1 systemd System and Service Manager
5 * A new modprobe.d drop-in is now shipped by default that sets the
6 bonding module option max_bonds=0. This overrides the kernel default,
7 to avoid conflicts and ambiguity as to whether or not bond0 should be
8 managed by systemd-networkd or not. This resolves multiple issues
9 with bond0 properties not being applied, when bond0 is configured
10 with systemd-networkd. Distributors may choose to not package this,
11 however in that case users will be prevented from correctly managing
12 bond0 interface using systemd-networkd.
14 * systemd-analyze gained new verbs "get-log-level" and "get-log-target"
15 which print the logging level and target of the system manager,
16 respectively. They complement the existing "set-log-level" and
17 "set-log-target" verbs, which can be used to change those values.
19 * journald.conf gained a new boolean setting ReadKMsg= which defaults
20 to on. If turned off kernel log messages will not be read by
21 systemd-journald and not be included in the logs. It also gained a
22 new setting LineMax= for configuring the maximum line length to allow
23 when converting STDOUT/STDERR log streams into individual log
24 records. The new default for this value is 48K, up from the previous
27 * A new setting RuntimeDirectoryPreserve= for units has been added,
28 which allows more detailed control of what to do with a runtime
29 directory configured with RuntimeDirectory= (i.e. a directory below
30 /run or $XDG_RUNTIME_DIR) after a unit is stopped.
32 * The RuntimeDirectory= setting for units gained support for creating
33 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
34 one top-level directory.
36 * Units gained new options StateDirectory=, CacheDirectory=,
37 LogsDirectory= and ConfigurationDirectory= which are closely related
38 to RuntimeDirectory= but manage per-service directories below
39 /var/lib, /var/cache, /var/log and /etc. By making use of this it is
40 possible to write unit files which when activated automatically gain
41 properly owned service specific directories in these locations, thus
42 making unit files self-contained and increasing compatibility with
43 stateless systems and factory reset where /etc or /var are
44 unpopulated at boot. Matching these new settings there's also
45 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
46 ConfigurationDirectoryMode= for configuring the access mode of these
49 * Automake support has been removed from this release. systemd is now
52 * systemd-journald will now aggressively cache client metadata during
53 runtime, speeding up log write performance under pressure. This comes
54 at a small price though: as much of the metadata is read
55 asynchronously from /proc/ (and isn't implicitly attached to log
56 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
57 metadata stored alongside a log entry might be slightly
58 out-of-date. Previously it could only be slightly newer than the log
59 message. The time window is small however, and given that the kernel
60 is unlikely to be improved anytime soon in this regard, this appears
63 * nss-myhostname/systemd-resolved will now by default synthesize an
64 A/AAAA resource record for the "_gateway" hostname, pointing to the
65 current default IP gateway. Previously it did that for the "gateway"
66 name, hampering adoption, as some distributions wanted to leave that
67 host name open for local use. The old behaviour may still be
68 requested at build time.
70 * systemd-networkd's [Address] section in .network files gained a new
71 Scope= setting for configuring the IP address scope. The [Network]
72 section gained a new boolean setting ConfigureWithoutCarrier= that
73 tells systemd-networkd to ignore link sensing when configuring the
74 device. The [DHCP] section gained a new Anonymize= boolean option for
75 turning on a number of options suggested in RFC 7844. A new
76 [RoutingPolicyRule] section has been added for configuring the IP
77 routing policy. The [Route] section has gained support for a new
78 Type= setting which permits configuring
79 blackhole/unreachable/prohibit routes.
81 * The [VRF] section in .netdev files gained a new Table= setting for
82 configuring the routing table to use. The [Tunnel] section gained a
83 new Independent= boolean field for configuring tunnels independent of
84 an underlying network interface. The [Bridge] section gained a new
85 GroupForwardMask= option for configuration of propagation of link
86 local frames between bridge ports.
88 * The WakeOnLan= setting in .link files gained support for a number of
89 new modes. A new TCP6SegmentationOffload= setting has been added for
90 configuring TCP/IPv6 hardware segmentation offload.
92 * The IPv6 RA sender implementation may now optionally send out RDNSS
93 and RDNSSL records for supplying DNS configuration to peers.
95 * systemd-nspawn gained support for a new --system-call-filter= command
96 line option for adding/removing entries in the default system call
97 filter it applies. Moreover systemd-nspawn has been changed to
98 implement a system call whitelist instead of a blacklist.
100 * systemd-run gained support for a new --pipe command line option. If
101 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
102 are directly passed on to the activated transient service
103 binary. This allows invoking arbitrary processes as systemd services
104 (for example to take benefit of dependency management, accounting
105 management, resource management or log management that is done
106 automatically for services) — while still allowing them to be
107 integrated in a classic UNIX shell pipeline.
109 * When a service sends RELOAD=1 via sd_notify() and reload propagation
110 using ReloadPropagationTo= is configured, a reload is now propagated
111 to configured units. (Previously this was only done on explicitly
112 requested reloads, using "systemctl reload" or an equivalent
115 * For each service unit a restart counter is now kept: it is increased
116 each time the service is restarted due to Restart=, and may be
117 queried using "systemctl show -p NRestarts …".
119 * New system call filter groups @setuid, @memlock, @signal and
120 @timer have been added, for usage with SystemCallFilter=
121 in unit files and the new --system-call-filter= command line option
122 of systemd-nspawn (see above).
124 * ExecStart= lines in unit files gained two new modifiers: when a
125 command line is prefixed with "!" the command will be executed as
126 configured, except for the credentials applied by
127 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
128 "+", but does still apply namespacing options unlike "+". There's
129 also "!!" now, which is mostly identical, but becomes a NOP on
130 systems that support ambient capabilities. This is useful to write
131 unit files that work with ambient capabilities where possible but
132 automatically fall back to traditional privilege dropping mechanisms
133 on systems where this is not supported.
135 * ListenNetlink= settings in socket units now support RDMA netlink
138 * A new unit file setting LockPersonality= has been added which permits
139 locking down the chosen execution domain ("personality") of a service
142 * A new special target "getty-pre.target" has been added, which is
143 ordered before all text logins, and may be used to order services
144 before, that shall run before these textual logins acquire access to
147 * systemd will now attempt to load the virtio-rng.ko kernel module very
148 early on if a VM environment supporting this is detected. This should
149 improve entropy during early boot in virtualized environments.
151 * A _netdev option is now supported in /etc/crypttab that operates in a
152 similar way as the same option in /etc/fstab: it permits configuring
153 encrypted devices that need to be ordered after the network coming
154 up. Following this logic, two new special targets
155 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
156 added that are to cryptsetup.target what
157 remote-fs.target/remote-fs-pre.target are to local-fs.target.
159 * Service units gained a new UnsetEnvironment= setting which permits
160 unsetting specific environment variables for specific services that
161 are normally passed to it (for example in order to mask out locale
162 settings for specific services that can't deal with it).
164 * Units acquired a new boolean option IPAccounting=. When turned on, IP
165 traffic accounting (packet count as well as byte count) is done for
166 the service, and shown as part of "systemctl status" or "systemd-run
169 * Service units acquired two new options IPAddressAllow= and
170 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
171 for configuring a simple IP access control list for all sockets of
172 the unit. These options are available also on .slice and .socket
173 units, permitting flexible access list configuration for individual
174 services as well as groups of services (as defined by a slice unit),
175 including system-wide. Note that IP ACLs configured this way are
176 enforced on every single IPv4 and IPv6 socket created by any process
177 of the service unit, and apply to ingress as well as egress traffic.
179 * If CPUAccounting= or IPAccounting= is turned on for a unit a new,
180 recognizable log message is generated each time the unit is stopped,
181 containing information about the consumed resources of this
184 * A new setting KeyringMode= has been added to unit files, which may be
185 used to control how the kernel keyring is set up for executed
188 * "systemctl poweroff", "systemctl reboot", "systemctl halt",
189 "systemctl kexec" and "systemctl exit" are now always asynchronous in
190 behaviour (that is: these commands return immediately after the
191 operation was enqueued instead of waiting until the operation was
192 completed). Previously, "systemctl poweroff" and "systemctl reboot"
193 were asynchronous on systems using systemd-logind (i.e. almost
194 always, and like they were on sysvinit), and the other three commands
195 were unconditionally synchronous. With this release this is cleaned
196 up, and callers will see the same asynchronous behaviour on all
197 systems for all five operations.
199 * systemd-logind gained new Halt() and CanHalt() bus calls for halting
202 * .timer units now accept calendar specifications in other timezones
203 than UTC or the local timezone.
205 * The tmpfiles snippet var.conf has been changed to create
206 /var/log/btmp with access mode 0660 instead of 0600. It has been
207 owned by the "utmp" group already, and it appears to be generally
208 understood that members of "utmp" can modify/flush the
209 utmp/wtmp/lastlog/btmp databases. Previously this was implemented
210 correctly for all these database excepts btmp, which has been opened
211 up like this now too. Note that while the other databases are
212 world-readable (i.e. 0644), btmp is not and remains more restrictive.
214 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
215 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
216 Burchardt, b1tninja, bengal, Benjamin Berg, Benjamin Robin, Charles
217 Huber, Christian Hesse, Daniel Berrange, Daniel Mack, Daniel Rusek,
218 dasj19, Davide Cavalca, Dimitri John Ledkov, Diogo Pereira, Djalal
219 Harouni, dkg, dmig, Dmitry Torokhov, ettavolt, Evgeny Vereshchagin,
220 Fabio Kung, Felipe Sateler, Franck Bui, g0tar, Hans de Goede, Harald
221 Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov, Jakub Wilk, Jan
222 Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen, John Lin,
223 jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg Thalheim,
224 Jouke Witteveen, juga0, Justin Michaud, Kai-Heng Feng, Lennart
225 Poettering, Lion Yang, Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn,
226 Marcel Hollerbach, Marcus Lundblad, Martin Pitt, Michael Biebl, Michael
227 Grzeschik, Michal Sekletar, Mike Gilbert, Neil Brown, Nicolas Iooss,
228 Patrik Flykt, pEJipE, Russell Stuart, S. Fan, Shengyao Xue, Stefan
229 Pietsch, Susant Sahani, Tejun Heo, Thomas Miller, Thomas Sailer, Tobias
230 Hunger, Tom Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø,
231 userwithuid, Vito Caputo, vliaskov, WaLyong Cho, William Douglas, Xiang
232 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
238 * Meson is now supported as build system in addition to Automake. It is
239 our plan to remove Automake in one of our next releases, so that
240 Meson becomes our exclusive build system. Hence, please start using
241 the Meson build system in your downstream packaging. There's plenty
242 of documentation around how to use Meson, the extremely brief
245 ./autogen.sh && ./configure && make && sudo make install
249 meson build && ninja -C build && sudo ninja -C build install
251 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
252 which permits configuring a timeout on the time a job is
253 running. This is particularly useful for setting timeouts on jobs for
256 * Unit files gained two new options ConditionUser= and ConditionGroup=
257 for conditionalizing units based on the identity of the user/group
258 running a systemd user instance.
260 * systemd-networkd now understands a new FlowLabel= setting in the
261 [VXLAN] section of .network files, as well as a Priority= in
262 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
263 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
264 gained support for configuration of GENEVE links, and IPv6 address
265 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
267 * .link files now understand a new Port= setting.
269 * systemd-networkd's DHCP support gained support for DHCP option 119
270 (domain search list).
272 * systemd-networkd gained support for serving IPv6 address ranges using
273 the Router Advertisment protocol. The new .network configuration
274 section [IPv6Prefix] may be used to configure the ranges to
275 serve. This is implemented based on a new, minimal, native server
276 implementation of RA.
278 * journalctl's --output= switch gained support for a new parameter
279 "short-iso-precise" for a mode where timestamps are shown as precise
282 * systemd-udevd's "net_id" builtin may now generate stable network
283 interface names from IBM PowerVM VIO devices as well as ACPI platform
286 * MulticastDNS support in systemd-resolved may now be explicitly
287 enabled/disabled using the new MulticastDNS= configuration file
290 * systemd-resolved may now optionally use libidn2 instead of the libidn
291 for processing internationalized domain names. Support for libidn2
292 should be considered experimental and should not be enabled by
295 * "machinectl pull-tar" and related call may now do verification of
296 downloaded images using SUSE-style .sha256 checksum files in addition
297 to the already existing support for validating using Ubuntu-style
300 * sd-bus gained support for a new sd_bus_message_appendv() call which
301 is va_list equivalent of sd_bus_message_append().
303 * sd-boot gained support for validating images using SHIM/MOK.
305 * The SMACK code learnt support for "onlycap".
307 * systemd-mount --umount is now much smarter in figuring out how to
308 properly unmount a device given its mount or device path.
310 * The code to call libnss_dns as a fallback from libnss_resolve when
311 the communication with systemd-resolved fails was removed. This
312 fallback was redundant and interfered with the [!UNAVAIL=return]
313 suffix. See nss-resolve(8) for the recommended configuration.
315 * systemd-logind may now be restarted without losing state. It stores
316 the file descriptors for devices it manages in the system manager
317 using the FDSTORE= mechanism. Please note that further changes in
318 other components may be required to make use of this (for example
319 Xorg has code to listen for stops of systemd-logind and terminate
320 itself when logind is stopped or restarted, in order to avoid using
321 stale file descriptors for graphical devices, which is now
322 counterproductive and must be reverted in order for restarts of
323 systemd-logind to be safe. See
324 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
326 * All kernel install plugins are called with the environment variable
327 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
328 /etc/machine-id. If the file is missing or empty, the variable is
329 empty and BOOT_DIR_ABS is the path of a temporary directory which is
330 removed after all the plugins exit. So, if KERNEL_INSTALL_MACHINE_ID
331 is empty, all plugins should not put anything in BOOT_DIR_ABS.
333 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
334 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
335 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
336 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
337 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
338 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
339 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
340 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
341 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
342 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
343 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
344 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
345 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
346 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
347 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
348 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
349 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
350 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
351 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
352 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
353 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
354 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
355 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
356 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
357 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
358 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
359 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
360 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
361 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
368 * The "hybrid" control group mode has been modified to improve
369 compatibility with "legacy" cgroups-v1 setups. Specifically, the
370 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
371 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
372 cgroups-v1 hierarchy), the only externally visible change being that
373 the cgroups-v2 hierarchy is also mounted, to
374 /sys/fs/cgroup/unified. This should provide a large degree of
375 compatibility with "legacy" cgroups-v1, while taking benefit of the
376 better management capabilities of cgroups-v2.
378 * The default control group setup mode may be selected both a boot-time
379 via a set of kernel command line parameters (specifically:
380 systemd.unified_cgroup_hierarchy= and
381 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
382 default selected on the configure command line
383 (--with-default-hierarchy=). The upstream default is "hybrid"
384 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
385 this will change in a future systemd version to be "unified" (pure
386 cgroups-v2 mode). The third option for the compile time option is
387 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
388 distributions to default to "hybrid" mode for release distributions,
389 starting with v233. We recommend "unified" for development
390 distributions (specifically: distributions such as Fedora's rawhide)
391 as that's where things are headed in the long run. Use "legacy" for
392 greatest stability and compatibility only.
394 * Note one current limitation of "unified" and "hybrid" control group
395 setup modes: the kernel currently does not permit the systemd --user
396 instance (i.e. unprivileged code) to migrate processes between two
397 disconnected cgroup subtrees, even if both are managed and owned by
398 the user. This effectively means "systemd-run --user --scope" doesn't
399 work when invoked from outside of any "systemd --user" service or
400 scope. Specifically, it is not supported from session scopes. We are
401 working on fixing this in a future systemd version. (See #3388 for
402 further details about this.)
404 * DBus policy files are now installed into /usr rather than /etc. Make
405 sure your system has dbus >= 1.9.18 running before upgrading to this
406 version, or override the install path with --with-dbuspolicydir= .
408 * All python scripts shipped with systemd (specifically: the various
409 tests written in Python) now require Python 3.
411 * systemd unit tests can now run standalone (without the source or
412 build directories), and can be installed into /usr/lib/systemd/tests/
413 with 'make install-tests'.
415 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
416 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
419 * Support for the %c, %r, %R specifiers in unit files has been
420 removed. Specifiers are not supposed to be dependent on configuration
421 in the unit file itself (so that they resolve the same regardless
422 where used in the unit files), but these specifiers were influenced
423 by the Slice= option.
425 * The shell invoked by debug-shell.service now defaults to /bin/sh in
426 all cases. If distributions want to use a different shell for this
427 purpose (for example Fedora's /sbin/sushell) they need to specify
428 this explicitly at configure time using --with-debug-shell=.
