with a new kernel command line switch: net.ifnames=0.
7484 * A new libsystemd-bus module has been added that implements a
7485 pretty complete D-Bus client library. For details see:
7487 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
7489 * journald will now explicitly flush the journal files to disk
7490 at the latest 5min after each write. The file will then also
7491 be marked offline until the next write. This should increase
7492 reliability in case of a crash. The synchronization delay
7493 can be configured via SyncIntervalSec= in journald.conf.
7495 * There's a new remote-fs-setup.target unit that can be used
7496 to pull in specific services when at least one remote file
7497 system is to be mounted.
7499 * There are new targets timers.target and paths.target as
7500 canonical targets to pull user timer and path units in
7501 from. This complements sockets.target with a similar
7502 purpose for socket units.
7504 * libudev gained a new call udev_device_set_attribute_value()
7505 to set sysfs attributes of a device.
7507 * The udev daemon now sets the default number of worker
7508 processes executed in parallel based on the number of available
7509 CPUs instead of the amount of available RAM. This is supposed
7510 to provide a more reliable default and limit a too aggressive
7511 parallelism for setups with 1000s of devices connected.
7513 Contributions from: Auke Kok, Colin Walters, Cristian
7514 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
7515 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
7516 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
7517 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
7518 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
7519 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
7520 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
7521 Zbigniew Jędrzejewski-Szmek
7525 * Configuration of unit files may now be extended via drop-in
7526 files without having to edit/override the unit files
7527 themselves. More specifically, if the administrator wants to
7528 change one value for a service file foobar.service he can
7529 now do so by dropping in a configuration snippet into
7530 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
7531 will load all these snippets and apply them on top of the
7532 main unit configuration file, possibly extending or
7533 overriding its settings. Using these drop-in snippets is
7534 generally nicer than the two earlier options for changing
7535 unit files locally: copying the files from
7536 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
7537 them there; or creating a new file in /etc/systemd/system/
7538 that incorporates the original one via ".include". Drop-in
7539 snippets into these .d/ directories can be placed in any
7540 directory systemd looks for units in, and the usual
7541 overriding semantics between /usr/lib, /etc and /run apply
7544 * Most unit file settings which take lists of items can now be
7545 reset by assigning the empty string to them. For example,
7546 normally, settings such as Environment=FOO=BAR append a new
7547 environment variable assignment to the environment block,
7548 each time they are used. By assigning Environment= the empty
7549 string the environment block can be reset to empty. This is
7550 particularly useful with the .d/*.conf drop-in snippets
7551 mentioned above, since this adds the ability to reset list
7552 settings from vendor unit files via these drop-ins.
7554 * systemctl gained a new "list-dependencies" command for
7555 listing the dependencies of a unit recursively.
7557 * Inhibitors are now honored and listed by "systemctl
7558 suspend", "systemctl poweroff" (and similar) too, not only
7559 GNOME. These commands will also list active sessions by
7562 * Resource limits (as exposed by the various control group
7563 controllers) can now be controlled dynamically at runtime
7564 for all units. More specifically, you can now use a command
7565 like "systemctl set-cgroup-attr foobar.service cpu.shares
7566 2000" to alter the CPU shares a specific service gets. These
7567 settings are stored persistently on disk, and thus allow the
7568 administrator to easily adjust the resource usage of
7569 services with a few simple commands. This dynamic resource
7570 management logic is also available to other programs via the
7571 bus. Almost any kernel cgroup attribute and controller is
7574 * systemd-vconsole-setup will now copy all font settings to
7575 all allocated VTs, where it previously applied them only to
7578 * libsystemd-login gained the new sd_session_get_tty() API
7581 * This release drops support for a few legacy or
7582 distribution-specific LSB facility names when parsing init
7583 scripts: $x-display-manager, $mail-transfer-agent,
7584 $mail-transport-agent, $mail-transfer-agent, $smtp,
7585 $null. Also, the mail-transfer-agent.target unit backing
7586 this has been removed. Distributions which want to retain
7587 compatibility with this should carry the burden for
7588 supporting this themselves and patch support for these back
7589 in, if they really need to. Also, the facilities $syslog and
7590 $local_fs are now ignored, since systemd does not support
7591 early-boot LSB init scripts anymore, and these facilities
7592 are implied anyway for normal services. syslog.target has
7595 * There are new bus calls on PID1's Manager object for
7596 cancelling jobs, and removing snapshot units. Previously,
7597 both calls were only available on the Job and Snapshot
7600 * systemd-journal-gatewayd gained SSL support.
7602 * The various "environment" files, such as /etc/locale.conf
7603 now support continuation lines with a backslash ("\") as
7604 last character in the line, similarly in style (but different)
7605 to how this is supported in shells.
7607 * For normal user processes the _SYSTEMD_USER_UNIT= field is
7608 now implicitly appended to every log entry logged. systemctl
7609 has been updated to filter by this field when operating on a
7610 user systemd instance.
7612 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
7613 CAP_AUDIT_CONTROL capabilities to the capabilities set for
7614 the container. This makes it easier to boot unmodified
7615 Fedora systems in a container, which however still requires
7616 audit=0 to be passed on the kernel command line. Auditing in
7617 kernel and userspace is unfortunately still too broken in
7618 context of containers, hence we recommend compiling it out
7619 of the kernel or using audit=0. Hopefully this will be fixed
7620 one day for good in the kernel.
7622 * nspawn gained the new --bind= and --bind-ro= parameters to
7623 bind mount specific directories from the host into the
7626 * nspawn will now mount its own devpts file system instance
7627 into the container, in order not to leak pty devices from
7628 the host into the container.
7630 * systemd will now read the firmware boot time performance
7631 information from the EFI variables, if the used boot loader
7632 supports this, and takes it into account for boot performance
7633 analysis via "systemd-analyze". This is currently supported
7634 only in conjunction with Gummiboot, but could be supported
7635 by other boot loaders too. For details see:
7637 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
7639 * A new generator has been added that automatically mounts the
7640 EFI System Partition (ESP) to /boot, if that directory
7641 exists, is empty, and no other file system has been
7642 configured to be mounted there.
7644 * logind will now send out PrepareForSleep(false) out
7645 unconditionally, after coming back from suspend. This may be
7646 used by applications as asynchronous notification for
7647 system resume events.
7649 * "systemctl unlock-sessions" has been added, that allows
7650 unlocking the screens of all user sessions at once, similar
7651 to how "systemctl lock-sessions" already locked all users
7652 sessions. This is backed by a new D-Bus call UnlockSessions().
7654 * "loginctl seat-status" will now show the master device of a
7655 seat. (i.e. the device of a seat that needs to be around for
7656 the seat to be considered available, usually the graphics
7659 * tmpfiles gained a new "X" line type, that allows
7660 configuration of files and directories (with wildcards) that
7661 shall be excluded from automatic cleanup ("aging").
7663 * udev default rules set the device node permissions now only
7664 at "add" events, and do not change them any longer with a
7665 later "change" event.
7667 * The log messages for lid events and power/sleep keypresses
7668 now carry a message ID.
7670 * We now have a substantially larger unit test suite, but this
7671 continues to be work in progress.
7673 * udevadm hwdb gained a new --root= parameter to change the
7674 root directory to operate relative to.
7676 * logind will now issue a background sync() request to the kernel
7677 early at shutdown, so that dirty buffers are flushed to disk early
7678 instead of at the last moment, in order to optimize shutdown
7681 * A new bootctl tool has been added that is an interface for
7682 certain boot loader operations. This is currently a preview
7683 and is likely to be extended into a small mechanism daemon
7684 like timedated, localed, hostnamed, and can be used by
7685 graphical UIs to enumerate available boot options, and
7686 request boot into firmware operations.
7688 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
7689 the rest of the package. It also has been updated to work
7690 correctly in initrds.
7692 * polkit previously has been runtime optional, and is now also
7693 compile time optional via a configure switch.
7695 * systemd-analyze has been reimplemented in C. Also "systemctl
7696 dot" has moved into systemd-analyze.
7698 * "systemctl status" with no further parameters will now print
7699 the status of all active or failed units.
7701 * Operations such as "systemctl start" can now be executed
7702 with a new mode "--irreversible" which may be used to queue
7703 operations that cannot accidentally be reversed by a later
7704 job queuing. This is by default used to make shutdown
7705 requests more robust.
7707 * The Python API of systemd now gained a new module for
7708 reading journal files.
7710 * A new tool kernel-install has been added that can install
7711 kernel images according to the Boot Loader Specification:
7713 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
7715 * Boot time console output has been improved to provide
7716 animated boot time output for hanging jobs.
7718 * A new tool systemd-activate has been added which can be used
7719 to test socket activation with, directly from the command
7720 line. This should make it much easier to test and debug
7721 socket activation in daemons.
7723 * journalctl gained a new "--reverse" (or -r) option to show
7724 journal output in reverse order (i.e. newest line first).
7726 * journalctl gained a new "--pager-end" (or -e) option to jump
7727 to immediately jump to the end of the journal in the
7728 pager. This is only supported in conjunction with "less".
7730 * journalctl gained a new "--user-unit=" option, that works
7731 similarly to "--unit=" but filters for user units rather than
7734 * A number of unit files to ease adoption of systemd in
7735 initrds has been added. This moves some minimal logic from
7736 the various initrd implementations into systemd proper.
7738 * The journal files are now owned by a new group
7739 "systemd-journal", which exists specifically to allow access
7740 to the journal, and nothing else. Previously, we used the
7741 "adm" group for that, which however possibly covers more
7742 than just journal/log file access. This new group is now
7743 already used by systemd-journal-gatewayd to ensure this
7744 daemon gets access to the journal files and as little else
7745 as possible. Note that "make install" will also set FS ACLs
7746 up for /var/log/journal to give "adm" and "wheel" read
7747 access to it, in addition to "systemd-journal" which owns
7748 the journal files. We recommend that packaging scripts also
7749 add read access to "adm" + "wheel" to /var/log/journal, and
7750 all existing/future journal files. To normal users and
7751 administrators little changes, however packagers need to
7752 ensure to create the "systemd-journal" system group at
7753 package installation time.
7755 * The systemd-journal-gatewayd now runs as unprivileged user
7756 systemd-journal-gateway:systemd-journal-gateway. Packaging
7757 scripts need to create these system user/group at
7760 * timedated now exposes a new boolean property CanNTP that
7761 indicates whether a local NTP service is available or not.
7763 * systemd-detect-virt will now also detect xen PVs
7765 * The pstore file system is now mounted by default, if it is
7768 * In addition to the SELinux and IMA policies we will now also
7769 load SMACK policies at early boot.
7771 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
7772 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
7773 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
7774 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
7775 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
7776 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
7777 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
7778 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
7779 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
7780 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
7781 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
7782 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
7783 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
7784 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
7788 * Timer units now support calendar time events in addition to
7789 monotonic time events. That means you can now trigger a unit
7790 based on a calendar time specification such as "Thu,Fri
7791 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
7792 or fifth day of any month of the year 2013, given that it is
7793 a thursday or friday. This brings timer event support
7794 considerably closer to cron's capabilities. For details on
7795 the supported calendar time specification language see
7798 * udev now supports a number of different naming policies for
7799 network interfaces for predictable names, and a combination
7800 of these policies is now the default. Please see this wiki
7801 document for details:
7803 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
7805 * Auke Kok's bootchart implementation has been added to the
7806 systemd tree. It is an optional component that can graph the
7807 boot in quite some detail. It is one of the best bootchart
7808 implementations around and minimal in its code and
7811 * nss-myhostname has been integrated into the systemd source
7812 tree. nss-myhostname guarantees that the local hostname
7813 always stays resolvable via NSS. It has been a weak
7814 requirement of systemd-hostnamed since a long time, and
7815 since its code is actually trivial we decided to just
7816 include it in systemd's source tree. It can be turned off
7817 with a configure switch.
7819 * The read-ahead logic is now capable of properly detecting
7820 whether a btrfs file system is on SSD or rotating media, in
7821 order to optimize the read-ahead scheme. Previously, it was
7822 only capable of detecting this on traditional file systems
7825 * In udev, additional device properties are now read from the
7826 IAB in addition to the OUI database. Also, Bluetooth company
7827 identities are attached to the devices as well.
7829 * In service files %U may be used as specifier that is
7830 replaced by the configured user name of the service.
7832 * nspawn may now be invoked without a controlling TTY. This
7833 makes it suitable for invocation as its own service. This
7834 may be used to set up a simple containerized server system
7835 using only core OS tools.
7837 * systemd and nspawn can now accept socket file descriptors
7838 when they are started for socket activation. This enables
7839 implementation of socket activated nspawn
7840 containers. i.e. think about autospawning an entire OS image
7841 when the first SSH or HTTP connection is received. We expect
7842 that similar functionality will also be added to libvirt-lxc
7845 * journalctl will now suppress ANSI color codes when
7846 presenting log data.
7848 * systemctl will no longer show control group information for
7849 a unit if the control group is empty anyway.
7851 * logind can now automatically suspend/hibernate/shutdown the
7854 * /etc/machine-info and hostnamed now also expose the chassis
7855 type of the system. This can be used to determine whether
7856 the local system is a laptop, desktop, handset or
7857 tablet. This information may either be configured by the
7858 user/vendor or is automatically determined from ACPI and DMI
7859 information if possible.
7861 * A number of polkit actions are now bound together with "imply"
7862 rules. This should simplify creating UIs because many actions
7863 will now authenticate similar ones as well.
7865 * Unit files learnt a new condition ConditionACPower= which
7866 may be used to conditionalize a unit depending on whether an
7867 AC power source is connected or not, of whether the system
7868 is running on battery power.
7870 * systemctl gained a new "is-failed" verb that may be used in
7871 shell scripts and suchlike to check whether a specific unit
7872 is in the "failed" state.