430 * The confirmation spawn prompt has been reworked to offer the
433 (c)ontinue, proceed without asking anymore
434 (D)ump, show the state of the unit
435 (f)ail, don't execute the command and pretend it failed
437 (i)nfo, show a short summary of the unit
438 (j)obs, show jobs that are in progress
439 (s)kip, don't execute the command and pretend it succeeded
440 (y)es, execute the command
442 The 'n' choice for the confirmation spawn prompt has been removed,
443 because its meaning was confusing.
445 The prompt may now also be redirected to an alternative console by
446 specifying the console as parameter to systemd.confirm_spawn=.
448 * Services of Type=notify require a READY=1 notification to be sent
449 during startup. If no such message is sent, the service now fails,
450 even if the main process exited with a successful exit code.
452 * Services that fail to start up correctly now always have their
453 ExecStopPost= commands executed. Previously, they'd enter "failed"
454 state directly, without executing these commands.
456 * The option MulticastDNS= of network configuration files has acquired
457 an actual implementation. With MulticastDNS=yes a host can resolve
458 names of remote hosts and reply to mDNS A and AAAA requests.
460 * When units are about to be started an additional check is now done to
461 ensure that all dependencies of type BindsTo= (when used in
462 combination with After=) have been started.
464 * systemd-analyze gained a new verb "syscall-filter" which shows which
465 system call groups are defined for the SystemCallFilter= unit file
466 setting, and which system calls they contain.
468 * A new system call filter group "@filesystem" has been added,
469 consisting of various file system related system calls. Group
470 "@reboot" has been added, covering reboot, kexec and shutdown related
471 calls. Finally, group "@swap" has been added covering swap
472 configuration related calls.
474 * A new unit file option RestrictNamespaces= has been added that may be
475 used to restrict access to the various process namespace types the
476 Linux kernel provides. Specifically, it may be used to take away the
477 right for a service unit to create additional file system, network,
478 user, and other namespaces. This sandboxing option is particularly
479 relevant due to the high amount of recently discovered namespacing
480 related vulnerabilities in the kernel.
482 * systemd-udev's .link files gained support for a new AutoNegotiation=
483 setting for configuring Ethernet auto-negotiation.
485 * systemd-networkd's .network files gained support for a new
486 ListenPort= setting in the [DHCP] section to explicitly configure the
487 UDP client port the DHCP client shall listen on.
489 * .network files gained a new Unmanaged= boolean setting for explicitly
490 excluding one or more interfaces from management by systemd-networkd.
492 * The systemd-networkd ProxyARP= option has been renamed to
493 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
494 renamed to ReduceARPProxy=. The old names continue to be available
497 * systemd-networkd gained support for configuring IPv6 Proxy NDP
498 addresses via the new IPv6ProxyNDPAddress= .network file setting.
500 * systemd-networkd's bonding device support gained support for two new
501 configuration options ActiveSlave= and PrimarySlave=.
503 * The various options in the [Match] section of .network files gained
504 support for negative matching.
506 * New systemd-specific mount options are now understood in /etc/fstab:
508 x-systemd.mount-timeout= may be used to configure the maximum
509 permitted runtime of the mount command.
511 x-systemd.device-bound may be set to bind a mount point to its
512 backing device unit, in order to automatically remove a mount point
513 if its backing device is unplugged. This option may also be
514 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
515 on the block device, which is now automatically set for all CDROM
516 drives, so that mounted CDs are automatically unmounted when they are
517 removed from the drive.
519 x-systemd.after= and x-systemd.before= may be used to explicitly
520 order a mount after or before another unit or mount point.
522 * Enqueued start jobs for device units are now automatically garbage
523 collected if there are no jobs waiting for them anymore.
525 * systemctl list-jobs gained two new switches: with --after, for every
526 queued job the jobs it's waiting for are shown; with --before the
527 jobs which it's blocking are shown.
529 * systemd-nspawn gained support for ephemeral boots from disk images
530 (or in other words: --ephemeral and --image= may now be
531 combined). Moreover, ephemeral boots are now supported for normal
532 directories, even if the backing file system is not btrfs. Of course,
533 if the file system does not support file system snapshots or
534 reflinks, the initial copy operation will be relatively expensive, but
535 this should still be suitable for many use cases.
537 * Calendar time specifications in .timer units now support
538 specifications relative to the end of a month by using "~" instead of
539 "-" as separator between month and day. For example, "*-02~03" means
540 "the third last day in February". In addition a new syntax for
541 repeated events has been added using the "/" character. For example,
542 "9..17/2:00" means "every two hours from 9am to 5pm".
544 * systemd-socket-proxyd gained a new parameter --connections-max= for
545 configuring the maximum number of concurrent connections.
547 * sd-id128 gained a new API for generating unique IDs for the host in a
548 way that does not leak the machine ID. Specifically,
549 sd_id128_get_machine_app_specific() derives an ID based on the
550 machine ID a in well-defined, non-reversible, stable way. This is
551 useful whenever an identifier for the host is needed but where the
552 identifier shall not be useful to identify the system beyond the
553 scope of the application itself. (Internally this uses HMAC-SHA256 as
554 keyed hash function using the machine ID as input.)
556 * NotifyAccess= gained a new supported value "exec". When set
557 notifications are accepted from all processes systemd itself invoked,
558 including all control processes.
560 * .nspawn files gained support for defining overlay mounts using the
561 Overlay= and OverlayReadOnly= options. Previously this functionality
562 was only available on the systemd-nspawn command line.
564 * systemd-nspawn's --bind= and --overlay= options gained support for
565 bind/overlay mounts whose source lies within the container tree by
566 prefixing the source path with "+".
568 * systemd-nspawn's --bind= and --overlay= options gained support for
569 automatically allocating a temporary source directory in /var/tmp
570 that is removed when the container dies. Specifically, if the source
571 directory is specified as empty string this mechanism is selected. An
572 example usage is --overlay=+/var::/var, which creates an overlay
573 mount based on the original /var contained in the image, overlayed
574 with a temporary directory in the host's /var/tmp. This way changes
575 to /var are automatically flushed when the container shuts down.
577 * systemd-nspawn --image= option does now permit raw file system block
578 devices (in addition to images containing partition tables, as
581 * The disk image dissection logic in systemd-nspawn gained support for
582 automatically setting up LUKS encrypted as well as Verity protected
583 partitions. When a container is booted from an encrypted image the
584 passphrase is queried at start-up time. When a container with Verity
585 data is started, the root hash is search in a ".roothash" file
586 accompanying the disk image (alternatively, pass the root hash via
587 the new --root-hash= command line option).
589 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
590 be used to dissect disk images the same way as systemd-nspawn does
591 it, following the Bootable Partition Specification. It may even be
592 used to mount disk images with complex partition setups (including
593 LUKS and Verity partitions) to a local host directory, in order to
594 inspect them. This tool is not considered public API (yet), and is
595 thus not installed into /usr/bin. Please do not rely on its
596 existence, since it might go away or be changed in later systemd
599 * A new generator "systemd-verity-generator" has been added, similar in
600 style to "systemd-cryptsetup-generator", permitting automatic setup of
601 Verity root partitions when systemd boots up. In order to make use of
602 this your partition setup should follow the Discoverable Partitions
603 Specification, and the GPT partition ID of the root file system
604 partition should be identical to the upper 128bit of the Verity root
605 hash. The GPT partition ID of the Verity partition protecting it
606 should be the lower 128bit of the Verity root hash. If the partition
607 image follows this model it is sufficient to specify a single
608 "roothash=" kernel command line argument to both configure which root
609 image and verity partition to use as well as the root hash for
610 it. Note that systemd-nspawn's Verity support follows the same
611 semantics, meaning that disk images with proper Verity data in place
612 may be booted in containers with systemd-nspawn as well as on
613 physical systems via the verity generator. Also note that the "mkosi"
614 tool available at https://github.com/systemd/mkosi has been updated
615 to generate Verity protected disk images following this scheme. In
616 fact, it has been updated to generate disk images that optionally
617 implement a complete UEFI SecureBoot trust chain, involving a signed
618 kernel and initrd image that incorporates such a root hash as well as
619 a Verity-enabled root partition.
621 * The hardware database (hwdb) udev supports has been updated to carry
622 accelerometer quirks.
624 * All system services are now run with a fresh kernel keyring set up
625 for them. The invocation ID is stored by default in it, thus
626 providing a safe, non-overridable way to determine the invocation
629 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
630 options for bind mounting arbitrary paths in a service-specific
631 way. When these options are used, arbitrary host or service files and
632 directories may be mounted to arbitrary locations in the service's
635 * Documentation has been added that lists all of systemd's low-level
636 environment variables:
638 https://github.com/systemd/systemd/blob/master/ENVIRONMENT.md
640 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
641 whether a specific socket file descriptor matches a specified socket
644 * systemd-firstboot has been updated to check for the
645 systemd.firstboot= kernel command line option. It accepts a boolean
646 and when set to false the first boot questions are skipped.
648 * systemd-fstab-generator has been updated to check for the
649 systemd.volatile= kernel command line option, which either takes an
650 optional boolean parameter or the special value "state". If used the
651 system may be booted in a "volatile" boot mode. Specifically,
652 "systemd.volatile" is used, the root directory will be mounted as
653 tmpfs, and only /usr is mounted from the actual root file system. If
654 "systemd.volatile=state" is used, the root directory will be mounted
655 as usual, but /var is mounted as tmpfs. This concept provides similar
656 functionality as systemd-nspawn's --volatile= option, but provides it
657 on physical boots. Use this option for implementing stateless
658 systems, or testing systems with all state and/or configuration reset
659 to the defaults. (Note though that many distributions are not
660 prepared to boot up without a populated /etc or /var, though.)
662 * systemd-gpt-auto-generator gained support for LUKS encrypted root
663 partitions. Previously it only supported LUKS encrypted partitions
664 for all other uses, except for the root partition itself.
666 * Socket units gained support for listening on AF_VSOCK sockets for
667 communication in virtualized QEMU environments.
669 * The "configure" script gained a new option --with-fallback-hostname=
670 for specifying the fallback hostname to use if none is configured in
671 /etc/hostname. For example, by specifying
672 --with-fallback-hostname=fedora it is possible to default to a
673 hostname of "fedora" on pristine installations.
675 * systemd-cgls gained support for a new --unit= switch for listing only
676 the control groups of a specific unit. Similar --user-unit= has been
677 added for listing only the control groups of a specific user unit.
679 * systemd-mount gained a new --umount switch for unmounting a mount or
680 automount point (and all mount/automount points below it).
682 * systemd will now refuse full configuration reloads (via systemctl
683 daemon-reload and related calls) unless at least 16MiB of free space
684 are available in /run. This is a safety precaution in order to ensure
685 that generators can safely operate after the reload completed.
687 * A new unit file option RootImage= has been added, which has a similar
688 effect as RootDirectory= but mounts the service's root directory from
689 a disk image instead of plain directory. This logic reuses the same
690 image dissection and mount logic that systemd-nspawn already uses,
691 and hence supports any disk images systemd-nspawn supports, including
692 those following the Discoverable Partition Specification, as well as
693 Verity enabled images. This option enables systemd to run system
694 services directly off disk images acting as resource bundles,
695 possibly even including full integrity data.
697 * A new MountAPIVFS= unit file option has been added, taking a boolean
698 argument. If enabled /proc, /sys and /dev (collectively called the
699 "API VFS") will be mounted for the service. This is only relevant if
700 RootDirectory= or RootImage= is used for the service, as these mounts
701 are of course in place in the host mount namespace anyway.
703 * systemd-nspawn gained support for a new --pivot-root= switch. If
704 specified the root directory within the container image is pivoted to
705 the specified mount point, while the original root disk is moved to a
706 different place. This option enables booting of ostree images
707 directly with systemd-nspawn.
709 * The systemd build scripts will no longer complain if the NTP server
710 addresses are not changed from the defaults. Google now supports
711 these NTP servers officially. We still recommend downstreams to
712 properly register an NTP pool with the NTP pool project though.
714 * coredumpctl gained a new "--reverse" option for printing the list
715 of coredumps in reverse order.
717 * coredumpctl will now show additional information about truncated and
718 inaccessible coredumps, as well as coredumps that are still being
719 processed. It also gained a new --quiet switch for suppressing
720 additional informational message in its output.
722 * coredumpctl gained support for only showing coredumps newer and/or
723 older than specific timestamps, using the new --since= and --until=
724 options, reminiscent of journalctl's options by the same name.
726 * The systemd-coredump logic has been improved so that it may be reused
727 to collect backtraces in non-compiled languages, for example in
728 scripting languages such as Python.
730 * machinectl will now show the UID shift of local containers, if user
731 namespacing is enabled for them.
733 * systemd will now optionally run "environment generator" binaries at
734 configuration load time. They may be used to add environment
735 variables to the environment block passed to services invoked. One
736 user environment generator is shipped by default that sets up
737 environment variables based on files dropped into /etc/environment.d
738 and ~/.config/environment.d/.
740 * systemd-resolved now includes the new, recently published 2017 DNSSEC
743 * hostnamed has been updated to report a new chassis type of
744 "convertible" to cover "foldable" laptops that can both act as a
745 tablet and as a laptop, such as various Lenovo Yoga devices.
747 Contributions from: Adrián López, Alexander Galanin, Alexander
748 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
749 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
750 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
751 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
752 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
753 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
754 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
755 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
756 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
757 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
758 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
759 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
760 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
761 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
762 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
763 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
764 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
765 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
766 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
767 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
768 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
769 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
770 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
771 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
772 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
773 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
774 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
781 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
782 RestrictAddressFamilies= enabled. These sandboxing options should
783 generally be compatible with the various external udev call-out
784 binaries we are aware of, however there may be exceptions, in
785 particular when exotic languages for these call-outs are used. In
786 this case, consider turning off these settings locally.
788 * The new RemoveIPC= option can be used to remove IPC objects owned by
789 the user or group of a service when that service exits.
791 * The new ProtectKernelModules= option can be used to disable explicit
792 load and unload operations of kernel modules by a service. In
793 addition access to /usr/lib/modules is removed if this option is set.
795 * ProtectSystem= option gained a new value "strict", which causes the
796 whole file system tree with the exception of /dev, /proc, and /sys,
797 to be remounted read-only for a service.
799 * The new ProtectKernelTunables= option can be used to disable
800 modification of configuration files in /sys and /proc by a service.
801 Various directories and files are remounted read-only, so access is
802 restricted even if the file permissions would allow it.
804 * The new ProtectControlGroups= option can be used to disable write
805 access by a service to /sys/fs/cgroup.
807 * Various systemd services have been hardened with
808 ProtectKernelTunables=yes, ProtectControlGroups=yes,
809 RestrictAddressFamilies=.
811 * Support for dynamically creating users for the lifetime of a service
812 has been added. If DynamicUser=yes is specified, user and group IDs
813 will be allocated from the range 61184..65519 for the lifetime of the
814 service. They can be resolved using the new nss-systemd.so NSS
815 module. The module must be enabled in /etc/nsswitch.conf. Services
816 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
817 any resources allocated by the service will be cleaned up when the
818 service exits. They also have ProtectHome=read-only and
819 ProtectSystem=strict enabled, so they are not able to make any
820 permanent modifications to the system.
822 * The nss-systemd module also always resolves root and nobody, making
823 it possible to have no /etc/passwd or /etc/group files in minimal
824 container or chroot environments.
826 * Services may be started with their own user namespace using the new
827 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
828 under which the service is running are mapped. All other users are
831 * Support for the cgroup namespace has been added to systemd-nspawn. If
832 supported by kernel, the container system started by systemd-nspawn
833 will have its own view of the cgroup hierarchy. This new behaviour
834 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
836 * The new MemorySwapMax= option can be used to limit the maximum swap
837 usage under the unified cgroup hierarchy.
839 * Support for the CPU controller in the unified cgroup hierarchy has
840 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
841 options. This controller requires out-of-tree patches for the kernel
842 and the support is provisional.
844 * Mount and automount units may now be created transiently
845 (i.e. dynamically at runtime via the bus API, instead of requiring
846 unit files in the file system).
848 * systemd-mount is a new tool which may mount file systems – much like
849 mount(8), optionally pulling in additional dependencies through
850 transient .mount and .automount units. For example, this tool
851 automatically runs fsck on a backing block device before mounting,
852 and allows the automount logic to be used dynamically from the
853 command line for establishing mount points. This tool is particularly
854 useful when dealing with removable media, as it will ensure fsck is
855 run – if necessary – before the first access and that the file system
856 is quickly unmounted after each access by utilizing the automount
857 logic. This maximizes the chance that the file system on the
858 removable media stays in a clean state, and if it isn't in a clean
859 state is fixed automatically.
861 * LazyUnmount=yes option for mount units has been added to expose the
862 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
865 * /efi will be used as the mount point of the EFI boot partition, if
866 the directory is present, and the mount point was not configured
867 through other means (e.g. fstab). If /efi directory does not exist,
868 /boot will be used as before. This makes it easier to automatically
869 mount the EFI partition on systems where /boot is used for something
872 * When operating on GPT disk images for containers, systemd-nspawn will
873 now mount the ESP to /boot or /efi according to the same rules as PID
874 1 running on a host. This allows tools like "bootctl" to operate
875 correctly within such containers, in order to make container images
876 bootable on physical systems.
878 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
880 * Two new user session targets have been added to support running
881 graphical sessions under the systemd --user instance:
882 graphical-session.target and graphical-session-pre.target. See
883 systemd.special(7) for a description of how those targets should be
886 * The vconsole initialization code has been significantly reworked to
887 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
888 support unicode keymaps. Font and keymap configuration will now be
889 copied to all allocated virtual consoles.
891 * FreeBSD's bhyve virtualization is now detected.
893 * Information recorded in the journal for core dumps now includes the
894 contents of /proc/mountinfo and the command line of the process at
895 the top of the process hierarchy (which is usually the init process
898 * systemd-journal-gatewayd learned the --directory= option to serve
899 files from the specified location.
901 * journalctl --root=… can be used to peruse the journal in the
902 /var/log/ directories inside of a container tree. This is similar to
903 the existing --machine= option, but does not require the container to
906 * The hardware database has been extended to support
907 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
910 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
911 specify the click rate for mice which include a horizontal wheel with
912 a click rate that is different than the one for the vertical wheel.
914 * systemd-run gained a new --wait option that makes service execution
915 synchronous. (Specifically, the command will not return until the
916 specified service binary exited.)
918 * systemctl gained a new --wait option that causes the start command to
919 wait until the units being started have terminated again.
921 * A new journal output mode "short-full" has been added which displays
922 timestamps with abbreviated English day names and adds a timezone
923 suffix. Those timestamps include more information than the default
924 "short" output mode, and can be passed directly to journalctl's
925 --since= and --until= options.
927 * /etc/resolv.conf will be bind-mounted into containers started by
928 systemd-nspawn, if possible, so any changes to resolv.conf contents
929 are automatically propagated to the container.
931 * The number of instances for socket-activated services originating
932 from a single IP address can be limited with
933 MaxConnectionsPerSource=, extending the existing setting of
936 * systemd-networkd gained support for vcan ("Virtual CAN") interface
939 * .netdev and .network configuration can now be extended through
942 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
943 Segmentation Offload, Generic Receive Offload, Large Receive Offload
944 can be enabled and disabled using the new UDPSegmentationOffload=,
945 TCPSegmentationOffload=, GenericSegmentationOffload=,
946 GenericReceiveOffload=, LargeReceiveOffload= options in the
947 [Link] section of .link files.
949 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
950 Port VLAN ID can be configured for bridge devices using the new STP=,
951 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
952 section of .netdev files.
954 * The route table to which routes received over DHCP or RA should be
955 added can be configured with the new RouteTable= option in the [DHCP]
956 and [IPv6AcceptRA] sections of .network files.
958 * The Address Resolution Protocol can be disabled on links managed by
959 systemd-networkd using the ARP=no setting in the [Link] section of
962 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
963 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
964 encode information about the result and exit codes of the current
965 service runtime cycle.
967 * systemd-sysctl will now configure kernel parameters in the order
968 they occur in the configuration files. This matches what sysctl
969 has been traditionally doing.
971 * kernel-install "plugins" that are executed to perform various
972 tasks after a new kernel is added and before an old one is removed
973 can now return a special value to terminate the procedure and
974 prevent any later plugins from running.
976 * Journald's SplitMode=login setting has been deprecated. It has been
977 removed from documentation, and its use is discouraged. In a future
978 release it will be completely removed, and made equivalent to current
979 default of SplitMode=uid.
981 * Storage=both option setting in /etc/systemd/coredump.conf has been
982 removed. With fast LZ4 compression storing the core dump twice is not
985 * The --share-system systemd-nspawn option has been replaced with an
986 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
987 this functionality is discouraged. In addition the variables
988 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
989 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
990 individual namespaces.
992 * "machinectl list" now shows the IP address of running containers in
993 the output, as well as OS release information.
995 * "loginctl list" now shows the TTY of each session in the output.
997 * sd-bus gained new API calls sd_bus_track_set_recursive(),
998 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
999 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
1000 tracking objects in a "recursive" mode, where a single client can be
1001 counted multiple times, if it takes multiple references.
1003 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
1004 sd_bus_get_exit_on_disconnect(). They may be used to to make a
1005 process using sd-bus automatically exit if the bus connection is
1008 * Bus clients of the service manager may now "pin" loaded units into
1009 memory, by taking an explicit reference on them. This is useful to
1010 ensure the client can retrieve runtime data about the service even
1011 after the service completed execution. Taking such a reference is
1012 available only for privileged clients and should be helpful to watch
1013 running services in a race-free manner, and in particular collect
1014 information about exit statuses and results.
1016 * The nss-resolve module has been changed to strictly return UNAVAIL
1017 when communication via D-Bus with resolved failed, and NOTFOUND when
1018 a lookup completed but was negative. This means it is now possible to
1019 neatly configure fallbacks using nsswitch.conf result checking
1020 expressions. Taking benefit of this, the new recommended
1021 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
1023 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
1025 * A new setting CtrlAltDelBurstAction= has been added to
1026 /etc/systemd/system.conf which may be used to configure the precise
1027 behaviour if the user on the console presses Ctrl-Alt-Del more often
1028 than 7 times in 2s. Previously this would unconditionally result in
1029 an expedited, immediate reboot. With this new setting the precise
1030 operation may be configured in more detail, and also turned off
1033 * In .netdev files two new settings RemoteChecksumTx= and
1034 RemoteChecksumRx= are now understood that permit configuring the
1035 remote checksumming logic for VXLAN networks.
1037 * The service manager learnt a new "invocation ID" concept for invoked
1038 services. Each runtime cycle of a service will get a new invocation
1039 ID (a 128bit random UUID) assigned that identifies the current
1040 run of the service uniquely and globally. A new invocation ID
1041 is generated each time a service starts up. The journal will store
1042 the invocation ID of a service along with any logged messages, thus
1043 making the invocation ID useful for matching the online runtime of a
1044 service with the offline log data it generated in a safe way without
1045 relying on synchronized timestamps. In many ways this new service
1046 invocation ID concept is similar to the kernel's boot ID concept that
1047 uniquely and globally identifies the runtime of each boot. The
1048 invocation ID of a service is passed to the service itself via an
1049 environment variable ($INVOCATION_ID). A new bus call
1050 GetUnitByInvocationID() has been added that is similar to GetUnit()
1051 but instead of retrieving the bus path for a unit by its name
1052 retrieves it by its invocation ID. The returned path is valid only as
1053 long as the passed invocation ID is current.
1055 * systemd-resolved gained a new "DNSStubListener" setting in
1056 resolved.conf. It either takes a boolean value or the special values
1057 "udp" and "tcp", and configures whether to enable the stub DNS
1058 listener on 127.0.0.53:53.
1060 * IP addresses configured via networkd may now carry additional
1061 configuration settings supported by the kernel. New options include:
1062 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
1063 PrefixRoute=, AutoJoin=.
1065 * The PAM configuration fragment file for "user@.service" shipped with
1066 systemd (i.e. the --user instance of systemd) has been stripped to
1067 the minimum necessary to make the system boot. Previously, it
1068 contained Fedora-specific stanzas that did not apply to other
1069 distributions. It is expected that downstream distributions add
1070 additional configuration lines, matching their needs to this file,
1071 using it only as rough template of what systemd itself needs. Note
1072 that this reduced fragment does not even include an invocation of
1073 pam_limits which most distributions probably want to add, even though
1074 systemd itself does not need it. (There's also the new build time
1075 option --with-pamconfdir=no to disable installation of the PAM
1078 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
1079 capability is now also dropped from its set (in addition to
1080 CAP_SYS_MKNOD as before).
1082 * In service unit files it is now possible to connect a specific named
1083 file descriptor with stdin/stdout/stdout of an executed service. The
1084 name may be specified in matching .socket units using the
1085 FileDescriptorName= setting.
1087 * A number of journal settings may now be configured on the kernel
1088 command line. Specifically, the following options are now understood:
1089 systemd.journald.max_level_console=,
1090 systemd.journald.max_level_store=,
1091 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
1092 systemd.journald.max_level_wall=.
1094 * "systemctl is-enabled --full" will now show by which symlinks a unit
1095 file is enabled in the unit dependency tree.
1097 * Support for VeraCrypt encrypted partitions has been added to the
1098 "cryptsetup" logic and /etc/crypttab.
1100 * systemd-detect-virt gained support for a new --private-users switch
1101 that checks whether the invoking processes are running inside a user
1102 namespace. Similar, a new special value "private-users" for the
1103 existing ConditionVirtualization= setting has been added, permitting
1104 skipping of specific units in user namespace environments.
1106 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
1107 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
1108 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
1109 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
1110 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
1111 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
1112 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
1113 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
1114 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
1115 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
1116 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
1117 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
1118 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
1119 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
1120 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
1121 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
1122 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
1123 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
1124 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
1125 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
1126 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
1127 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
1128 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
1129 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
1130 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
1131 Jędrzejewski-Szmek, Zeal Jagannatha
1133 — Santa Fe, 2016-11-03
1137 * In service units the various ExecXYZ= settings have been extended
1138 with an additional special character as first argument of the
1139 assigned value: if the character '+' is used the specified command
1140 line it will be run with full privileges, regardless of User=,
1141 Group=, CapabilityBoundingSet= and similar options. The effect is
1142 similar to the existing PermissionsStartOnly= option, but allows
1143 configuration of this concept for each executed command line
1146 * Services may now alter the service watchdog timeout at runtime by
1147 sending a WATCHDOG_USEC= message via sd_notify().
1149 * MemoryLimit= and related unit settings now optionally take percentage
1150 specifications. The percentage is taken relative to the amount of
1151 physical memory in the system (or in case of containers, the assigned
1152 amount of memory). This allows scaling service resources neatly with
1153 the amount of RAM available on the system. Similarly, systemd-logind's
1154 RuntimeDirectorySize= option now also optionally takes percentage
1157 * In similar fashion TasksMax= takes percentage values now, too. The
1158 value is taken relative to the configured maximum number of processes
1159 on the system. The per-service task maximum has been changed to 15%
1160 using this functionality. (Effectively this is an increase of 512 →
1161 4915 for service units, given the kernel's default pid_max setting.)
1163 * Calendar time specifications in .timer units now understand a ".."
1164 syntax for time ranges. Example: "4..7:10" may now be used for
1165 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
1168 * The InaccessableDirectories=, ReadOnlyDirectories= and
1169 ReadWriteDirectories= unit file settings have been renamed to
1170 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
1171 applied to all kinds of file nodes, and not just directories, with
1172 the exception of symlinks. Specifically these settings may now be
1173 used on block and character device nodes, UNIX sockets and FIFOS as
1174 well as regular files. The old names of these settings remain
1175 available for compatibility.
1177 * systemd will now log about all service processes it kills forcibly
1178 (using SIGKILL) because they remained after the clean shutdown phase
1179 of the service completed. This should help identifying services that
1180 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
1181 systemd-logind's configuration a similar log message is generated for
1182 processes killed at the end of each session due to this setting.
1184 * systemd will now set the $JOURNAL_STREAM environment variable for all
1185 services whose stdout/stderr are connected to the Journal (which
1186 effectively means by default: all services). The variable contains
1187 the device and inode number of the file descriptor used for
1188 stdout/stderr. This may be used by invoked programs to detect whether
1189 their stdout/stderr is connected to the Journal, in which case they
1190 can switch over to direct Journal communication, thus being able to
1191 pass extended, structured metadata along with their log messages. As
1192 one example, this is now used by glib's logging primitives.
1194 * When using systemd's default tmp.mount unit for /tmp, the mount point
1195 will now be established with the "nosuid" and "nodev" options. This
1196 avoids privilege escalation attacks that put traps and exploits into
1197 /tmp. However, this might cause problems if you e. g. put container
1198 images or overlays into /tmp; if you need this, override tmp.mount's
1199 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
1202 * systemd now supports the "memory" cgroup controller also on
1205 * The systemd-cgtop tool now optionally takes a control group path as
1206 command line argument. If specified, the control group list shown is
1207 limited to subgroups of that group.
1209 * The SystemCallFilter= unit file setting gained support for
1210 pre-defined, named system call filter sets. For example
1211 SystemCallFilter=@clock is now an effective way to make all clock
1212 changing-related system calls unavailable to a service. A number of
1213 similar pre-defined groups are defined. Writing system call filters
1214 for system services is simplified substantially with this new
1215 concept. Accordingly, all of systemd's own, long-running services now
1216 enable system call filtering based on this, by default.
1218 * A new service setting MemoryDenyWriteExecute= has been added, taking
1219 a boolean value. If turned on, a service may no longer create memory
1220 mappings that are writable and executable at the same time. This
1221 enhances security for services where this is enabled as it becomes
1222 harder to dynamically write and then execute memory in exploited
1223 service processes. This option has been enabled for all of systemd's
1224 own long-running services.
1226 * A new RestrictRealtime= service setting has been added, taking a
1227 boolean argument. If set the service's processes may no longer
1228 acquire realtime scheduling. This improves security as realtime
1229 scheduling may otherwise be used to easily freeze the system.
1231 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
1232 value. This may be used for requesting that the system manager inside
1233 of the container reports start-up completion to nspawn which then
1234 propagates this notification further to the service manager
1235 supervising nspawn itself. A related option NotifyReady= in .nspawn
1236 files has been added too. This functionality allows ordering of the
1237 start-up of multiple containers using the usual systemd ordering
1240 * machinectl gained a new command "stop" that is an alias for
1243 * systemd-resolved gained support for contacting DNS servers on
1244 link-local IPv6 addresses.
1246 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
1247 its caches. A method call for requesting the same operation has been
1248 added to the bus API too, and is made available via "systemd-resolve
1251 * systemd-resolve gained a new --status switch. If passed a brief
1252 summary of the used DNS configuration with per-interface information
1255 * resolved.conf gained a new Cache= boolean option, defaulting to
1256 on. If turned off local DNS caching is disabled. This comes with a
1257 performance penalty in particular when DNSSEC is enabled. Note that
1258 resolved disables its internal caching implicitly anyway, when the
1259 configured DNS server is on a host-local IP address such as ::1 or
1260 127.0.0.1, thus automatically avoiding double local caching.
1262 * systemd-resolved now listens on the local IP address 127.0.0.53:53
1263 for DNS requests. This improves compatibility with local programs
1264 that do not use the libc NSS or systemd-resolved's bus APIs for name
1265 resolution. This minimal DNS service is only available to local
1266 programs and does not implement the full DNS protocol, but enough to
1267 cover local DNS clients. A new, static resolv.conf file, listing just
1268 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
1269 now recommended to make /etc/resolv.conf a symlink to this file in
1270 order to route all DNS lookups to systemd-resolved, regardless if
1271 done via NSS, the bus API or raw DNS packets. Note that this local
1272 DNS service is not as fully featured as the libc NSS or
1273 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
1274 used to deliver link-local address information (as this implies
1275 sending a local interface index along), LLMNR/mDNS support via this
1276 interface is severely restricted. It is thus strongly recommended for
1277 all applications to use the libc NSS API or native systemd-resolved
1280 * systemd-networkd's bridge support learned a new setting
1281 VLANFiltering= for controlling VLAN filtering. Moreover a new section
1282 in .network files has been added for configuring VLAN bridging in
1283 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
1285 * systemd-networkd's IPv6 Router Advertisement code now makes use of
1286 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
1287 now be acquired without relying on DHCPv6. Two new options
1288 UseDomains= and UseDNS= have been added to configure this behaviour.
1290 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
1291 renamed IPv6AcceptRA=, without altering its behaviour. The old
1292 setting name remains available for compatibility reasons.
1294 * The systemd-networkd VTI/VTI6 tunneling support gained new options
1295 Key=, InputKey= and OutputKey=.
1297 * systemd-networkd gained support for VRF ("Virtual Routing Function")
1298 interface configuration.
1300 * "systemctl edit" may now be used to create new unit files by
1301 specifying the --force switch.
1303 * sd-event gained a new function sd_event_get_iteration() for
1304 requesting the current iteration counter of the event loop. It starts
1305 at zero and is increased by one with each event loop iteration.