7874 * The EnvironmentFile= setting in unit files now supports file
7875 globbing, and can hence be used to easily read a number of
7876 environment files at once.
7878 * systemd will no longer detect and recognize specific
7879 distributions. All distribution-specific #ifdeffery has been
7880 removed, systemd is now fully generic and
7881 distribution-agnostic. Effectively, not too much is lost as
7882 a lot of the code is still accessible via explicit configure
7883 switches. However, support for some distribution specific
7884 legacy configuration file formats has been dropped. We
7885 recommend distributions to simply adopt the configuration
7886 files everybody else uses now and convert the old
7887 configuration from packaging scripts. Most distributions
7888 already did that. If that's not possible or desirable,
7889 distributions are welcome to forward port the specific
7890 pieces of code locally from the git history.
7892 * When logging a message about a unit systemd will now always
7893 log the unit name in the message meta data.
7895 * localectl will now also discover system locale data that is
7896 not stored in locale archives, but directly unpacked.
7898 * logind will no longer unconditionally use framebuffer
7899 devices as seat masters, i.e. as devices that are required
7900 to be existing before a seat is considered preset. Instead,
7901 it will now look for all devices that are tagged as
7902 "seat-master" in udev. By default, framebuffer devices will
7903 be marked as such, but depending on local systems, other
7904 devices might be marked as well. This may be used to
7905 integrate graphics cards using closed source drivers (such
7906 as NVidia ones) more nicely into logind. Note however, that
7907 we recommend using the open source NVidia drivers instead,
7908 and no udev rules for the closed-source drivers will be
7909 shipped from us upstream.
7911 Contributions from: Adam Williamson, Alessandro Crismani, Auke
7912 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
7913 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
7914 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
7915 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
7916 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
7917 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
7918 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
7919 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
7920 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
7921 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
7922 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
7927 * udev gained support for loading additional device properties
7928 from an indexed database that is keyed by vendor/product IDs
7929 and similar device identifiers. For the beginning this
7930 "hwdb" is populated with data from the well-known PCI and
7931 USB database, but also includes PNP, ACPI and OID data. In
7932 the longer run this indexed database shall grow into
7933 becoming the one central database for non-essential
7934 userspace device metadata. Previously, data from the PCI/USB
7935 database was only attached to select devices, since the
7936 lookup was a relatively expensive operation due to O(n) time
7937 complexity (with n being the number of entries in the
7938 database). Since this is now O(1), we decided to add in this
7939 data for all devices where this is available, by
7940 default. Note that the indexed database needs to be rebuilt
7941 when new data files are installed. To achieve this you need
7942 to update your packaging scripts to invoke "udevadm hwdb
7943 --update" after installation of hwdb data files. For
7944 RPM-based distributions we introduced the new
7945 %udev_hwdb_update macro for this purpose.
7947 * The Journal gained support for the "Message Catalog", an
7948 indexed database to link up additional information with
7949 journal entries. For further details please check:
7951 https://www.freedesktop.org/wiki/Software/systemd/catalog
7953 The indexed message catalog database also needs to be
7954 rebuilt after installation of message catalog files. Use
7955 "journalctl --update-catalog" for this. For RPM-based
7956 distributions we introduced the %journal_catalog_update
7957 macro for this purpose.
7959 * The Python Journal bindings gained support for the standard
7960 Python logging framework.
7962 * The Journal API gained new functions for checking whether
7963 the underlying file system of a journal file is capable of
7964 properly reporting file change notifications, or whether
7965 applications that want to reflect journal changes "live"
7966 need to recheck journal files continuously in appropriate
7969 * It is now possible to set the "age" field for tmpfiles
7970 entries to 0, indicating that files matching this entry
7971 shall always be removed when the directories are cleaned up.
7973 * coredumpctl gained a new "gdb" verb which invokes gdb
7974 right-away on the selected coredump.
7976 * There's now support for "hybrid sleep" on kernels that
7977 support this, in addition to "suspend" and "hibernate". Use
7978 "systemctl hybrid-sleep" to make use of this.
7980 * logind's HandleSuspendKey= setting (and related settings)
7981 now gained support for a new "lock" setting to simply
7982 request the screen lock on all local sessions, instead of
7983 actually executing a suspend or hibernation.
7985 * systemd will now mount the EFI variables file system by
7988 * Socket units now gained support for configuration of the
7989 SMACK security label.
7991 * timedatectl will now output the time of the last and next
7992 daylight saving change.
7994 * We dropped support for various legacy and distro-specific
7995 concepts, such as insserv, early-boot SysV services
7996 (i.e. those for non-standard runlevels such as 'b' or 'S')
7997 or ArchLinux /etc/rc.conf support. We recommend the
7998 distributions who still need support this to either continue
7999 to maintain the necessary patches downstream, or find a
8000 different solution. (Talk to us if you have questions!)
8002 * Various systemd components will now bypass polkit checks for
8003 root and otherwise handle properly if polkit is not found to
8004 be around. This should fix most issues for polkit-less
8005 systems. Quite frankly this should have been this way since
8006 day one. It is absolutely our intention to make systemd work
8007 fine on polkit-less systems, and we consider it a bug if
8008 something does not work as it should if polkit is not around.
8010 * For embedded systems it is now possible to build udev and
8011 systemd without blkid and/or kmod support.
8013 * "systemctl switch-root" is now capable of switching root
8014 more than once. I.e. in addition to transitions from the
8015 initrd to the host OS it is now possible to transition to
8016 further OS images from the host. This is useful to implement
8017 offline updating tools.
8019 * Various other additions have been made to the RPM macros
8020 shipped with systemd. Use %udev_rules_update() after
8021 installing new udev rules files. %_udevhwdbdir,
8022 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
8023 %_sysctldir are now available which resolve to the right
8024 directories for packages to place various data files in.
8026 * journalctl gained the new --full switch (in addition to
8027 --all, to disable ellipsation for long messages.
8029 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
8030 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
8031 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
8032 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
8033 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
8034 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
8035 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
8036 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
8037 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
8041 * journalctl gained new --since= and --until= switches to
8042 filter by time. It also now supports nice filtering for
8043 units via --unit=/-u.
8045 * Type=oneshot services may use ExecReload= and do the
8048 * The journal daemon now supports time-based rotation and
8049 vacuuming, in addition to the usual disk-space based
8052 * The journal will now index the available field values for
8053 each field name. This enables clients to show pretty drop
8054 downs of available match values when filtering. The bash
8055 completion of journalctl has been updated
8056 accordingly. journalctl gained a new switch -F to list all
8057 values a certain field takes in the journal database.
8059 * More service events are now written as structured messages
8060 to the journal, and made recognizable via message IDs.
8062 * The timedated, localed and hostnamed mini-services which
8063 previously only provided support for changing time, locale
8064 and hostname settings from graphical DEs such as GNOME now
8065 also have a minimal (but very useful) text-based client
8066 utility each. This is probably the nicest way to changing
8067 these settings from the command line now, especially since
8068 it lists available options and is fully integrated with bash
8071 * There's now a new tool "systemd-coredumpctl" to list and
8072 extract coredumps from the journal.