1307 * A new rpm macro %systemd_ordering is provided by the macros.systemd
1308 file. It can be used in lieu of %systemd_requires in packages which
1309 don't use any systemd functionality and are intended to be installed
1310 in minimal containers without systemd present. This macro provides
1311 ordering dependencies to ensure that if the package is installed in
1312 the same rpm transaction as systemd, systemd will be installed before
1313 the scriptlets for the package are executed, allowing unit presets
1316 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
1317 been added to simplify packaging of generators.
1319 * The os-release file gained VERSION_CODENAME field for the
1320 distribution nickname (e.g. VERSION_CODENAME=woody).
1322 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
1323 can be set to disable parsing of metadata and the creation
1324 of persistent symlinks for that device.
1326 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
1327 to make them available to logged-in users has been reverted.
1329 * Much of the common code of the various systemd components is now
1330 built into an internal shared library libsystemd-shared-231.so
1331 (incorporating the systemd version number in the name, to be updated
1332 with future releases) that the components link to. This should
1333 decrease systemd footprint both in memory during runtime and on
1334 disk. Note that the shared library is not for public use, and is
1335 neither API not ABI stable, but is likely to change with every new
1336 released update. Packagers need to make sure that binaries
1337 linking to libsystemd-shared.so are updated in step with the
1340 * Configuration for "mkosi" is now part of the systemd
1341 repository. mkosi is a tool to easily build legacy-free OS images,
1342 and is available on github: https://github.com/systemd/mkosi. If
1343 "mkosi" is invoked in the build tree a new raw OS image is generated
1344 incorporating the systemd sources currently being worked on and a
1345 clean, fresh distribution installation. The generated OS image may be
1346 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
1347 UEFI PC. This functionality is particularly useful to easily test
1348 local changes made to systemd in a pristine, defined environment. See
1349 HACKING for details.
1351 * configure learned the --with-support-url= option to specify the
1352 distribution's bugtracker.
1354 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
1355 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
1356 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
1357 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
1358 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
1359 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
1360 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
1361 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
1362 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
1363 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
1364 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
1365 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
1366 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
1367 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
1368 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
1369 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
1370 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
1371 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
1372 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
1374 — Berlin, 2016-07-25
1378 * DNSSEC is now turned on by default in systemd-resolved (in
1379 "allow-downgrade" mode), but may be turned off during compile time by
1380 passing "--with-default-dnssec=no" to "configure" (and of course,
1381 during runtime with DNSSEC= in resolved.conf). We recommend
1382 downstreams to leave this on at least during development cycles and
1383 report any issues with the DNSSEC logic upstream. We are very
1384 interested in collecting feedback about the DNSSEC validator and its
1385 limitations in the wild. Note however, that DNSSEC support is
1386 probably nothing downstreams should turn on in stable distros just
1387 yet, as it might create incompatibilities with a few DNS servers and
1388 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
1389 automatically whenever we detect such incompatible setups, but there
1390 might be systems we do not cover yet. Hence: please help us testing
1391 the DNSSEC code, leave this on where you can, report back, but then
1392 again don't consider turning this on in your stable, LTS or
1393 production release just yet. (Note that you have to enable
1394 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
1395 and its DNSSEC mode for host name resolution from local
1398 * systemd-resolve conveniently resolves DANE records with the --tlsa
1399 option and OPENPGPKEY records with the --openpgp option. It also
1400 supports dumping raw DNS record data via the new --raw= switch.
1402 * systemd-logind will now by default terminate user processes that are
1403 part of the user session scope unit (session-XX.scope) when the user
1404 logs out. This behavior is controlled by the KillUserProcesses=
1405 setting in logind.conf, and the previous default of "no" is now
1406 changed to "yes". This means that user sessions will be properly
1407 cleaned up after, but additional steps are necessary to allow
1408 intentionally long-running processes to survive logout.
1410 While the user is logged in at least once, user@.service is running,
1411 and any service that should survive the end of any individual login
1412 session can be started at a user service or scope using systemd-run.
1413 systemd-run(1) man page has been extended with an example which shows
1414 how to run screen in a scope unit underneath user@.service. The same
1415 command works for tmux.
1417 After the user logs out of all sessions, user@.service will be
1418 terminated too, by default, unless the user has "lingering" enabled.
1419 To effectively allow users to run long-term tasks even if they are
1420 logged out, lingering must be enabled for them. See loginctl(1) for
1421 details. The default polkit policy was modified to allow users to
1422 set lingering for themselves without authentication.
1424 Previous defaults can be restored at compile time by the
1425 --without-kill-user-processes option to "configure".
1427 * systemd-logind gained new configuration settings SessionsMax= and
1428 InhibitorsMax=, both with a default of 8192. It will not register new
1429 user sessions or inhibitors above this limit.
1431 * systemd-logind will now reload configuration on SIGHUP.
1433 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
1434 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
1435 enable. Also, support for the "io" cgroup controller in the unified
1436 hierarchy has been added, so that the "memory", "pids" and "io" are
1437 now the controllers that are supported on the unified hierarchy.
1439 WARNING: it is not possible to use previous systemd versions with
1440 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
1441 is necessary to also update systemd in the initramfs if using the
1442 unified hierarchy. An updated SELinux policy is also required.
1444 * LLDP support has been extended, and both passive (receive-only) and
1445 active (sender) modes are supported. Passive mode ("routers-only") is
1446 enabled by default in systemd-networkd. Active LLDP mode is enabled
1447 by default for containers on the internal network. The "networkctl
1448 lldp" command may be used to list information gathered. "networkctl
1449 status" will also show basic LLDP information on connected peers now.
1451 * The IAID and DUID unique identifier sent in DHCP requests may now be
1452 configured for the system and each .network file managed by
1453 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
1455 * systemd-networkd gained support for configuring proxy ARP support for
1456 each interface, via the ProxyArp= setting in .network files. It also
1457 gained support for configuring the multicast querier feature of
1458 bridge devices, via the new MulticastQuerier= setting in .netdev
1459 files. Similarly, snooping on the IGMP traffic can be controlled
1460 via the new setting MulticastSnooping=.
1462 A new setting PreferredLifetime= has been added for addresses
1463 configured in .network file to configure the lifetime intended for an
1466 The systemd-networkd DHCP server gained the option EmitRouter=, which
1467 defaults to yes, to configure whether the DHCP Option 3 (Router)
1470 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
1471 systemd-socket-activate and installed into /usr/bin. It is now fully
1474 * systemd-journald now uses separate threads to flush changes to disk
1475 when closing journal files, thus reducing impact of slow disk I/O on
1476 logging performance.
1478 * The sd-journal API gained two new calls
1479 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
1480 can be used to open journal files using file descriptors instead of
1481 file or directory paths. sd_journal_open_container() has been
1482 deprecated, sd_journal_open_directory_fd() should be used instead
1483 with the flag SD_JOURNAL_OS_ROOT.
1485 * journalctl learned a new output mode "-o short-unix" that outputs log
1486 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
1487 UTC). It also gained support for a new --no-hostname setting to
1488 suppress the hostname column in the family of "short" output modes.
1490 * systemd-ask-password now optionally skips printing of the password to
1491 stdout with --no-output which can be useful in scripts.
1493 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
1494 (devices tagged with ID_MAKER_TOOL) are now tagged with
1495 "uaccess" and are available to logged in users.
1497 * The DeviceAllow= unit setting now supports specifiers (with "%").
1499 * "systemctl show" gained a new --value switch, which allows print a
1500 only the contents of a specific unit property, without also printing
1501 the property's name. Similar support was added to "show*" verbs
1502 of loginctl and machinectl that output "key=value" lists.
1504 * A new unit type "generated" was added for files dynamically generated
1505 by generator tools. Similarly, a new unit type "transient" is used
1506 for unit files created using the runtime API. "systemctl enable" will
1507 refuse to operate on such files.
1509 * A new command "systemctl revert" has been added that may be used to
1510 revert to the vendor version of a unit file, in case local changes
1511 have been made by adding drop-ins or overriding the unit file.
1513 * "machinectl clean" gained a new verb to automatically remove all or
1514 just hidden container images.
1516 * systemd-tmpfiles gained support for a new line type "e" for emptying
1517 directories, if they exist, without creating them if they don't.
1519 * systemd-nspawn gained support for automatically patching the UID/GIDs
1520 of the owners and the ACLs of all files and directories in a
1521 container tree to match the UID/GID user namespacing range selected
1522 for the container invocation. This mode is enabled via the new
1523 --private-users-chown switch. It also gained support for
1524 automatically choosing a free, previously unused UID/GID range when
1525 starting a container, via the new --private-users=pick setting (which
1526 implies --private-users-chown). Together, these options for the first
1527 time make user namespacing for nspawn containers fully automatic and
1528 thus deployable. The systemd-nspawn@.service template unit file has
1529 been changed to use this functionality by default.
1531 * systemd-nspawn gained a new --network-zone= switch, that allows
1532 creating ad-hoc virtual Ethernet links between multiple containers,
1533 that only exist as long as at least one container referencing them is
1534 running. This allows easy connecting of multiple containers with a
1535 common link that implements an Ethernet broadcast domain. Each of
1536 these network "zones" may be named relatively freely by the user, and
1537 may be referenced by any number of containers, but each container may
1538 only reference one of these "zones". On the lower level, this is
1539 implemented by an automatically managed bridge network interface for
1540 each zone, that is created when the first container referencing its
1541 zone is created and removed when the last one referencing its zone
1544 * The default start timeout may now be configured on the kernel command
1545 line via systemd.default_timeout_start_sec=. It was already
1546 configurable via the DefaultTimeoutStartSec= option in
1547 /etc/systemd/system.conf.
1549 * Socket units gained a new TriggerLimitIntervalSec= and
1550 TriggerLimitBurst= setting to configure a limit on the activation
1551 rate of the socket unit.
1553 * The LimitNICE= setting now optionally takes normal UNIX nice values
1554 in addition to the raw integer limit value. If the specified
1555 parameter is prefixed with "+" or "-" and is in the range -20..19 the
1556 value is understood as UNIX nice value. If not prefixed like this it
1557 is understood as raw RLIMIT_NICE limit.
1559 * Note that the effect of the PrivateDevices= unit file setting changed
1560 slightly with this release: the per-device /dev file system will be
1561 mounted read-only from this version on, and will have "noexec"
1562 set. This (minor) change of behavior might cause some (exceptional)
1563 legacy software to break, when PrivateDevices=yes is set for its
1564 service. Please leave PrivateDevices= off if you run into problems
1567 * systemd-bootchart has been split out to a separate repository:
1568 https://github.com/systemd/systemd-bootchart
1570 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
1571 merged into the kernel in its current form.
1573 * The compatibility libraries libsystemd-daemon.so,
1574 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
1575 which have been deprecated since systemd-209 have been removed along
1576 with the corresponding pkg-config files. All symbols provided by
1577 those libraries are provided by libsystemd.so.
1579 * The Capabilities= unit file setting has been removed (it is ignored
1580 for backwards compatibility). AmbientCapabilities= and
1581 CapabilityBoundingSet= should be used instead.
1583 * A new special target has been added, initrd-root-device.target,
1584 which creates a synchronization point for dependencies of the root
1585 device in early userspace. Initramfs builders must ensure that this
1586 target is now included in early userspace.
1588 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
1589 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
1590 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
1591 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
1592 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
1593 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
1594 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
1595 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
1596 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
1597 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
1598 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
1599 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
1600 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
1601 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
1602 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
1603 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
1604 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
1605 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
1606 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
1607 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
1608 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
1609 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
1610 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
1611 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
1612 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
1615 — Fairfax, 2016-05-21
1619 * The systemd-resolved DNS resolver service has gained a substantial
1620 set of new features, most prominently it may now act as a DNSSEC
1621 validating stub resolver. DNSSEC mode is currently turned off by
1622 default, but is expected to be turned on by default in one of the
1623 next releases. For now, we invite everybody to test the DNSSEC logic
1624 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
1625 service also gained a full set of D-Bus interfaces, including calls
1626 to configure DNS and DNSSEC settings per link (for use by external
1627 network management software). systemd-resolved and systemd-networkd
1628 now distinguish between "search" and "routing" domains. The former
1629 are used to qualify single-label names, the latter are used purely
1630 for routing lookups within certain domains to specific links.
1631 resolved now also synthesizes RRs for all entries from /etc/hosts.
1633 * The systemd-resolve tool (which is a client utility for
1634 systemd-resolved) has been improved considerably and is now fully
1635 supported and documented. Hence it has moved from /usr/lib/systemd to
1638 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
1641 * The coredump collection logic has been reworked: when a coredump is
1642 collected it is now written to disk, compressed and processed
1643 (including stacktrace extraction) from a new instantiated service
1644 systemd-coredump@.service, instead of directly from the
1645 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
1646 processing large coredumps can take up a substantial amount of
1647 resources and time, and this previously happened entirely outside of
1648 systemd's service supervision. With the new logic the core_pattern
1649 hook only does minimal metadata collection before passing off control
1650 to the new instantiated service, which is configured with a time
1651 limit, a nice level and other settings to minimize negative impact on
1652 the rest of the system. Also note that the new logic will honour the
1653 RLIMIT_CORE setting of the crashed process, which now allows users
1654 and processes to turn off coredumping for their processes by setting
1657 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
1658 and all forked processes by default. Previously, PID 1 would leave
1659 the setting at "0" for all processes, as set by the kernel. Note that
1660 the resource limit traditionally has no effect on the generated
1661 coredumps on the system if the /proc/sys/kernel/core_pattern hook
1662 logic is used. Since the limit is now honoured (see above) its
1663 default has been changed so that the coredumping logic is enabled by
1664 default for all processes, while allowing specific opt-out.
1666 * When the stacktrace is extracted from processes of system users, this
1667 is now done as "systemd-coredump" user, in order to sandbox this
1668 potentially security sensitive parsing operation. (Note that when
1669 processing coredumps of normal users this is done under the user ID
1670 of process that crashed, as before.) Packagers should take notice
1671 that it is now necessary to create the "systemd-coredump" system user
1672 and group at package installation time.
1674 * The systemd-activate socket activation testing tool gained support
1675 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
1676 and --seqpacket switches. It also has been extended to support both
1677 new-style and inetd-style file descriptor passing. Use the new
1678 --inetd switch to request inetd-style file descriptor passing.
1680 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
1681 variable, which takes a boolean value. If set to false, ANSI color
1682 output is disabled in the tools even when run on a terminal that
1685 * The VXLAN support in networkd now supports two new settings
1686 DestinationPort= and PortRange=.
1688 * A new systemd.machine_id= kernel command line switch has been added,
1689 that may be used to set the machine ID in /etc/machine-id if it is
1690 not initialized yet. This command line option has no effect if the
1691 file is already initialized.
1693 * systemd-nspawn gained a new --as-pid2 switch that invokes any
1694 specified command line as PID 2 rather than PID 1 in the
1695 container. In this mode PID 1 is a minimal stub init process that
1696 implements the special POSIX and Linux semantics of PID 1 regarding
1697 signal and child process management. Note that this stub init process
1698 is implemented in nspawn itself and requires no support from the
1699 container image. This new logic is useful to support running
1700 arbitrary commands in the container, as normal processes are
1701 generally not prepared to run as PID 1.
1703 * systemd-nspawn gained a new --chdir= switch for setting the current
1704 working directory for the process started in the container.
1706 * "journalctl /dev/sda" will now output all kernel log messages for
1707 specified device from the current boot, in addition to all devices
1708 that are parents of it. This should make log output about devices
1709 pretty useful, as long as kernel drivers attach enough metadata to
1710 the log messages. (The usual SATA drivers do.)
1712 * The sd-journal API gained two new calls
1713 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
1714 that report whether log data from /run or /var has been found.
1716 * journalctl gained a new switch "--fields" that prints all journal
1717 record field names currently in use in the journal. This is backed
1718 by two new sd-journal API calls sd_journal_enumerate_fields() and
1719 sd_journal_restart_fields().
1721 * Most configurable timeouts in systemd now expect an argument of
1722 "infinity" to turn them off, instead of "0" as before. The semantics
1723 from now on is that a timeout of "0" means "now", and "infinity"
1724 means "never". To maintain backwards compatibility, "0" continues to
1725 turn off previously existing timeout settings.
1727 * "systemctl reload-or-try-restart" has been renamed to "systemctl
1728 try-reload-or-restart" to clarify what it actually does: the "try"
1729 logic applies to both reloading and restarting, not just restarting.
1730 The old name continues to be accepted for compatibility.
1732 * On boot-up, when PID 1 detects that the system clock is behind the
1733 release date of the systemd version in use, the clock is now set
1734 to the latter. Previously, this was already done in timesyncd, in order
1735 to avoid running with clocks set to the various clock epochs such as
1736 1902, 1938 or 1970. With this change the logic is now done in PID 1
1737 in addition to timesyncd during early boot-up, so that it is enforced
1738 before the first process is spawned by systemd. Note that the logic
1739 in timesyncd remains, as it is more comprehensive and ensures
1740 clock monotonicity by maintaining a persistent timestamp file in
1741 /var. Since /var is generally not available in earliest boot or the
1742 initrd, this part of the logic remains in timesyncd, and is not done
1745 * Support for tweaking details in net_cls.class_id through the
1746 NetClass= configuration directive has been removed, as the kernel
1747 people have decided to deprecate that controller in cgroup v2.
1748 Userspace tools such as nftables are moving over to setting rules
1749 that are specific to the full cgroup path of a task, which obsoletes
1750 these controllers anyway. The NetClass= directive is kept around for
1751 legacy compatibility reasons. For a more in-depth description of the
1752 kernel change, please refer to the respective upstream commit:
1754 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
1756 * A new service setting RuntimeMaxSec= has been added that may be used
1757 to specify a maximum runtime for a service. If the timeout is hit, the
1758 service is terminated and put into a failure state.
1760 * A new service setting AmbientCapabilities= has been added. It allows
1761 configuration of additional Linux process capabilities that are
1762 passed to the activated processes. This is only available on very
1765 * The process resource limit settings in service units may now be used
1766 to configure hard and soft limits individually.
1768 * The various libsystemd APIs such as sd-bus or sd-event now publicly
1769 expose support for gcc's __attribute__((cleanup())) C extension.
1770 Specifically, for many object destructor functions alternative
1771 versions have been added that have names suffixed with "p" and take a
1772 pointer to a pointer to the object to destroy, instead of just a
1773 pointer to the object itself. This is useful because these destructor
1774 functions may be used directly as parameters to the cleanup
1775 construct. Internally, systemd has been a heavy user of this GCC
1776 extension for a long time, and with this change similar support is
1777 now available to consumers of the library outside of systemd. Note
1778 that by using this extension in your sources compatibility with old
1779 and strictly ANSI compatible C compilers is lost. However, all gcc or
1780 LLVM versions of recent years support this extension.
1782 * Timer units gained support for a new setting RandomizedDelaySec= that
1783 allows configuring some additional randomized delay to the configured
1784 time. This is useful to spread out timer events to avoid load peaks in
1785 clusters or larger setups.
1787 * Calendar time specifications now support sub-second accuracy.
1789 * Socket units now support listening on SCTP and UDP-lite protocol
1792 * The sd-event API now comes with a full set of man pages.
1794 * Older versions of systemd contained experimental support for
1795 compressing journal files and coredumps with the LZ4 compressor that
1796 was not compatible with the lz4 binary (due to API limitations of the
1797 lz4 library). This support has been removed; only support for files
1798 compatible with the lz4 binary remains. This LZ4 logic is now
1799 officially supported and no longer considered experimental.
1801 * The dkr image import logic has been removed again from importd. dkr's
1802 micro-services focus doesn't fit into the machine image focus of
1803 importd, and quickly got out of date with the upstream dkr API.
1805 * Creation of the /run/lock/lockdev/ directory was dropped from
1806 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
1807 been available for many years. If you still need this, you need to
1808 create your own tmpfiles.d config file with:
1810 d /run/lock/lockdev 0775 root lock -
1812 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
1813 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
1814 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
1815 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
1816 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
1817 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
1818 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
1819 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
1820 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
1821 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
1822 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
1823 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
1824 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
1825 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
1826 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
1827 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
1828 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
1829 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
1830 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
1832 — Berlin, 2016-02-11
1836 * A number of properties previously only settable in unit
1837 files are now also available as properties to set when
1838 creating transient units programmatically via the bus, as it
1839 is exposed with systemd-run's --property=
1840 setting. Specifically, these are: SyslogIdentifier=,
1841 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
1842 EnvironmentFile=, ReadWriteDirectories=,
1843 ReadOnlyDirectories=, InaccessibleDirectories=,
1844 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
1846 * When creating transient services via the bus API it is now
1847 possible to pass in a set of file descriptors to use as
1848 STDIN/STDOUT/STDERR for the invoked process.
1850 * Slice units may now be created transiently via the bus APIs,
1851 similar to the way service and scope units may already be
1852 created transiently.
1854 * Wherever systemd expects a calendar timestamp specification
1855 (like in journalctl's --since= and --until= switches) UTC
1856 timestamps are now supported. Timestamps suffixed with "UTC"
1857 are now considered to be in Universal Time Coordinated
1858 instead of the local timezone. Also, timestamps may now
1859 optionally be specified with sub-second accuracy. Both of
1860 these additions also apply to recurring calendar event
1861 specification, such as OnCalendar= in timer units.
1863 * journalctl gained a new "--sync" switch that asks the
1864 journal daemon to write all so far unwritten log messages to
1865 disk and sync the files, before returning.
1867 * systemd-tmpfiles learned two new line types "q" and "Q" that
1868 operate like "v", but also set up a basic btrfs quota
1869 hierarchy when used on a btrfs file system with quota
1872 * tmpfiles' "v", "q" and "Q" will now create a plain directory
1873 instead of a subvolume (even on a btrfs file system) if the
1874 root directory is a plain directory, and not a
1875 subvolume. This should simplify things with certain chroot()
1876 environments which are not aware of the concept of btrfs
1879 * systemd-detect-virt gained a new --chroot switch to detect
1880 whether execution takes place in a chroot() environment.
1882 * CPUAffinity= now takes CPU index ranges in addition to
1885 * The various memory-related resource limit settings (such as
1886 LimitAS=) now understand the usual K, M, G, ... suffixes to
1887 the base of 1024 (IEC). Similar, the time-related resource
1888 limit settings understand the usual min, h, day, ...
1891 * There's a new system.conf setting DefaultTasksMax= to
1892 control the default TasksMax= setting for services and
1893 scopes running on the system. (TasksMax= is the primary
1894 setting that exposes the "pids" cgroup controller on systemd
1895 and was introduced in the previous systemd release.) The
1896 setting now defaults to 512, which means services that are
1897 not explicitly configured otherwise will only be able to
1898 create 512 processes or threads at maximum, from this
1899 version on. Note that this means that thread- or
1900 process-heavy services might need to be reconfigured to set
1901 TasksMax= to a higher value. It is sufficient to set
1902 TasksMax= in these specific unit files to a higher value, or
1903 even "infinity". Similar, there's now a logind.conf setting
1904 UserTasksMax= that defaults to 4096 and limits the total
1905 number of processes or tasks each user may own
1906 concurrently. nspawn containers also have the TasksMax=
1907 value set by default now, to 8192. Note that all of this
1908 only has an effect if the "pids" cgroup controller is
1909 enabled in the kernel. The general benefit of these changes
1910 should be a more robust and safer system, that provides a
1911 certain amount of per-service fork() bomb protection.
1913 * systemd-nspawn gained the new --network-veth-extra= switch
1914 to define additional and arbitrarily-named virtual Ethernet
1915 links between the host and the container.
1917 * A new service execution setting PassEnvironment= has been
1918 added that allows importing select environment variables
1919 from PID1's environment block into the environment block of
1922 * Timer units gained support for a new RemainAfterElapse=
1923 setting which takes a boolean argument. It defaults to on,
1924 exposing behaviour unchanged to previous releases. If set to
1925 off, timer units are unloaded after they elapsed if they
1926 cannot elapse again. This is particularly useful for
1927 transient timer units, which shall not stay around longer
1928 than until they first elapse.
1930 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
1931 default now (the kernel default is 16). This is beneficial
1932 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
1933 allows substantially larger numbers of queued
1934 datagrams. This should increase the capability of systemd to
1935 parallelize boot-up, as logging and sd_notify() are unlikely
1936 to stall execution anymore. If you need to change the value
1937 from the new defaults, use the usual sysctl.d/ snippets.
1939 * The compression framing format used by the journal or
1940 coredump processing has changed to be in line with what the
1941 official LZ4 tools generate. LZ4 compression support in
1942 systemd was considered unsupported previously, as the format
1943 was not compatible with the normal tools. With this release
1944 this has changed now, and it is hence safe for downstream
1945 distributions to turn it on. While not compressing as well
1946 as the XZ, LZ4 is substantially faster, which makes
1947 it a good default choice for the compression logic in the
1948 journal and in coredump handling.
1950 * Any reference to /etc/mtab has been dropped from
1951 systemd. The file has been obsolete since a while, but
1952 systemd refused to work on systems where it was incorrectly
1953 set up (it should be a symlink or non-existent). Please make
1954 sure to update to util-linux 2.27.1 or newer in conjunction
1955 with this systemd release, which also drops any reference to
1956 /etc/mtab. If you maintain a distribution make sure that no
1957 software you package still references it, as this is a
1958 likely source of bugs. There's also a glibc bug pending,
1959 asking for removal of any reference to this obsolete file:
1961 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
1963 Note that only util-linux versions built with
1964 --enable-libmount-force-mountinfo are supported.
1966 * Support for the ".snapshot" unit type has been removed. This
1967 feature turned out to be little useful and little used, and
1968 has now been removed from the core and from systemctl.
1970 * The dependency types RequiresOverridable= and
1971 RequisiteOverridable= have been removed from systemd. They
1972 have been used only very sparingly to our knowledge and
1973 other options that provide a similar effect (such as
1974 systemctl --mode=ignore-dependencies) are much more useful
1975 and commonly used. Moreover, they were only half-way
1976 implemented as the option to control behaviour regarding
1977 these dependencies was never added to systemctl. By removing
1978 these dependency types the execution engine becomes a bit
1979 simpler. Unit files that use these dependencies should be
1980 changed to use the non-Overridable dependency types
1981 instead. In fact, when parsing unit files with these
1982 options, that's what systemd will automatically convert them
1983 too, but it will also warn, asking users to fix the unit
1984 files accordingly. Removal of these dependency types should
1985 only affect a negligible number of unit files in the wild.
1987 * Behaviour of networkd's IPForward= option changed
1988 (again). It will no longer maintain a per-interface setting,
1989 but propagate one way from interfaces where this is enabled
1990 to the global kernel setting. The global setting will be
1991 enabled when requested by a network that is set up, but
1992 never be disabled again. This change was made to make sure
1993 IPv4 and IPv6 behaviour regarding packet forwarding is
1994 similar (as the Linux IPv6 stack does not support
1995 per-interface control of this setting) and to minimize
1998 * In unit files the behaviour of %u, %U, %h, %s has
1999 changed. These specifiers will now unconditionally resolve
2000 to the various user database fields of the user that the
2001 systemd instance is running as, instead of the user
2002 configured in the specific unit via User=. Note that this
2003 effectively doesn't change much, as resolving of these
2004 specifiers was already turned off in the --system instance
2005 of systemd, as we cannot do NSS lookups from PID 1. In the
2006 --user instance of systemd these specifiers where correctly
2007 resolved, but hardly made any sense, since the user instance
2008 lacks privileges to do user switches anyway, and User= is
2009 hence useless. Moreover, even in the --user instance of
2010 systemd behaviour was awkward as it would only take settings
2011 from User= assignment placed before the specifier into
2012 account. In order to unify and simplify the logic around
2013 this the specifiers will now always resolve to the
2014 credentials of the user invoking the manager (which in case
2015 of PID 1 is the root user).
2017 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
2018 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
2019 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
2020 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
2021 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
2022 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
2023 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
2024 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
2025 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
2026 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
2027 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
2028 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
2029 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
2030 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
2033 — Berlin, 2015-11-18
2037 * systemd now depends on util-linux v2.27. More specifically,
2038 the newly added mount monitor feature in libmount now
2039 replaces systemd's former own implementation.
2041 * libmount mandates /etc/mtab not to be regular file, and
2042 systemd now enforces this condition at early boot.
2043 /etc/mtab has been deprecated and warned about for a very
2044 long time, so systems running systemd should already have
2045 stopped having this file around as anything else than a
2046 symlink to /proc/self/mounts.
2048 * Support for the "pids" cgroup controller has been added. It
2049 allows accounting the number of tasks in a cgroup and
2050 enforcing limits on it. This adds two new setting
2051 TasksAccounting= and TasksMax= to each unit, as well as a
2052 global option DefaultTasksAccounting=.
2054 * Support for the "net_cls" cgroup controller has been added.
2055 It allows assigning a net class ID to each task in the
2056 cgroup, which can then be used in firewall rules and traffic
2057 shaping configurations. Note that the kernel netfilter net
2058 class code does not currently work reliably for ingress
2059 packets on unestablished sockets.
2061 This adds a new config directive called NetClass= to CGroup
2062 enabled units. Allowed values are positive numbers for fixed
2063 assignments and "auto" for picking a free value
2066 * 'systemctl is-system-running' now returns 'offline' if the
2067 system is not booted with systemd. This command can now be
2068 used as a substitute for 'systemd-notify --booted'.
2070 * Watchdog timeouts have been increased to 3 minutes for all
2071 in-tree service files. Apparently, disk IO issues are more
2072 frequent than we hoped, and user reported >1 minute waiting
2075 * 'machine-id-commit' functionality has been merged into
2076 'machine-id-setup --commit'. The separate binary has been
2079 * The WorkingDirectory= directive in unit files may now be set
2080 to the special value '~'. In this case, the working
2081 directory is set to the home directory of the user
2082 configured in User=.
2084 * "machinectl shell" will now open the shell in the home
2085 directory of the selected user by default.
2087 * The CrashChVT= configuration file setting is renamed to
2088 CrashChangeVT=, following our usual logic of not
2089 abbreviating unnecessarily. The old directive is still
2090 supported for compat reasons. Also, this directive now takes
2091 an integer value between 1 and 63, or a boolean value. The
2092 formerly supported '-1' value for disabling stays around for
2095 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
2096 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
2097 RootDirectory= properties can now be set for transient
2100 * The systemd-analyze tool gained a new "set-log-target" verb
2101 to change the logging target the system manager logs to
2102 dynamically during runtime. This is similar to how
2103 "systemd-analyze set-log-level" already changes the log
2106 * In nspawn /sys is now mounted as tmpfs, with only a selected
2107 set of subdirectories mounted in from the real sysfs. This
2108 enhances security slightly, and is useful for ensuring user
2109 namespaces work correctly.
2111 * Support for USB FunctionFS activation has been added. This
2112 allows implementation of USB gadget services that are
2113 activated as soon as they are requested, so that they don't
2114 have to run continuously, similar to classic socket
2117 * The "systemctl exit" command now optionally takes an
2118 additional parameter that sets the exit code to return from
2119 the systemd manager when exiting. This is only relevant when
2120 running the systemd user instance, or when running the
2121 system instance in a container.
2123 * sd-bus gained the new API calls sd_bus_path_encode_many()
2124 and sd_bus_path_decode_many() that allow easy encoding and
2125 decoding of multiple identifier strings inside a D-Bus
2126 object path. Another new call sd_bus_default_flush_close()
2127 has been added to flush and close per-thread default
2130 * systemd-cgtop gained support for a -M/--machine= switch to
2131 show the control groups within a certain container only.
2133 * "systemctl kill" gained support for an optional --fail
2134 switch. If specified the requested operation will fail of no
2135 processes have been killed, because the unit had no
2136 processes attached, or similar.
2138 * A new systemd.crash_reboot=1 kernel command line option has
2139 been added that triggers a reboot after crashing. This can
2140 also be set through CrashReboot= in systemd.conf.
2142 * The RuntimeDirectory= setting now understands unit
2143 specifiers like %i or %f.
2145 * A new (still internal) library API sd-ipv4acd has been added,
2146 that implements address conflict detection for IPv4. It's
2147 based on code from sd-ipv4ll, and will be useful for
2148 detecting DHCP address conflicts.
2150 * File descriptors passed during socket activation may now be
2151 named. A new API sd_listen_fds_with_names() is added to
2152 access the names. The default names may be overridden,
2153 either in the .socket file using the FileDescriptorName=
2154 parameter, or by passing FDNAME= when storing the file
2155 descriptors using sd_notify().
2157 * systemd-networkd gained support for:
2159 - Setting the IPv6 Router Advertisement settings via
2160 IPv6AcceptRouterAdvertisements= in .network files.
2162 - Configuring the HelloTimeSec=, MaxAgeSec= and
2163 ForwardDelaySec= bridge parameters in .netdev files.
2165 - Configuring PreferredSource= for static routes in
2168 * The "ask-password" framework used to query for LUKS harddisk
2169 passwords or SSL passwords during boot gained support for
2170 caching passwords in the kernel keyring, if it is
2171 available. This makes sure that the user only has to type in
2172 a passphrase once if there are multiple objects to unlock
2173 with the same one. Previously, such password caching was
2174 available only when Plymouth was used; this moves the
2175 caching logic into the systemd codebase itself. The
2176 "systemd-ask-password" utility gained a new --keyname=
2177 switch to control which kernel keyring key to use for
2178 caching a password in. This functionality is also useful for
2179 enabling display managers such as gdm to automatically
2180 unlock the user's GNOME keyring if its passphrase, the
2181 user's password and the harddisk password are the same, if
2182 gdm-autologin is used.