8074 * We now install a README each in /var/log/ and
8075 /etc/rc.d/init.d explaining where the system logs and init
8076 scripts went. This hopefully should help folks who go to
8077 that dirs and look into the otherwise now empty void and
8078 scratch their heads.
8080 * When user-services are invoked (by systemd --user) the
8081 $MANAGERPID env var is set to the PID of systemd.
8083 * SIGRTMIN+24 when sent to a --user instance will now result
8084 in immediate termination of systemd.
8086 * gatewayd received numerous feature additions such as a
8087 "follow" mode, for live syncing and filtering.
8089 * browse.html now allows filtering and showing detailed
8090 information on specific entries. Keyboard navigation and
8091 mouse screen support has been added.
8093 * gatewayd/journalctl now supports HTML5/JSON
8094 Server-Sent-Events as output.
8096 * The SysV init script compatibility logic will now
8097 heuristically determine whether a script supports the
8098 "reload" verb, and only then make this available as
8101 * "systemctl status --follow" has been removed, use "journalctl
8104 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
8105 have been removed since they are hardly useful to be
8108 * And I'd like to take the opportunity to specifically mention
8109 Zbigniew for his great contributions. Zbigniew, you rock!
8111 Contributions from: Andrew Eikum, Christian Hesse, Colin
8112 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
8113 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
8114 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
8115 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
8116 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
8117 Jędrzejewski-Szmek, Сковорода Никита Андреевич
8121 * If /etc/vconsole.conf is non-existent or empty we will no
8122 longer load any console font or key map at boot by
8123 default. Instead the kernel defaults will be left
8124 intact. This is definitely the right thing to do, as no
8125 configuration should mean no configuration, and hard-coding
8126 font names that are different on all archs is probably a bad
8127 idea. Also, the kernel default key map and font should be
8128 good enough for most cases anyway, and mostly identical to
8129 the userspace fonts/key maps we previously overloaded them
8130 with. If distributions want to continue to default to a
8131 non-kernel font or key map they should ship a default
8132 /etc/vconsole.conf with the appropriate contents.
8134 Contributions from: Colin Walters, Daniel J Walsh, Dave
8135 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
8136 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
8140 * journalctl gained a new --cursor= switch to show entries
8141 starting from the specified location in the journal.
8143 * We now enforce a size limit on journal entry fields exported
8144 with "-o json" in journalctl. Fields larger than 4K will be
8145 assigned null. This can be turned off with --all.
8147 * An (optional) journal gateway daemon is now available as
8148 "systemd-journal-gatewayd.service". This service provides
8149 access to the journal via HTTP and JSON. This functionality
8150 will be used to implement live log synchronization in both
8151 pull and push modes, but has various other users too, such
8152 as easy log access for debugging of embedded devices. Right
8153 now it is already useful to retrieve the journal via HTTP:
8155 # systemctl start systemd-journal-gatewayd.service
8156 # wget http://localhost:19531/entries
8158 This will download the journal contents in a
8159 /var/log/messages compatible format. The same as JSON:
8161 # curl -H"Accept: application/json" http://localhost:19531/entries
8163 This service is also accessible via a web browser where a
8164 single static HTML5 app is served that uses the JSON logic
8165 to enable the user to do some basic browsing of the
8166 journal. This will be extended later on. Here's an example
8167 screenshot of this app in its current state:
8169 http://0pointer.de/public/journal-gatewayd
8171 Contributions from: Kay Sievers, Lennart Poettering, Robert
8172 Milasan, Tom Gundersen
8176 * The bash completion logic is now available for journalctl
8179 * We do not mount the "cpuset" controller anymore together with
8180 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
8181 started if no parameters are assigned to it. "cpuset" hence
8182 broke code that assumed it could create "cpu" groups and
8185 * journalctl -f will now subscribe to terminal size changes,
8186 and line break accordingly.
8188 Contributions from: Dave Reisner, Kay Sievers, Lennart
8189 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
8193 * nspawn will now create a symlink /etc/localtime in the
8194 container environment, copying the host's timezone
8195 setting. Previously this has been done via a bind mount, but
8196 since symlinks cannot be bind mounted this has now been
8197 changed to create/update the appropriate symlink.
8199 * journalctl -n's line number argument is now optional, and
8200 will default to 10 if omitted.
8202 * journald will now log the maximum size the journal files may
8203 take up on disk. This is particularly useful if the default
8204 built-in logic of determining this parameter from the file
8205 system size is used. Use "systemctl status
8206 systemd-journald.service" to see this information.
8208 * The multi-seat X wrapper tool has been stripped down. As X
8209 is now capable of enumerating graphics devices via udev in a
8210 seat-aware way the wrapper is not strictly necessary
8211 anymore. A stripped down temporary stop-gap is still shipped
8212 until the upstream display managers have been updated to
8213 fully support the new X logic. Expect this wrapper to be
8214 removed entirely in one of the next releases.
8216 * HandleSleepKey= in logind.conf has been split up into
8217 HandleSuspendKey= and HandleHibernateKey=. The old setting
8218 is not available anymore. X11 and the kernel are
8219 distinguishing between these keys and we should too. This
8220 also means the inhibition lock for these keys has been split
8223 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
8224 Poettering, Lukas Nykryn, Václav Pavlín
8228 * Whenever a unit changes state we will now log this to the
8229 journal and show along the unit's own log output in
8232 * ConditionPathIsMountPoint= can now properly detect bind
8233 mount points too. (Previously, a bind mount of one file
8234 system to another place in the same file system could not be
8235 detected as mount, since they shared struct stat's st_dev
8238 * We will now mount the cgroup controllers cpu, cpuacct,
8239 cpuset and the controllers net_cls, net_prio together by
8242 * nspawn containers will now have a virtualized boot
8243 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
8244 over with a randomized ID at container initialization). This
8245 has the effect of making "journalctl -b" do the right thing
8248 * The JSON output journal serialization has been updated not
8249 to generate "endless" list objects anymore, but rather one
8250 JSON object per line. This is more in line how most JSON
8251 parsers expect JSON objects. The new output mode
8252 "json-pretty" has been added to provide similar output, but
8253 neatly aligned for readability by humans.
8255 * We dropped all explicit sync() invocations in the shutdown
8256 code. The kernel does this implicitly anyway in the kernel
8257 reboot() syscall. halt(8)'s -n option is now a compatibility
8260 * We now support virtualized reboot() in containers, as
8261 supported by newer kernels. We will fall back to exit() if
8262 CAP_SYS_REBOOT is not available to the container. Also,
8263 nspawn makes use of this now and will actually reboot the
8264 container if the containerized OS asks for that.
8266 * journalctl will only show local log output by default
8267 now. Use --merge (-m) to show remote log output, too.