2184 * When downloading tar or raw images using "machinectl
2185 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
2186 file is now also downloaded, if it is available and stored
2187 next to the image file.
2189 * Units of type ".socket" gained a new boolean setting
2190 Writable= which is only useful in conjunction with
2191 ListenSpecial=. If true, enables opening the specified
2192 special file in O_RDWR mode rather than O_RDONLY mode.
2194 * systemd-rfkill has been reworked to become a singleton
2195 service that is activated through /dev/rfkill on each rfkill
2196 state change and saves the settings to disk. This way,
2197 systemd-rfkill is now compatible with devices that exist
2198 only intermittendly, and even restores state if the previous
2199 system shutdown was abrupt rather than clean.
2201 * The journal daemon gained support for vacuuming old journal
2202 files controlled by the number of files that shall remain,
2203 in addition to the already existing control by size and by
2204 date. This is useful as journal interleaving performance
2205 degrades with too many separate journal files, and allows
2206 putting an effective limit on them. The new setting defaults
2207 to 100, but this may be changed by setting SystemMaxFiles=
2208 and RuntimeMaxFiles= in journald.conf. Also, the
2209 "journalctl" tool gained the new --vacuum-files= switch to
2210 manually vacuum journal files to leave only the specified
2211 number of files in place.
2213 * udev will now create /dev/disk/by-path links for ATA devices
2214 on kernels where that is supported.
2216 * Galician, Serbian, Turkish and Korean translations were added.
2218 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
2219 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
2220 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
2221 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
2222 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
2223 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
2224 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
2225 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
2226 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
2227 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
2228 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
2229 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
2230 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
2231 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
2232 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
2233 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
2234 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
2235 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
2237 — Berlin, 2015-10-07
2241 * The DHCP implementation of systemd-networkd gained a set of
2244 - The DHCP server now supports emitting DNS and NTP
2245 information. It may be enabled and configured via
2246 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
2247 and NTP information is enabled, but no servers are
2248 configured, the corresponding uplink information (if there
2249 is any) is propagated.
2251 - Server and client now support transmission and reception
2252 of timezone information. It can be configured via the
2253 newly introduced network options UseTimezone=,
2254 EmitTimezone=, and Timezone=. Transmission of timezone
2255 information is enabled between host and containers by
2256 default now: the container will change its local timezone
2257 to what the host has set.
2259 - Lease timeouts can now be configured via
2260 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
2262 - The DHCP server improved on the stability of
2263 leases. Clients are more likely to get the same lease
2264 information back, even if the server loses state.
2266 - The DHCP server supports two new configuration options to
2267 control the lease address pool metrics, PoolOffset= and
2270 * The encapsulation limit of tunnels in systemd-networkd may
2271 now be configured via 'EncapsulationLimit='. It allows
2272 modifying the maximum additional levels of encapsulation
2273 that are permitted to be prepended to a packet.
2275 * systemd now supports the concept of user buses replacing
2276 session buses, if used with dbus-1.10 (and enabled via dbus
2277 --enable-user-session). It previously only supported this on
2278 kdbus-enabled systems, and this release expands this to
2279 'dbus-daemon' systems.
2281 * systemd-networkd now supports predictable interface names
2284 * systemd now optionally supports the new Linux kernel
2285 "unified" control group hierarchy. If enabled via the kernel
2286 command-line option 'systemd.unified_cgroup_hierarchy=1',
2287 systemd will try to mount the unified cgroup hierarchy
2288 directly on /sys/fs/cgroup. If not enabled, or not
2289 available, systemd will fall back to the legacy cgroup
2290 hierarchy setup, as before. Host system and containers can
2291 mix and match legacy and unified hierarchies as they
2292 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
2293 environment variable to individually select the hierarchy to
2294 use for executed containers. By default, nspawn will use the
2295 unified hierarchy for the containers if the host uses the
2296 unified hierarchy, and the legacy hierarchy otherwise.
2297 Please note that at this point the unified hierarchy is an
2298 experimental kernel feature and is likely to change in one
2299 of the next kernel releases. Therefore, it should not be
2300 enabled by default in downstream distributions yet. The
2301 minimum required kernel version for the unified hierarchy to
2302 work is 4.2. Note that when the unified hierarchy is used
2303 for the first time delegated access to controllers is
2304 safe. Because of this systemd-nspawn containers will get
2305 access to controllers now, as will systemd user
2306 sessions. This means containers and user sessions may now
2307 manage their own resources, partitioning up what the system
2310 * A new special scope unit "init.scope" has been introduced
2311 that encapsulates PID 1 of the system. It may be used to
2312 determine resource usage and enforce resource limits on PID
2313 1 itself. PID 1 hence moved out of the root of the control
2316 * The cgtop tool gained support for filtering out kernel
2317 threads when counting tasks in a control group. Also, the
2318 count of processes is now recursively summed up by
2319 default. Two options -k and --recursive= have been added to
2320 revert to old behaviour. The tool has also been updated to
2321 work correctly in containers now.
2323 * systemd-nspawn's --bind= and --bind-ro= options have been
2324 extended to allow creation of non-recursive bind mounts.
2326 * libsystemd gained two new calls sd_pid_get_cgroup() and
2327 sd_peer_get_cgroup() which return the control group path of
2328 a process or peer of a connected AF_UNIX socket. This
2329 function call is particularly useful when implementing
2330 delegated subtrees support in the control group hierarchy.
2332 * The "sd-event" event loop API of libsystemd now supports
2333 correct dequeuing of real-time signals, without losing
2336 * When systemd requests a PolicyKit decision when managing
2337 units it will now add additional fields to the request,
2338 including unit name and desired operation. This enables more
2339 powerful PolicyKit policies, that make decisions depending
2340 on these parameters.
2342 * nspawn learnt support for .nspawn settings files, that may
2343 accompany the image files or directories of containers, and
2344 may contain additional settings for the container. This is
2345 an alternative to configuring container parameters via the
2346 nspawn command line.
2348 Contributions from: Cristian Rodríguez, Daniel Mack, David
2349 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
2350 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
2351 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
2352 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
2353 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
2354 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
2355 Andersen, Tom Gundersen, Torstein Husebø
2357 — Berlin, 2015-09-08
2361 * machinectl gained a new verb 'shell' which opens a fresh
2362 shell on the target container or the host. It is similar to
2363 the existing 'login' command of machinectl, but spawns the
2364 shell directly without prompting for username or
2365 password. The pseudo machine '.host' now refers to the local
2366 host and is used by default. Hence, 'machinectl shell' can
2367 be used as replacement for 'su -' which spawns a session as
2368 a fresh systemd unit in a way that is fully isolated from
2369 the originating session.
2371 * systemd-networkd learned to cope with private-zone DHCP
2372 options and allows other programs to query the values.
2374 * SELinux access control when enabling/disabling units is no
2375 longer enforced with this release. The previous
2376 implementation was incorrect, and a new corrected
2377 implementation is not yet available. As unit file operations
2378 are still protected via PolicyKit and D-Bus policy this is
2379 not a security problem. Yet, distributions which care about
2380 optimal SELinux support should probably not stabilize on
2383 * sd-bus gained support for matches of type "arg0has=", that
2384 test for membership of strings in string arrays sent in bus
2387 * systemd-resolved now dumps the contents of its DNS and LLMNR
2388 caches to the logs on reception of the SIGUSR1 signal. This
2389 is useful to debug DNS behaviour.
2391 * The coredumpctl tool gained a new --directory= option to
2392 operate on journal files in a specific directory.
2394 * "systemctl reboot" and related commands gained a new
2395 "--message=" option which may be used to set a free-text
2396 wall message when shutting down or rebooting the
2397 system. This message is also logged, which is useful for
2398 figuring out the reason for a reboot or shutdown a
2401 * The "systemd-resolve-host" tool's -i switch now takes
2402 network interface numbers as alternative to interface names.
2404 * A new unit file setting for services has been introduced:
2405 UtmpMode= allows configuration of how precisely systemd
2406 handles utmp and wtmp entries for the service if this is
2407 enabled. This allows writing services that appear similar to
2408 user sessions in the output of the "w", "who", "last" and
2411 * systemd-resolved will now locally synthesize DNS resource
2412 records for the "localhost" and "gateway" domains as well as
2413 the local hostname. This should ensure that clients querying
2414 RRs via resolved will get similar results as those going via
2415 NSS, if nss-myhostname is enabled.
2417 Contributions from: Alastair Hughes, Alex Crawford, Daniel
2418 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
2419 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
2420 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
2421 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
2422 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
2423 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
2424 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
2425 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
2426 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
2427 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
2428 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
2430 — Berlin, 2015-08-27
2434 * The systemd-efi-boot-generator functionality was merged into
2435 systemd-gpt-auto-generator.
2437 * systemd-networkd now supports Group Policy for vxlan
2438 devices. It can be enabled via the new boolean configuration
2439 option called 'GroupPolicyExtension='.
2441 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
2442 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
2443 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
2445 — Berlin, 2015-07-31
2449 * The python-systemd code has been removed from the systemd repository.
2450 A new repository has been created which accommodates the code from
2451 now on, and we kindly ask distributions to create a separate package
2452 for this: https://github.com/systemd/python-systemd
2454 * The systemd daemon will now reload its main configuration
2455 (/etc/systemd/system.conf) on daemon-reload.
2457 * sd-dhcp now exposes vendor specific extensions via
2458 sd_dhcp_lease_get_vendor_specific().
2460 * systemd-networkd gained a number of new configuration options.
2462 - A new boolean configuration option for TAP devices called
2463 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
2464 device, thus allowing to send and receive GSO packets.
2466 - A new tunnel configuration option called 'CopyDSCP='.
2467 If enabled, the DSCP field of ip6 tunnels is copied into the
2468 decapsulated packet.
2470 - A set of boolean bridge configuration options were added.
2471 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
2472 and 'UnicastFlood=' are now parsed by networkd and applied to the
2473 respective bridge link device via the respective IFLA_BRPORT_*
2476 - A new string configuration option to override the hostname sent
2477 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
2478 is true, networkd will use the configured hostname instead of the
2479 system hostname when sending DHCP requests.
2481 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
2482 networkd will configure the IPv6 flow-label of the tunnel device
2483 according to RFC2460.
2485 - The 'macvtap' virtual network devices are now supported, similar to
2486 the already supported 'macvlan' devices.
2488 * systemd-resolved now implements RFC5452 to improve resilience against
2489 cache poisoning. Additionally, source port randomization is enabled
2490 by default to further protect against DNS spoofing attacks.
2492 * nss-mymachines now supports translating UIDs and GIDs of running
2493 containers with user-namespaces enabled. If a container 'foo'
2494 translates a host uid 'UID' to the container uid 'TUID', then
2495 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
2496 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
2497 mapped as 'vg-foo-TGID'.
2499 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
2500 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
2501 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
2502 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
2503 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
2504 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
2505 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
2506 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
2507 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
2508 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
2510 — Berlin, 2015-07-29
2514 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
2515 There are no known issues with current sysfs, and udev does not need
2516 or should be used to work around such bugs.
2518 * udev does no longer enable USB HID power management. Several reports
2519 indicate, that some devices cannot handle that setting.
2521 * The udev accelerometer helper was removed. The functionality
2522 is now fully included in iio-sensor-proxy. But this means,
2523 older iio-sensor-proxy versions will no longer provide
2524 accelerometer/orientation data with this systemd version.
2525 Please upgrade iio-sensor-proxy to version 1.0.
2527 * networkd gained a new configuration option IPv6PrivacyExtensions=
2528 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
2529 for Stateless Address") on selected networks.
2531 * For the sake of fewer build-time dependencies and less code in the
2532 main repository, the python bindings are about to be removed in the
2533 next release. A new repository has been created which accommodates
2534 the code from now on, and we kindly ask distributions to create a
2535 separate package for this. The removal will take place in v223.
2537 https://github.com/systemd/python-systemd
2539 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
2540 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
2541 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
2542 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
2543 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
2544 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
2545 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
2546 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
2547 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
2548 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
2550 — Berlin, 2015-07-07
2554 * The sd-bus.h and sd-event.h APIs have now been declared
2555 stable and have been added to the official interface of
2556 libsystemd.so. sd-bus implements an alternative D-Bus client
2557 library, that is relatively easy to use, very efficient and
2558 supports both classic D-Bus as well as kdbus as transport
2559 backend. sd-event is a generic event loop abstraction that
2560 is built around Linux epoll, but adds features such as event
2561 prioritization or efficient timer handling. Both APIs are good
2562 choices for C programs looking for a bus and/or event loop
2563 implementation that is minimal and does not have to be
2564 portable to other kernels.
2566 * kdbus support is no longer compile-time optional. It is now
2567 always built-in. However, it can still be disabled at
2568 runtime using the kdbus=0 kernel command line setting, and
2569 that setting may be changed to default to off, by specifying
2570 --disable-kdbus at build-time. Note though that the kernel
2571 command line setting has no effect if the kdbus.ko kernel
2572 module is not installed, in which case kdbus is (obviously)
2573 also disabled. We encourage all downstream distributions to
2574 begin testing kdbus by adding it to the kernel images in the
2575 development distributions, and leaving kdbus support in
2578 * The minimal required util-linux version has been bumped to
2581 * Support for chkconfig (--enable-chkconfig) was removed in
2582 favor of calling an abstraction tool
2583 /lib/systemd/systemd-sysv-install. This needs to be
2584 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
2585 in README for details.
2587 * If there's a systemd unit and a SysV init script for the
2588 same service name, and the user executes "systemctl enable"
2589 for it (or a related call), then this will now enable both
2590 (or execute the related operation on both), not just the
2593 * The libudev API documentation has been converted from gtkdoc
2596 * gudev has been removed from the systemd tree, it is now an
2599 * The systemd-cgtop tool learnt a new --raw switch to generate
2600 "raw" (machine parsable) output.
2602 * networkd's IPForwarding= .network file setting learnt the
2603 new setting "kernel", which ensures that networkd does not
2604 change the IP forwarding sysctl from the default kernel
2607 * The systemd-logind bus API now exposes a new boolean
2608 property "Docked" that reports whether logind considers the
2609 system "docked", i.e. connected to a docking station or not.
2611 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
2612 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
2613 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
2614 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
2615 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
2616 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
2617 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
2618 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
2619 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
2620 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
2621 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
2622 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
2623 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
2624 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
2625 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
2626 Fink, Zbigniew Jędrzejewski-Szmek
2628 — Berlin, 2015-06-19
2632 * The gudev library has been extracted into a separate repository
2633 available at: https://git.gnome.org/browse/libgudev/
2634 It is now managed as part of the Gnome project. Distributions
2635 are recommended to pass --disable-gudev to systemd and use
2636 gudev from the Gnome project instead. gudev is still included
2637 in systemd, for now. It will be removed soon, though. Please
2638 also see the announcement-thread on systemd-devel:
2639 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
2641 * systemd now exposes a CPUUsageNSec= property for each
2642 service unit on the bus, that contains the overall consumed
2643 CPU time of a service (the sum of what each process of the
2644 service consumed). This value is only available if
2645 CPUAccounting= is turned on for a service, and is then shown
2646 in the "systemctl status" output.
2648 * Support for configuring alternative mappings of the old SysV
2649 runlevels to systemd targets has been removed. They are now
2650 hardcoded in a way that runlevels 2, 3, 4 all map to
2651 multi-user.target and 5 to graphical.target (which
2652 previously was already the default behaviour).
2654 * The auto-mounter logic gained support for mount point
2655 expiry, using a new TimeoutIdleSec= setting in .automount
2656 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
2658 * The EFI System Partition (ESP) as mounted to /boot by
2659 systemd-efi-boot-generator will now be unmounted
2660 automatically after 2 minutes of not being used. This should
2661 minimize the risk of ESP corruptions.
2663 * New /etc/fstab options x-systemd.requires= and
2664 x-systemd.requires-mounts-for= are now supported to express
2665 additional dependencies for mounts. This is useful for
2666 journalling file systems that support external journal
2667 devices or overlay file systems that require underlying file
2668 systems to be mounted.
2670 * systemd does not support direct live-upgrades (via systemctl
2671 daemon-reexec) from versions older than v44 anymore. As no
2672 distribution we are aware of shipped such old versions in a
2673 stable release this should not be problematic.
2675 * When systemd forks off a new per-connection service instance
2676 it will now set the $REMOTE_ADDR environment variable to the
2677 remote IP address, and $REMOTE_PORT environment variable to
2678 the remote IP port. This behaviour is similar to the
2679 corresponding environment variables defined by CGI.