8269 * libsystemd-journal gained the new sd_journal_get_usage()
8270 call to determine the current disk usage of all journal
8271 files. This is exposed in the new "journalctl --disk-usage"
8274 * journald gained a new configuration setting SplitMode= in
8275 journald.conf which may be used to control how user journals
8276 are split off. See journald.conf(5) for details.
8278 * A new condition type ConditionFileNotEmpty= has been added.
8280 * tmpfiles' "w" lines now support file globbing, to write
8281 multiple files at once.
8283 * We added Python bindings for the journal submission
8284 APIs. More Python APIs for a number of selected APIs will
8285 likely follow. Note that we intend to add native bindings
8286 only for the Python language, as we consider it common
8287 enough to deserve bindings shipped within systemd. There are
8288 various projects outside of systemd that provide bindings
8289 for languages such as PHP or Lua.
8291 * Many conditions will now resolve specifiers such as %i. In
8292 addition, PathChanged= and related directives of .path units
8293 now support specifiers as well.
8295 * There's now a new RPM macro definition for the system preset
8298 * journald will now warn if it ca not forward a message to the
8299 syslog daemon because its socket is full.
8301 * timedated will no longer write or process /etc/timezone,
8302 except on Debian. As we do not support late mounted /usr
8303 anymore /etc/localtime always being a symlink is now safe,
8304 and hence the information in /etc/timezone is not necessary
8307 * logind will now always reserve one VT for a text getty (VT6
8308 by default). Previously if more than 6 X sessions where
8309 started they took up all the VTs with auto-spawned gettys,
8310 so that no text gettys were available anymore.
8312 * udev will now automatically inform the btrfs kernel logic
8313 about btrfs RAID components showing up. This should make
8314 simple hotplug based btrfs RAID assembly work.
8316 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
8317 (but not for its children which will stay at the kernel
8318 default). This should allow setups with a lot more listening
8321 * systemd will now always pass the configured timezone to the
8322 kernel at boot. timedated will do the same when the timezone
8325 * logind's inhibition logic has been updated. By default,
8326 logind will now handle the lid switch, the power and sleep
8327 keys all the time, even in graphical sessions. If DEs want
8328 to handle these events on their own they should take the new
8329 handle-power-key, handle-sleep-key and handle-lid-switch
8330 inhibitors during their runtime. A simple way to achieve
8331 that is to invoke the DE wrapped in an invocation of:
8333 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
8335 * Access to unit operations is now checked via SELinux taking
8336 the unit file label and client process label into account.
8338 * systemd will now notify the administrator in the journal
8339 when he over-mounts a non-empty directory.
8341 * There are new specifiers that are resolved in unit files,
8342 for the host name (%H), the machine ID (%m) and the boot ID
8345 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
8346 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
8347 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
8348 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
8349 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
8350 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
8351 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
8355 * Support for reading structured kernel messages from
8356 /dev/kmsg has now been added and is enabled by default.
8358 * Support for reading kernel messages from /proc/kmsg has now
8359 been removed. If you want kernel messages in the journal
8360 make sure to run a recent kernel (>= 3.5) that supports
8361 reading structured messages from /dev/kmsg (see
8362 above). /proc/kmsg is now exclusive property of classic
8363 syslog daemons again.
8365 * The libudev API gained the new
8366 udev_device_new_from_device_id() call.
8368 * The logic for file system namespace (ReadOnlyDirectory=,
8369 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
8370 require pivot_root() anymore. This means fewer temporary
8371 directories are created below /tmp for this feature.
8373 * nspawn containers will now see and receive all submounts
8374 made on the host OS below the root file system of the
8377 * Forward Secure Sealing is now supported for Journal files,
8378 which provide cryptographical sealing of journal files so
8379 that attackers cannot alter log history anymore without this
8380 being detectable. Lennart will soon post a blog story about
8381 this explaining it in more detail.
8383 * There are two new service settings RestartPreventExitStatus=
8384 and SuccessExitStatus= which allow configuration of exit
8385 status (exit code or signal) which will be excepted from the
8386 restart logic, resp. consider successful.
8388 * journalctl gained the new --verify switch that can be used
8389 to check the integrity of the structure of journal files and
8390 (if Forward Secure Sealing is enabled) the contents of
8393 * nspawn containers will now be run with /dev/stdin, /dev/fd/
8394 and similar symlinks pre-created. This makes running shells
8395 as container init process a lot more fun.
8397 * The fstab support can now handle PARTUUID= and PARTLABEL=
8400 * A new ConditionHost= condition has been added to match
8401 against the hostname (with globs) and machine ID. This is
8402 useful for clusters where a single OS image is used to
8403 provision a large number of hosts which shall run slightly
8404 different sets of services.
8406 * Services which hit the restart limit will now be placed in a
8409 Contributions from: Bertram Poettering, Dave Reisner, Huang
8410 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
8411 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
8415 * When running in --user mode systemd will now become a
8416 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
8417 tree a lot more organized.
8419 * A new PartOf= unit dependency type has been introduced that
8420 may be used to group services in a natural way.
8422 * "systemctl enable" may now be used to enable instances of
8425 * journalctl now prints error log levels in red, and
8426 warning/notice log levels in bright white. It also supports
8427 filtering by log level now.
8429 * cgtop gained a new -n switch (similar to top), to configure
8430 the maximum number of iterations to run for. It also gained
8431 -b, to run in batch mode (accepting no input).
8433 * The suffix ".service" may now be omitted on most systemctl
8434 command lines involving service unit names.
8436 * There's a new bus call in logind to lock all sessions, as
8437 well as a loginctl verb for it "lock-sessions".
8439 * libsystemd-logind.so gained a new call sd_journal_perror()
8440 that works similar to libc perror() but logs to the journal
8441 and encodes structured information about the error number.
8443 * /etc/crypttab entries now understand the new keyfile-size=
8446 * shutdown(8) now can send a (configurable) wall message when
8447 a shutdown is cancelled.
8449 * The mount propagation mode for the root file system will now
8450 default to "shared", which is useful to make containers work
8451 nicely out-of-the-box so that they receive new mounts from
8452 the host. This can be undone locally by running "mount
8453 --make-rprivate /" if needed.
8455 * The prefdm.service file has been removed. Distributions
8456 should maintain this unit downstream if they intend to keep
8457 it around. However, we recommend writing normal unit files
8458 for display managers instead.
8460 * Since systemd is a crucial part of the OS we will now
8461 default to a number of compiler switches that improve
8462 security (hardening) such as read-only relocations, stack
8463 protection, and suchlike.
8465 * The TimeoutSec= setting for services is now split into
8466 TimeoutStartSec= and TimeoutStopSec= to allow configuration
8467 of individual time outs for the start and the stop phase of
8470 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
8471 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
8472 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
8473 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
8474 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
8475 Gundersen, Zbigniew Jędrzejewski-Szmek
8479 * The journal and id128 C APIs are now fully documented as man
8482 * Extra safety checks have been added when transitioning from
8483 the initial RAM disk to the main system to avoid accidental
8486 * /etc/crypttab entries now understand the new keyfile-offset=
8489 * systemctl -t can now be used to filter by unit load state.