2681 * systemd-networkd gained support for uplink failure
2682 detection. The BindCarrier= option allows binding interface
2683 configuration dynamically to the link sense of other
2684 interfaces. This is useful to achieve behaviour like in
2687 * systemd-networkd gained support for configuring the DHCP
2688 client identifier to use when requesting leases.
2690 * systemd-networkd now has a per-network UseNTP= option to
2691 configure whether NTP server information acquired via DHCP
2692 is passed on to services like systemd-timesyncd.
2694 * systemd-networkd gained support for vti6 tunnels.
2696 * Note that systemd-networkd manages the sysctl variable
2697 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
2698 it is configured for since v219. The variable controls IP
2699 forwarding, and is a per-interface alternative to the global
2700 /proc/sys/net/ipv[46]/ip_forward. This setting is
2701 configurable in the IPForward= option, which defaults to
2702 "no". This means if networkd is used for an interface it is
2703 no longer sufficient to set the global sysctl option to turn
2704 on IP forwarding! Instead, the .network file option
2705 IPForward= needs to be turned on! Note that the
2706 implementation of this behaviour was broken in v219 and has
2709 * Many bonding and vxlan options are now configurable in
2712 * systemd-nspawn gained a new --property= setting to set unit
2713 properties for the container scope. This is useful for
2714 setting resource parameters (e.g. "CPUShares=500") on
2715 containers started from the command line.
2717 * systemd-nspawn gained a new --private-users= switch to make
2718 use of user namespacing available on recent Linux kernels.
2720 * systemd-nspawn may now be called as part of a shell pipeline
2721 in which case the pipes used for stdin and stdout are passed
2722 directly to the process invoked in the container, without
2723 indirection via a pseudo tty.
2725 * systemd-nspawn gained a new switch to control the UNIX
2726 signal to use when killing the init process of the container
2729 * systemd-nspawn gained a new --overlay= switch for mounting
2730 overlay file systems into the container using the new kernel
2733 * When a container image is imported via systemd-importd and
2734 the host file system is not btrfs, a loopback block device
2735 file is created in /var/lib/machines.raw with a btrfs file
2736 system inside. It is then mounted to /var/lib/machines to
2737 enable btrfs features for container management. The loopback
2738 file and btrfs file system is grown as needed when container
2739 images are imported via systemd-importd.
2741 * systemd-machined/systemd-importd gained support for btrfs
2742 quota, to enforce container disk space limits on disk. This
2743 is exposed in "machinectl set-limit".
2745 * systemd-importd now can import containers from local .tar,
2746 .raw and .qcow2 images, and export them to .tar and .raw. It
2747 can also import dkr v2 images now from the network (on top
2750 * systemd-importd gained support for verifying downloaded
2751 images with gpg2 (previously only gpg1 was supported).
2753 * systemd-machined, systemd-logind, systemd: most bus calls
2754 are now accessible to unprivileged processes via
2755 PolicyKit. Also, systemd-logind will now allow users to kill
2756 their own sessions without further privileges or
2759 * systemd-shutdownd has been removed. This service was
2760 previously responsible for implementing scheduled shutdowns
2761 as exposed in /usr/bin/shutdown's time parameter. This
2762 functionality has now been moved into systemd-logind and is
2763 accessible via a bus interface.
2765 * "systemctl reboot" gained a new switch --firmware-setup that
2766 can be used to reboot into the EFI firmware setup, if that
2767 is available. systemd-logind now exposes an API on the bus
2768 to trigger such reboots, in case graphical desktop UIs want
2769 to cover this functionality.
2771 * "systemctl enable", "systemctl disable" and "systemctl mask"
2772 now support a new "--now" switch. If specified the units
2773 that are enabled will also be started, and the ones
2774 disabled/masked also stopped.
2776 * The Gummiboot EFI boot loader tool has been merged into
2777 systemd, and renamed to "systemd-boot". The bootctl tool has been
2778 updated to support systemd-boot.
2780 * An EFI kernel stub has been added that may be used to create
2781 kernel EFI binaries that contain not only the actual kernel,
2782 but also an initrd, boot splash, command line and OS release
2783 information. This combined binary can then be signed as a
2784 single image, so that the firmware can verify it all in one
2785 step. systemd-boot has special support for EFI binaries created
2786 like this and can extract OS release information from them
2787 and show them in the boot menu. This functionality is useful
2788 to implement cryptographically verified boot schemes.
2790 * Optional support has been added to systemd-fsck to pass
2791 fsck's progress report to an AF_UNIX socket in the file
2794 * udev will no longer create device symlinks for all block
2795 devices by default. A blacklist for excluding special block
2796 devices from this logic has been turned into a whitelist
2797 that requires picking block devices explicitly that require
2800 * A new (currently still internal) API sd-device.h has been
2801 added to libsystemd. This modernized API is supposed to
2802 replace libudev eventually. In fact, already much of libudev
2803 is now just a wrapper around sd-device.h.
2805 * A new hwdb database for storing metadata about pointing
2806 stick devices has been added.
2808 * systemd-tmpfiles gained support for setting file attributes
2809 similar to the "chattr" tool with new 'h' and 'H' lines.
2811 * systemd-journald will no longer unconditionally set the
2812 btrfs NOCOW flag on new journal files. This is instead done
2813 with tmpfiles snippet using the new 'h' line type. This
2814 allows easy disabling of this logic, by masking the
2815 journal-nocow.conf tmpfiles file.
2817 * systemd-journald will now translate audit message types to
2818 human readable identifiers when writing them to the
2819 journal. This should improve readability of audit messages.
2821 * The LUKS logic gained support for the offset= and skip=
2822 options in /etc/crypttab, as previously implemented by
2825 * /usr/lib/os-release gained a new optional field VARIANT= for
2826 distributions that support multiple variants (such as a
2827 desktop edition, a server edition, ...)
2829 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
2830 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
2831 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
2832 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
2833 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
2834 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
2835 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
2836 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
2837 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
2838 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
2839 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
2840 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
2841 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
2842 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
2843 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
2844 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
2845 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
2846 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
2847 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
2848 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
2849 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
2850 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
2851 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
2852 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
2853 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
2854 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
2855 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
2857 — Berlin, 2015-05-22
2861 * Introduce a new API "sd-hwdb.h" for querying the hardware
2862 metadata database. With this minimal interface one can query
2863 and enumerate the udev hwdb, decoupled from the old libudev
2864 library. libudev's interface for this is now only a wrapper
2865 around sd-hwdb. A new tool systemd-hwdb has been added to
2866 interface with and update the database.
2868 * When any of systemd's tools copies files (for example due to
2869 tmpfiles' C lines) a btrfs reflink will attempted first,
2870 before bytewise copying is done.
2872 * systemd-nspawn gained a new --ephemeral switch. When
2873 specified a btrfs snapshot is taken of the container's root
2874 directory, and immediately removed when the container
2875 terminates again. Thus, a container can be started whose
2876 changes never alter the container's root directory, and are
2877 lost on container termination. This switch can also be used
2878 for starting a container off the root file system of the
2879 host without affecting the host OS. This switch is only
2880 available on btrfs file systems.
2882 * systemd-nspawn gained a new --template= switch. It takes the
2883 path to a container tree to use as template for the tree
2884 specified via --directory=, should that directory be
2885 missing. This allows instantiating containers dynamically,
2886 on first run. This switch is only available on btrfs file
2889 * When a .mount unit refers to a mount point on which multiple
2890 mounts are stacked, and the .mount unit is stopped all of
2891 the stacked mount points will now be unmounted until no
2892 mount point remains.
2894 * systemd now has an explicit notion of supported and
2895 unsupported unit types. Jobs enqueued for unsupported unit
2896 types will now fail with an "unsupported" error code. More
2897 specifically .swap, .automount and .device units are not
2898 supported in containers, .busname units are not supported on
2899 non-kdbus systems. .swap and .automount are also not
2900 supported if their respective kernel compile time options
2903 * machinectl gained support for two new "copy-from" and
2904 "copy-to" commands for copying files from a running
2905 container to the host or vice versa.
2907 * machinectl gained support for a new "bind" command to bind
2908 mount host directories into local containers. This is
2909 currently only supported for nspawn containers.
2911 * networkd gained support for configuring bridge forwarding
2912 database entries (fdb) from .network files.
2914 * A new tiny daemon "systemd-importd" has been added that can
2915 download container images in tar, raw, qcow2 or dkr formats,
2916 and make them available locally in /var/lib/machines, so
2917 that they can run as nspawn containers. The daemon can GPG
2918 verify the downloads (not supported for dkr, since it has no
2919 provisions for verifying downloads). It will transparently
2920 decompress bz2, xz, gzip compressed downloads if necessary,
2921 and restore sparse files on disk. The daemon uses privilege
2922 separation to ensure the actual download logic runs with
2923 fewer privileges than the daemon itself. machinectl has
2924 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
2925 make the functionality of importd available to the
2926 user. With this in place the Fedora and Ubuntu "Cloud"
2927 images can be downloaded and booted as containers unmodified
2928 (the Fedora images lack the appropriate GPG signature files
2929 currently, so they cannot be verified, but this will change
2930 soon, hopefully). Note that downloading images is currently
2931 only fully supported on btrfs.
2933 * machinectl is now able to list container images found in
2934 /var/lib/machines, along with some metadata about sizes of
2935 disk and similar. If the directory is located on btrfs and
2936 quota is enabled, this includes quota display. A new command
2937 "image-status" has been added that shows additional
2938 information about images.
2940 * machinectl is now able to clone container images
2941 efficiently, if the underlying file system (btrfs) supports
2942 it, with the new "machinectl clone" command. It also
2943 gained commands for renaming and removing images, as well as
2944 marking them read-only or read-write (supported also on
2945 legacy file systems).
2947 * networkd gained support for collecting LLDP network
2948 announcements, from hardware that supports this. This is
2949 shown in networkctl output.
2951 * systemd-run gained support for a new -t (--pty) switch for
2952 invoking a binary on a pty whose input and output is
2953 connected to the invoking terminal. This allows executing
2954 processes as system services while interactively
2955 communicating with them via the terminal. Most interestingly
2956 this is supported across container boundaries. Invoking
2957 "systemd-run -t /bin/bash" is an alternative to running a
2958 full login session, the difference being that the former
2959 will not register a session, nor go through the PAM session
2962 * tmpfiles gained support for a new "v" line type for creating
2963 btrfs subvolumes. If the underlying file system is a legacy
2964 file system, this automatically degrades to creating a
2965 normal directory. Among others /var/lib/machines is now
2966 created like this at boot, should it be missing.
2968 * The directory /var/lib/containers/ has been deprecated and
2969 been replaced by /var/lib/machines. The term "machines" has
2970 been used in the systemd context as generic term for both
2971 VMs and containers, and hence appears more appropriate for
2972 this, as the directory can also contain raw images bootable
2975 * systemd-nspawn when invoked with -M but without --directory=
2976 or --image= is now capable of searching for the container
2977 root directory, subvolume or disk image automatically, in
2978 /var/lib/machines. systemd-nspawn@.service has been updated
2979 to make use of this, thus allowing it to be used for raw
2982 * A new machines.target unit has been introduced that is
2983 supposed to group all containers/VMs invoked as services on
2984 the system. systemd-nspawn@.service has been updated to
2985 integrate with that.
2987 * machinectl gained a new "start" command, for invoking a
2988 container as a service. "machinectl start foo" is mostly
2989 equivalent to "systemctl start systemd-nspawn@foo.service",
2990 but handles escaping in a nicer way.
2992 * systemd-nspawn will now mount most of the cgroupfs tree
2993 read-only into each container, with the exception of the
2994 container's own subtree in the name=systemd hierarchy.
2996 * journald now sets the special FS_NOCOW file flag for its
2997 journal files. This should improve performance on btrfs, by
2998 avoiding heavy fragmentation when journald's write-pattern
2999 is used on COW file systems. It degrades btrfs' data
3000 integrity guarantees for the files to the same levels as for
3001 ext3/ext4 however. This should be OK though as journald does
3002 its own data integrity checks and all its objects are
3003 checksummed on disk. Also, journald should handle btrfs disk
3004 full events a lot more gracefully now, by processing SIGBUS
3005 errors, and not relying on fallocate() anymore.
3007 * When journald detects that journal files it is writing to
3008 have been deleted it will immediately start new journal
3011 * systemd now provides a way to store file descriptors
3012 per-service in PID 1. This is useful for daemons to ensure
3013 that fds they require are not lost during a daemon
3014 restart. The fds are passed to the daemon on the next
3015 invocation in the same way socket activation fds are
3016 passed. This is now used by journald to ensure that the
3017 various sockets connected to all the system's stdout/stderr
3018 are not lost when journald is restarted. File descriptors
3019 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
3020 an extension to sd_notify(). Note that a limit is enforced
3021 on the number of fds a service can store in PID 1, and it
3022 defaults to 0, so that no fds may be stored, unless this is
3023 explicitly turned on.
3025 * The default TERM variable to use for units connected to a
3026 terminal, when no other value is explicitly is set is now
3027 vt220 rather than vt102. This should be fairly safe still,
3028 but allows PgUp/PgDn work.
3030 * The /etc/crypttab option header= as known from Debian is now
3033 * "loginctl user-status" and "loginctl session-status" will
3034 now show the last 10 lines of log messages of the
3035 user/session following the status output. Similar,
3036 "machinectl status" will show the last 10 log lines
3037 associated with a virtual machine or container
3038 service. (Note that this is usually not the log messages
3039 done in the VM/container itself, but simply what the
3040 container manager logs. For nspawn this includes all console
3043 * "loginctl session-status" without further argument will now
3044 show the status of the session of the caller. Similar,
3045 "lock-session", "unlock-session", "activate",
3046 "enable-linger", "disable-linger" may now be called without
3047 session/user parameter in which case they apply to the
3048 caller's session/user.
3050 * An X11 session scriptlet is now shipped that uploads
3051 $DISPLAY and $XAUTHORITY into the environment of the systemd
3052 --user daemon if a session begins. This should improve
3053 compatibility with X11 enabled applications run as systemd
3056 * Generators are now subject to masking via /etc and /run, the
3057 same way as unit files.
3059 * networkd .network files gained support for configuring
3060 per-link IPv4/IPv6 packet forwarding as well as IPv4
3061 masquerading. This is by default turned on for veth links to
3062 containers, as registered by systemd-nspawn. This means that
3063 nspawn containers run with --network-veth will now get
3064 automatic routed access to the host's networks without any
3065 further configuration or setup, as long as networkd runs on
3068 * systemd-nspawn gained the --port= (-p) switch to expose TCP
3069 or UDP posts of a container on the host. With this in place
3070 it is possible to run containers with private veth links
3071 (--network-veth), and have their functionality exposed on
3072 the host as if their services were running directly on the
3075 * systemd-nspawn's --network-veth switch now gained a short
3076 version "-n", since with the changes above it is now truly
3077 useful out-of-the-box. The systemd-nspawn@.service has been
3078 updated to make use of it too by default.
3080 * systemd-nspawn will now maintain a per-image R/W lock, to
3081 ensure that the same image is not started more than once
3082 writable. (It's OK to run an image multiple times
3083 simultaneously in read-only mode.)
3085 * systemd-nspawn's --image= option is now capable of
3086 dissecting and booting MBR and GPT disk images that contain
3087 only a single active Linux partition. Previously it
3088 supported only GPT disk images with proper GPT type
3089 IDs. This allows running cloud images from major
3090 distributions directly with systemd-nspawn, without
3093 * In addition to collecting mouse dpi data in the udev
3094 hardware database, there's now support for collecting angle
3095 information for mouse scroll wheels. The database is
3096 supposed to guarantee similar scrolling behavior on mice
3097 that it knows about. There's also support for collecting
3098 information about Touchpad types.
3100 * udev's input_id built-in will now also collect touch screen
3101 dimension data and attach it to probed devices.
3103 * /etc/os-release gained support for a Distribution Privacy
3106 * networkd gained support for creating "ipvlan", "gretap",
3107 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
3109 * systemd-tmpfiles gained support for "a" lines for setting
3112 * systemd-nspawn will now mount /tmp in the container to
3113 tmpfs, automatically.
3115 * systemd now exposes the memory.usage_in_bytes cgroup
3116 attribute and shows it for each service in the "systemctl
3117 status" output, if available.
3119 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
3120 immediate reboot is triggered. This useful if shutdown is
3121 hung and is unable to complete, to expedite the
3122 operation. Note that this kind of reboot will still unmount
3123 all file systems, and hence should not result in fsck being
3126 * A .device unit for an optical block device will now be
3127 considered active only when a medium is in the drive. Also,
3128 mount units are now bound to their backing devices thus
3129 triggering automatic unmounting when devices become
3130 unavailable. With this in place systemd will now
3131 automatically unmount left-over mounts when a CD-ROM is
3132 ejected or an USB stick is yanked from the system.
3134 * networkd-wait-online now has support for waiting for
3135 specific interfaces only (with globbing), and for giving up
3136 after a configurable timeout.
3138 * networkd now exits when idle. It will be automatically
3139 restarted as soon as interfaces show up, are removed or
3140 change state. networkd will stay around as long as there is
3141 at least one DHCP state machine or similar around, that keep
3144 * networkd may now configure IPv6 link-local addressing in
3145 addition to IPv4 link-local addressing.
3147 * The IPv6 "token" for use in SLAAC may now be configured for
3148 each .network interface in networkd.
3150 * Routes configured with networkd may now be assigned a scope
3153 * networkd's [Match] sections now support globbing and lists
3154 of multiple space-separated matches per item.
3156 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
3157 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
3158 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
3159 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
3160 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
3161 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
3162 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
3163 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
3164 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
3165 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
3166 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
3167 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
3168 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
3169 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
3170 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
3171 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
3172 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
3173 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
3174 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
3175 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
3176 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
3177 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
3178 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
3179 Hoffmann, Zbigniew Jędrzejewski-Szmek
3181 — Berlin, 2015-02-16
3185 * When querying unit file enablement status (for example via
3186 "systemctl is-enabled"), a new state "indirect" is now known
3187 which indicates that a unit might not be enabled itself, but
3188 another unit listed in its Also= setting might be.
3190 * Similar to the various existing ConditionXYZ= settings for
3191 units, there are now matching AssertXYZ= settings. While
3192 failing conditions cause a unit to be skipped, but its job
3193 to succeed, failing assertions declared like this will cause
3194 a unit start operation and its job to fail.
3196 * hostnamed now knows a new chassis type "embedded".
3198 * systemctl gained a new "edit" command. When used on a unit
3199 file, this allows extending unit files with .d/ drop-in
3200 configuration snippets or editing the full file (after
3201 copying it from /usr/lib to /etc). This will invoke the
3202 user's editor (as configured with $EDITOR), and reload the
3203 modified configuration after editing.
3205 * "systemctl status" now shows the suggested enablement state
3206 for a unit, as declared in the (usually vendor-supplied)
3207 system preset files.
3209 * nss-myhostname will now resolve the single-label host name
3210 "gateway" to the locally configured default IP routing
3211 gateways, ordered by their metrics. This assigns a stable
3212 name to the used gateways, regardless which ones are
3213 currently configured. Note that the name will only be
3214 resolved after all other name sources (if nss-myhostname is
3215 configured properly) and should hence not negatively impact
3216 systems that use the single-label host name "gateway" in
3219 * systemd-inhibit now allows filtering by mode when listing
3222 * Scope and service units gained a new "Delegate" boolean
3223 property, which, when set, allows processes running inside the
3224 unit to further partition resources. This is primarily
3225 useful for systemd user instances as well as container
3228 * journald will now pick up audit messages directly from
3229 the kernel, and log them like any other log message. The
3230 audit fields are split up and fully indexed. This means that
3231 journalctl in many ways is now a (nicer!) alternative to
3232 ausearch, the traditional audit client. Note that this
3233 implements only a minimal audit client. If you want the
3234 special audit modes like reboot-on-log-overflow, please use
3235 the traditional auditd instead, which can be used in
3236 parallel to journald.
3238 * The ConditionSecurity= unit file option now understands the
3239 special string "audit" to check whether auditing is
3242 * journalctl gained two new commands --vacuum-size= and
3243 --vacuum-time= to delete old journal files until the
3244 remaining ones take up no more than the specified size on disk,
3245 or are not older than the specified time.
3247 * A new, native PPPoE library has been added to sd-network,
3248 systemd's library of light-weight networking protocols. This
3249 library will be used in a future version of networkd to
3250 enable PPPoE communication without an external pppd daemon.
3252 * The busctl tool now understands a new "capture" verb that
3253 works similar to "monitor", but writes a packet capture
3254 trace to STDOUT that can be redirected to a file which is
3255 compatible with libcap's capture file format. This can then
3256 be loaded in Wireshark and similar tools to inspect bus
3259 * The busctl tool now understands a new "tree" verb that shows
3260 the object trees of a specific service on the bus, or of all
3263 * The busctl tool now understands a new "introspect" verb that
3264 shows all interfaces and members of objects on the bus,
3265 including their signature and values. This is particularly
3266 useful to get more information about bus objects shown by
3267 the new "busctl tree" command.
3269 * The busctl tool now understands new verbs "call",
3270 "set-property" and "get-property" for invoking bus method
3271 calls, setting and getting bus object properties in a
3274 * busctl gained a new --augment-creds= argument that controls
3275 whether the tool shall augment credential information it
3276 gets from the bus with data from /proc, in a possibly
3279 * nspawn's --link-journal= switch gained two new values
3280 "try-guest" and "try-host" that work like "guest" and
3281 "host", but do not fail if the host has no persistent
3282 journalling enabled. -j is now equivalent to
3283 --link-journal=try-guest.
3285 * macvlan network devices created by nspawn will now have
3286 stable MAC addresses.
3288 * A new SmackProcessLabel= unit setting has been added, which
3289 controls the SMACK security label processes forked off by
3290 the respective unit shall use.
3292 * If compiled with --enable-xkbcommon, systemd-localed will
3293 verify x11 keymap settings by compiling the given keymap. It
3294 will spew out warnings if the compilation fails. This
3295 requires libxkbcommon to be installed.
3297 * When a coredump is collected, a larger number of metadata
3298 fields is now collected and included in the journal records
3299 created for it. More specifically, control group membership,
3300 environment variables, memory maps, working directory,
3301 chroot directory, /proc/$PID/status, and a list of open file
3302 descriptors is now stored in the log entry.
3304 * The udev hwdb now contains DPI information for mice. For
3307 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
3309 * All systemd programs that read standalone configuration
3310 files in /etc now also support a corresponding series of
3311 .conf.d configuration directories in /etc/, /run/,
3312 /usr/local/lib/, /usr/lib/, and (if configured with
3313 --enable-split-usr) /lib/. In particular, the following
3314 configuration files now have corresponding configuration
3315 directories: system.conf user.conf, logind.conf,
3316 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
3317 resolved.conf, timesyncd.conf, journal-remote.conf, and
3318 journal-upload.conf. Note that distributions should use the
3319 configuration directories in /usr/lib/; the directories in
3320 /etc/ are reserved for the system administrator.
3322 * systemd-rfkill will no longer take the rfkill device name
3323 into account when storing rfkill state on disk, as the name
3324 might be dynamically assigned and not stable. Instead, the
3325 ID_PATH udev variable combined with the rfkill type (wlan,
3326 bluetooth, ...) is used.
3328 * A new service systemd-machine-id-commit.service has been
3329 added. When used on systems where /etc is read-only during
3330 boot, and /etc/machine-id is not initialized (but an empty
3331 file), this service will copy the temporary machine ID
3332 created as replacement into /etc after the system is fully
3333 booted up. This is useful for systems that are freshly
3334 installed with a non-initialized machine ID, but should get
3335 a fixed machine ID for subsequent boots.
3337 * networkd's .netdev files now provide a large set of
3338 configuration parameters for VXLAN devices. Similarly, the
3339 bridge port cost parameter is now configurable in .network
3340 files. There's also new support for configuring IP source
3341 routing. networkd .link files gained support for a new
3342 OriginalName= match that is useful to match against the
3343 original interface name the kernel assigned. .network files
3344 may include MTU= and MACAddress= fields for altering the MTU
3345 and MAC address while being connected to a specific network
3348 * The LUKS logic gained supported for configuring
3349 UUID-specific key files. There's also new support for naming
3350 LUKS device from the kernel command line, using the new
3351 luks.name= argument.
3353 * Timer units may now be transiently created via the bus API
3354 (this was previously already available for scope and service
3355 units). In addition it is now possible to create multiple
3356 transient units at the same time with a single bus call. The
3357 "systemd-run" tool has been updated to make use of this for
3358 running commands on a specified time, in at(1)-style.
3360 * tmpfiles gained support for "t" lines, for assigning
3361 extended attributes to files. Among other uses this may be
3362 used to assign SMACK labels to files.
3364 Contributions from: Alin Rauta, Alison Chaiken, Andrej
3365 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
3366 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
3367 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
3368 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
3369 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
3370 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
3371 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
3372 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
3373 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
3374 Michael Biebl, Michael Chapman, Michael Marineau, Michal
3375 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
3376 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
3377 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
3378 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
3379 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
3380 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
3381 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
3383 — Berlin, 2014-12-10
3387 * journalctl gained the new options -t/--identifier= to match
3388 on the syslog identifier (aka "tag"), as well as --utc to
3389 show log timestamps in the UTC timezone. journalctl now also
3390 accepts -n/--lines=all to disable line capping in a pager.
3392 * journalctl gained a new switch, --flush, that synchronously
3393 flushes logs from /run/log/journal to /var/log/journal if
3394 persistent storage is enabled. systemd-journal-flush.service
3395 now waits until the operation is complete.
3397 * Services can notify the manager before they start a reload
3398 (by sending RELOADING=1) or shutdown (by sending
3399 STOPPING=1). This allows the manager to track and show the
3400 internal state of daemons and closes a race condition when
3401 the process is still running but has closed its D-Bus
3404 * Services with Type=oneshot do not have to have any ExecStart
3407 * User units are now loaded also from
3408 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
3409 /run/systemd/user directory that was already previously
3410 supported, but is under the control of the user.
3412 * Job timeouts (i.e. time-outs on the time a job that is
3413 queued stays in the run queue) can now optionally result in
3414 immediate reboot or power-off actions (JobTimeoutAction= and
3415 JobTimeoutRebootArgument=). This is useful on ".target"
3416 units, to limit the maximum time a target remains
3417 undispatched in the run queue, and to trigger an emergency
3418 operation in such a case. This is now used by default to
3419 turn off the system if boot-up (as defined by everything in
3420 basic.target) hangs and does not complete for at least
3421 15min. Also, if power-off or reboot hang for at least 30min
3422 an immediate power-off/reboot operation is triggered. This
3423 functionality is particularly useful to increase reliability
3424 on embedded devices, but also on laptops which might
3425 accidentally get powered on when carried in a backpack and
3426 whose boot stays stuck in a hard disk encryption passphrase
3429 * systemd-logind can be configured to also handle lid switch
3430 events even when the machine is docked or multiple displays
3431 are attached (HandleLidSwitchDocked= option).
3433 * A helper binary and a service have been added which can be
3434 used to resume from hibernation in the initramfs. A
3435 generator will parse the resume= option on the kernel
3436 command line to trigger resume.
3438 * A user console daemon systemd-consoled has been
3439 added. Currently, it is a preview, and will so far open a
3440 single terminal on each session of the user marked as
3441 Desktop=systemd-console.
3443 * Route metrics can be specified for DHCP routes added by
3446 * The SELinux context of socket-activated services can be set
3447 from the information provided by the networking stack
3448 (SELinuxContextFromNet= option).
3450 * Userspace firmware loading support has been removed and
3451 the minimum supported kernel version is thus bumped to 3.7.
3453 * Timeout for udev workers has been increased from 1 to 3
3454 minutes, but a warning will be printed after 1 minute to
3455 help diagnose kernel modules that take a long time to load.
3457 * Udev rules can now remove tags on devices with TAG-="foobar".
3459 * systemd's readahead implementation has been removed. In many
3460 circumstances it didn't give expected benefits even for
3461 rotational disk drives and was becoming less relevant in the
3462 age of SSDs. As none of the developers has been using
3463 rotating media anymore, and nobody stepped up to actively
3464 maintain this component of systemd it has now been removed.
3466 * Swap units can use Options= to specify discard options.
3467 Discard options specified for swaps in /etc/fstab are now
3470 * Docker containers are now detected as a separate type of
3473 * The Password Agent protocol gained support for queries where
3474 the user input is shown, useful e.g. for user names.
3475 systemd-ask-password gained a new --echo option to turn that
3478 * The default sysctl.d/ snippets will now set:
3480 net.core.default_qdisc = fq_codel
3482 This selects Fair Queuing Controlled Delay as the default
3483 queuing discipline for network interfaces. fq_codel helps
3484 fight the network bufferbloat problem. It is believed to be
3485 a good default with no tuning required for most workloads.
3486 Downstream distributions may override this choice. On 10Gbit
3487 servers that do not do forwarding, "fq" may perform better.
3488 Systems without a good clocksource should use "pfifo_fast".
3490 * If kdbus is enabled during build a new option BusPolicy= is
3491 available for service units, that allows locking all service
3492 processes into a stricter bus policy, in order to limit
3493 access to various bus services, or even hide most of them
3494 from the service's view entirely.
3496 * networkctl will now show the .network and .link file
3497 networkd has applied to a specific interface.
3499 * sd-login gained a new API call sd_session_get_desktop() to
3500 query which desktop environment has been selected for a
3503 * UNIX utmp support is now compile-time optional to support
3504 legacy-free systems.
3506 * systemctl gained two new commands "add-wants" and
3507 "add-requires" for pulling in units from specific targets
3510 * If the word "rescue" is specified on the kernel command line
3511 the system will now boot into rescue mode (aka
3512 rescue.target), which was previously available only by
3513 specifying "1" or "systemd.unit=rescue.target" on the kernel
3514 command line. This new kernel command line option nicely
3515 mirrors the already existing "emergency" kernel command line
3518 * New kernel command line options mount.usr=, mount.usrflags=,
3519 mount.usrfstype= have been added that match root=, rootflags=,
3520 rootfstype= but allow mounting a specific file system to
3523 * The $NOTIFY_SOCKET is now also passed to control processes of
3524 services, not only the main process.
3526 * This version reenables support for fsck's -l switch. This
3527 means at least version v2.25 of util-linux is required for
3528 operation, otherwise dead-locks on device nodes may
3529 occur. Again: you need to update util-linux to at least
3530 v2.25 when updating systemd to v217.
3532 * The "multi-seat-x" tool has been removed from systemd, as
3533 its functionality has been integrated into X servers 1.16,
3534 and the tool is hence redundant. It is recommended to update
3535 display managers invoking this tool to simply invoke X
3536 directly from now on, again.
3538 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
3539 message flag has been added for all of systemd's PolicyKit
3540 authenticated method calls has been added. In particular
3541 this now allows optional interactive authorization via
3542 PolicyKit for many of PID1's privileged operations such as
3543 unit file enabling and disabling.
3545 * "udevadm hwdb --update" learnt a new switch "--usr" for
3546 placing the rebuilt hardware database in /usr instead of
3547 /etc. When used only hardware database entries stored in
3548 /usr will be used, and any user database entries in /etc are
3549 ignored. This functionality is useful for vendors to ship a
3550 pre-built database on systems where local configuration is
3551 unnecessary or unlikely.
3553 * Calendar time specifications in .timer units now also
3554 understand the strings "semi-annually", "quarterly" and
3555 "minutely" as shortcuts (in addition to the preexisting
3556 "anually", "hourly", ...).
3558 * systemd-tmpfiles will now correctly create files in /dev
3559 at boot which are marked for creation only at boot. It is
3560 recommended to always create static device nodes with 'c!'
3561 and 'b!', so that they are created only at boot and not
3562 overwritten at runtime.
3564 * When the watchdog logic is used for a service (WatchdogSec=)
3565 and the watchdog timeout is hit the service will now be
3566 terminated with SIGABRT (instead of just SIGTERM), in order
3567 to make sure a proper coredump and backtrace is
3568 generated. This ensures that hanging services will result in
3569 similar coredump/backtrace behaviour as services that hit a
3572 Contributions from: Andreas Henriksson, Andrei Borzenkov,
3573 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
3574 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
3575 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
3576 Herrmann, David Sommerseth, David Strauss, Emil Renner
3577 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
3578 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
3579 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
3580 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
3581 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
3582 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
3583 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
3584 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
3585 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
3586 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
3587 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
3588 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
3589 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
3590 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
3591 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
3592 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
3595 — Berlin, 2014-10-28
3599 * timedated no longer reads NTP implementation unit names from
3600 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
3601 implementations should add a
3603 Conflicts=systemd-timesyncd.service
3605 to their unit files to take over and replace systemd's NTP
3606 default functionality.
3608 * systemd-sysusers gained a new line type "r" for configuring
3609 which UID/GID ranges to allocate system users/groups
3610 from. Lines of type "u" may now add an additional column
3611 that specifies the home directory for the system user to be
3612 created. Also, systemd-sysusers may now optionally read user
3613 information from STDIN instead of a file. This