8491 * The journal C API gained the new sd_journal_wait() call to
8492 make writing synchronous journal clients easier.
8494 * journalctl gained the new -D switch to show journals from a
8497 * journalctl now displays a special marker between log
8498 messages of two different boots.
8500 * The journal is now explicitly flushed to /var via a service
8501 systemd-journal-flush.service, rather than implicitly simply
8502 by seeing /var/log/journal to be writable.
8504 * journalctl (and the journal C APIs) can now match for much
8505 more complex expressions, with alternatives and
8508 * When transitioning from the initial RAM disk to the main
8509 system we will now kill all processes in a killing spree to
8510 ensure no processes stay around by accident.
8512 * Three new specifiers may be used in unit files: %u, %h, %s
8513 resolve to the user name, user home directory resp. user
8514 shell. This is useful for running systemd user instances.
8516 * We now automatically rotate journal files if their data
8517 object hash table gets a fill level > 75%. We also size the
8518 hash table based on the configured maximum file size. This
8519 together should lower hash collisions drastically and thus
8520 speed things up a bit.
8522 * journalctl gained the new "--header" switch to introspect
8523 header data of journal files.
8525 * A new setting SystemCallFilters= has been added to services
8526 which may be used to apply blacklists or whitelists to
8527 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
8529 * nspawn gained a new --link-journal= switch (and quicker: -j)
8530 to link the container journal with the host. This makes it
8531 very easy to centralize log viewing on the host for all
8532 guests while still keeping the journal files separated.
8534 * Many bugfixes and optimizations
8536 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
8537 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
8538 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
8543 * Several tools now understand kernel command line arguments,
8544 which are only read when run in an initial RAM disk. They
8545 usually follow closely their normal counterparts, but are
8548 * There's a new tool to analyze the readahead files that are
8549 automatically generated at boot. Use:
8551 /usr/lib/systemd/systemd-readahead analyze /.readahead
8553 * We now provide an early debug shell on tty9 if this enabled. Use:
8555 systemctl enable debug-shell.service
8557 * All plymouth related units have been moved into the Plymouth
8558 package. Please make sure to upgrade your Plymouth version
8561 * systemd-tmpfiles now supports getting passed the basename of
8562 a configuration file only, in which case it will look for it
8563 in all appropriate directories automatically.
8565 * udevadm info now takes a /dev or /sys path as argument, and
8566 does the right thing. Example:
8568 udevadm info /dev/sda
8569 udevadm info /sys/class/block/sda
8571 * systemctl now prints a warning if a unit is stopped but a
8572 unit that might trigger it continues to run. Example: a
8573 service is stopped but the socket that activates it is left
8576 * "systemctl status" will now mention if the log output was
8577 shortened due to rotation since a service has been started.
8579 * The journal API now exposes functions to determine the
8580 "cutoff" times due to rotation.
8582 * journald now understands SIGUSR1 and SIGUSR2 for triggering
8583 immediately flushing of runtime logs to /var if possible,
8584 resp. for triggering immediate rotation of the journal
8587 * It is now considered an error if a service is attempted to
8588 be stopped that is not loaded.
8590 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
8592 * systemd-analyze now supports Python 3
8594 * tmpfiles now supports cleaning up directories via aging
8595 where the first level dirs are always kept around but
8596 directories beneath it automatically aged. This is enabled
8597 by prefixing the age field with '~'.
8599 * Seat objects now expose CanGraphical, CanTTY properties
8600 which is required to deal with very fast bootups where the
8601 display manager might be running before the graphics drivers
8602 completed initialization.
8604 * Seat objects now expose a State property.
8606 * We now include RPM macros for service enabling/disabling
8607 based on the preset logic. We recommend RPM based
8608 distributions to make use of these macros if possible. This
8609 makes it simpler to reuse RPM spec files across
8612 * We now make sure that the collected systemd unit name is
8613 always valid when services log to the journal via
8616 * There's a new man page kernel-command-line(7) detailing all
8617 command line options we understand.
8619 * The fstab generator may now be disabled at boot by passing
8620 fstab=0 on the kernel command line.
8622 * A new kernel command line option modules-load= is now understood
8623 to load a specific kernel module statically, early at boot.
8625 * Unit names specified on the systemctl command line are now
8626 automatically escaped as needed. Also, if file system or
8627 device paths are specified they are automatically turned
8628 into the appropriate mount or device unit names. Example:
8630 systemctl status /home
8631 systemctl status /dev/sda
8633 * The SysVConsole= configuration option has been removed from
8634 system.conf parsing.
8636 * The SysV search path is no longer exported on the D-Bus
8639 * The Names= option has been removed from unit file parsing.
8641 * There's a new man page bootup(7) detailing the boot process.
8643 * Every unit and every generator we ship with systemd now
8644 comes with full documentation. The self-explanatory boot is
8647 * A couple of services gained "systemd-" prefixes in their
8648 name if they wrap systemd code, rather than only external
8649 code. Among them fsck@.service which is now
8650 systemd-fsck@.service.
8652 * The HaveWatchdog property has been removed from the D-Bus
8655 * systemd.confirm_spawn= on the kernel command line should now
8658 * There's a new man page crypttab(5) which details all options
8659 we actually understand.
8661 * systemd-nspawn gained a new --capability= switch to pass
8662 additional capabilities to the container.
8664 * timedated will now read known NTP implementation unit names
8665 from /usr/lib/systemd/ntp-units.d/*.list,
8666 systemd-timedated-ntp.target has been removed.
8668 * journalctl gained a new switch "-b" that lists log data of
8669 the current boot only.
8671 * The notify socket is in the abstract namespace again, in
8672 order to support daemons which chroot() at start-up.
8674 * There is a new Storage= configuration option for journald
8675 which allows configuration of where log data should go. This
8676 also provides a way to disable journal logging entirely, so
8677 that data collected is only forwarded to the console, the
8678 kernel log buffer or another syslog implementation.
8680 * Many bugfixes and optimizations
8682 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
8683 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
8684 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
8685 Shawn Landden, Tom Gundersen
8689 * "systemctl help <unit>" now shows the man page if one is
8692 * Several new man pages have been added.
8694 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
8695 MaxLevelConsole= can now be specified in
8696 journald.conf. These options allow reducing the amount of
8697 data stored on disk or forwarded by the log level.
8699 * TimerSlackNSec= can now be specified in system.conf for
8700 PID1. This allows system-wide power savings.
8702 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
8703 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
8708 * logind is now capable of (optionally) handling power and
8709 sleep keys as well as the lid switch.
8711 * journalctl now understands the syntax "journalctl
8712 /usr/bin/avahi-daemon" to get all log output of a specific
8715 * CapabilityBoundingSet= in system.conf now also influences
8716 the capability bound set of usermode helpers of the kernel.
8718 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
8719 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
8720 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
8721 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
8725 * Note that we skipped 139 releases here in order to set the
8726 new version to something that is greater than both udev's
8727 and systemd's most recent version number.
8729 * udev: all udev sources are merged into the systemd source tree now.
8730 All future udev development will happen in the systemd tree. It
8731 is still fully supported to use the udev daemon and tools without
8732 systemd running, like in initramfs or other init systems. Building
8733 udev though, will require the *build* of the systemd tree, but
8734 udev can be properly *run* without systemd.
8736 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
8737 should be used to create dead device nodes as workarounds for broken
8740 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
8741 no longer supported. udev_monitor_new_from_netlink() needs to be
8742 used to subscribe to events.
8744 * udev: when udevd is started by systemd, processes which are left
8745 behind by forking them off of udev rules, are unconditionally cleaned
8746 up and killed now after the event handling has finished. Services or
8747 daemons must be started as systemd services. Services can be
8748 pulled-in by udev to get started, but they can no longer be directly
8749 forked by udev rules.
8751 * udev: the daemon binary is called systemd-udevd now and installed
8752 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
8753 to adapt to that, create symlink, or rename the binary after building
8756 * libudev no longer provides these symbols:
8757 udev_monitor_from_socket()
8758 udev_queue_get_failed_list_entry()
8759 udev_get_{dev,sys,run}_path()
8760 The versions number was bumped and symbol versioning introduced.
8762 * systemd-loginctl and systemd-journalctl have been renamed
8763 to loginctl and journalctl to match systemctl.
8765 * The config files: /etc/systemd/systemd-logind.conf and
8766 /etc/systemd/systemd-journald.conf have been renamed to
8767 logind.conf and journald.conf. Package updates should rename
8768 the files to the new names on upgrade.
8770 * For almost all files the license is now LGPL2.1+, changed
8771 from the previous GPL2.0+. Exceptions are some minor stuff
8772 of udev (which will be changed to LGPL2.1 eventually, too),
8773 and the MIT licensed sd-daemon.[ch] library that is suitable
8774 to be used as drop-in files.
8776 * systemd and logind now handle system sleep states, in
8777 particular suspending and hibernating.
8779 * logind now implements a sleep/shutdown/idle inhibiting logic
8780 suitable for a variety of uses. Soonishly Lennart will blog
8781 about this in more detail.
8783 * var-run.mount and var-lock.mount are no longer provided
8784 (which previously bind mounted these directories to their new
8785 places). Distributions which have not converted these
8786 directories to symlinks should consider stealing these files
8787 from git history and add them downstream.
8789 * We introduced the Documentation= field for units and added
8790 this to all our shipped units. This is useful to make it
8791 easier to explore the boot and the purpose of the various
8794 * All smaller setup units (such as
8795 systemd-vconsole-setup.service) now detect properly if they
8796 are run in a container and are skipped when
8797 appropriate. This guarantees an entirely noise-free boot in
8798 Linux container environments such as systemd-nspawn.
8800 * A framework for implementing offline system updates is now
8801 integrated, for details see:
8802 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
8804 * A new service type Type=idle is available now which helps us
8805 avoiding ugly interleaving of getty output and boot status
8808 * There's now a system-wide CapabilityBoundingSet= option to
8809 globally reduce the set of capabilities for the
8810 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
8811 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
8812 even CAP_NET_ADMIN system-wide for secure systems.
8814 * There are now system-wide DefaultLimitXXX= options to
8815 globally change the defaults of the various resource limits
8816 for all units started by PID 1.
8818 * Harald Hoyer's systemd test suite has been integrated into
8819 systemd which allows easy testing of systemd builds in qemu
8820 and nspawn. (This is really awesome! Ask us for details!)
8822 * The fstab parser is now implemented as generator, not inside
8825 * systemctl will now warn you if .mount units generated from
8826 /etc/fstab are out of date due to changes in fstab that
8827 have not been read by systemd yet.
8829 * systemd is now suitable for usage in initrds. Dracut has
8830 already been updated to make use of this. With this in place
8831 initrds get a slight bit faster but primarily are much
8832 easier to introspect and debug since "systemctl status" in
8833 the host system can be used to introspect initrd services,
8834 and the journal from the initrd is kept around too.
8836 * systemd-delta has been added, a tool to explore differences
8837 between user/admin configuration and vendor defaults.
8839 * PrivateTmp= now affects both /tmp and /var/tmp.
8841 * Boot time status messages are now much prettier and feature
8842 proper english language. Booting up systemd has never been
8845 * Read-ahead pack files now include the inode number of all
8846 files to pre-cache. When the inode changes the pre-caching
8847 is not attempted. This should be nicer to deal with updated
8848 packages which might result in changes of read-ahead
8851 * We now temporaritly lower the kernel's read_ahead_kb variable
8852 when collecting read-ahead data to ensure the kernel's
8853 built-in read-ahead does not add noise to our measurements
8854 of necessary blocks to pre-cache.
8856 * There's now RequiresMountsFor= to add automatic dependencies
8857 for all mounts necessary for a specific file system path.
8859 * MountAuto= and SwapAuto= have been removed from
8860 system.conf. Mounting file systems at boot has to take place
8863 * nspawn now learned a new switch --uuid= to set the machine
8864 ID on the command line.
8866 * nspawn now learned the -b switch to automatically search
8869 * vt102 is now the default TERM for serial TTYs, upgraded from
8872 * systemd-logind now works on VT-less systems.
8874 * The build tree has been reorganized. The individual
8875 components now have directories of their own.
8877 * A new condition type ConditionPathIsReadWrite= is now available.
8879 * nspawn learned the new -C switch to create cgroups for the
8880 container in other hierarchies.
8882 * We now have support for hardware watchdogs, configurable in
8885 * The scheduled shutdown logic now has a public API.
8887 * We now mount /tmp as tmpfs by default, but this can be
8888 masked and /etc/fstab can override it.
8890 * Since udisks does not make use of /media anymore we are not
8891 mounting a tmpfs on it anymore.
8893 * journalctl gained a new --local switch to only interleave
8894 locally generated journal files.
8896 * We can now load the IMA policy at boot automatically.
8898 * The GTK tools have been split off into a systemd-ui.
8900 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
8901 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
8902 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
8903 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
8904 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
8905 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
8906 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
8907 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
8908 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
8913 * This is mostly a bugfix release
8915 * Support optional initialization of the machine ID from the
8916 KVM or container configured UUID.
8918 * Support immediate reboots with "systemctl reboot -ff"
8920 * Show /etc/os-release data in systemd-analyze output
8922 * Many bugfixes for the journal, including endianness fixes and
8923 ensuring that disk space enforcement works
8925 * sd-login.h is C++ compatible again
8927 * Extend the /etc/os-release format on request of the Debian
8930 * We now refuse non-UTF8 strings used in various configuration
8931 and unit files. This is done to ensure we do not pass invalid
8932 data over D-Bus or expose it elsewhere.
8934 * Register Mimo USB Screens as suitable for automatic seat
8937 * Read SELinux client context from journal clients in a race
8940 * Reorder configuration file lookup order. /etc now always
8941 overrides /run in order to allow the administrator to always
8942 and unconditionally override vendor-supplied or
8943 automatically generated data.
8945 * The various user visible bits of the journal now have man
8946 pages. We still lack man pages for the journal API calls
8949 * We now ship all man pages in HTML format again in the
8952 Contributions from: Dave Reisner, Dirk Eibach, Frederic
8953 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
8954 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
8959 * This is mostly a bugfix release
8961 * systems lacking /etc/os-release are no longer supported.
8963 * Various functionality updates to libsystemd-login.so
8965 * Track class of PAM logins to distinguish greeters from
8968 Contributions from: Kay Sievers, Lennart Poettering, Michael
8973 * This is an important bugfix release for v41.
8975 * Building man pages is now optional which should be useful
8976 for those building systemd from git but unwilling to install
8979 * Watchdog support for supervising services is now usable. In
8980 a future release support for hardware watchdogs
8981 (i.e. /dev/watchdog) will be added building on this.
8983 * Service start rate limiting is now configurable and can be
8984 turned off per service. When a start rate limit is hit a
8985 reboot can automatically be triggered.
8987 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
8989 Contributions from: Benjamin Franzke, Bill Nottingham,
8990 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
8991 Schmidt, Michał Górny, Piotr Drąg
8995 * The systemd binary is installed /usr/lib/systemd/systemd now;
8996 An existing /sbin/init symlink needs to be adapted with the
8999 * The code that loads kernel modules has been ported to invoke
9000 libkmod directly, instead of modprobe. This means we do not
9001 support systems with module-init-tools anymore.
9003 * Watchdog support is now already useful, but still not
9006 * A new kernel command line option systemd.setenv= is
9007 understood to set system wide environment variables
9008 dynamically at boot.
9010 * We now limit the set of capabilities of systemd-journald.
9012 * We now set SIGPIPE to ignore by default, since it only is
9013 useful in shell pipelines, and has little use in general
9014 code. This can be disabled with IgnoreSIPIPE=no in unit
9017 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
9018 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
9023 * This is mostly a bugfix release
9025 * We now expose the reason why a service failed in the
9026 "Result" D-Bus property.
9028 * Rudimentary service watchdog support (will be completed over
9029 the next few releases.)
9031 * When systemd forks off in order execute some service we will
9032 now immediately changes its argv[0] to reflect which process
9033 it will execute. This is useful to minimize the time window
9034 with a generic argv[0], which makes bootcharts more useful
9036 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
9037 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
9038 Mike Kazantsev, Ray Strode
9042 * This is mostly a test release, but incorporates many
9045 * New systemd-cgtop tool to show control groups by their
9048 * Linking against libacl for ACLs is optional again. If
9049 disabled, support tracking device access for active logins
9050 goes becomes unavailable, and so does access to the user
9051 journals by the respective users.
9053 * If a group "adm" exists, journal files are automatically
9054 owned by them, thus allow members of this group full access
9055 to the system journal as well as all user journals.
9057 * The journal now stores the SELinux context of the logging
9058 client for all entries.
9060 * Add C++ inclusion guards to all public headers
9062 * New output mode "cat" in the journal to print only text
9063 messages, without any meta data like date or time.
9065 * Include tiny X server wrapper as a temporary stop-gap to
9066 teach XOrg udev display enumeration. This is used by display
9067 managers such as gdm, and will go away as soon as XOrg
9068 learned native udev hotplugging for display devices.
9070 * Add new systemd-cat tool for executing arbitrary programs
9071 with STDERR/STDOUT connected to the journal. Can also act as
9072 BSD logger replacement, and does so by default.
9074 * Optionally store all locally generated coredumps in the
9075 journal along with meta data.
9077 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
9078 writing short strings to files (for usage for /sys), and for
9079 creating symlinks, character and block device nodes.
9081 * New unit file option ControlGroupPersistent= to make cgroups
9082 persistent, following the mechanisms outlined in
9083 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
9085 * Support multiple local RTCs in a sane way
9087 * No longer monopolize IO when replaying readahead data on
9088 rotating disks, since we might starve non-file-system IO to
9089 death, since fanotify() will not see accesses done by blkid,
9092 * Do not show kernel threads in systemd-cgls anymore, unless
9093 requested with new -k switch.
9095 Contributions from: Dan Horák, Kay Sievers, Lennart
9096 Poettering, Michal Schmidt
9100 * This is mostly a test release, but incorporates many
9103 * The git repository moved to:
9104 git://anongit.freedesktop.org/systemd/systemd
9105 ssh://git.freedesktop.org/git/systemd/systemd
9107 * First release with the journal
9108 http://0pointer.de/blog/projects/the-journal.html
9110 * The journal replaces both systemd-kmsg-syslogd and
9111 systemd-stdout-bridge.
9113 * New sd_pid_get_unit() API call in libsystemd-logind
9115 * Many systemadm clean-ups
9117 * Introduce remote-fs-pre.target which is ordered before all
9118 remote mounts and may be used to start services before all
9121 * Added Mageia support
9123 * Add bash completion for systemd-loginctl
9125 * Actively monitor PID file creation for daemons which exit in
9126 the parent process before having finished writing the PID
9127 file in the daemon process. Daemons which do this need to be
9128 fixed (i.e. PID file creation must have finished before the
9129 parent exits), but we now react a bit more gracefully to them.
9131 * Add colourful boot output, mimicking the well-known output
9132 of existing distributions.
9134 * New option PassCredentials= for socket units, for
9135 compatibility with a recent kernel ABI breakage.
9137 * /etc/rc.local is now hooked in via a generator binary, and
9138 thus will no longer act as synchronization point during
9141 * systemctl list-unit-files now supports --root=.
9143 * systemd-tmpfiles now understands two new commands: z, Z for
9144 relabelling files according to the SELinux database. This is
9145 useful to apply SELinux labels to specific files in /sys,
9148 * Output of SysV services is now forwarded to both the console
9149 and the journal by default, not only just the console.
9151 * New man pages for all APIs from libsystemd-login.
9153 * The build tree got reorganized and the build system is a
9154 lot more modular allowing embedded setups to specifically
9155 select the components of systemd they are interested in.
9157 * Support for Linux systems lacking the kernel VT subsystem is
9160 * configure's --with-rootdir= got renamed to
9161 --with-rootprefix= to follow the naming used by udev and
9164 * Unless specified otherwise we will now install to /usr instead
9165 of /usr/local by default.
9167 * Processes with '@' in argv[0][0] are now excluded from the
9168 final shut-down killing spree, following the logic explained
9170 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
9172 * All processes remaining in a service cgroup when we enter
9173 the START or START_PRE states are now killed with
9174 SIGKILL. That means it is no longer possible to spawn
9175 background processes from ExecStart= lines (which was never
9176 supported anyway, and bad style).
9178 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
9179 reloading of units together.
9181 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
9182 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
9183 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
9184 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
9185 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek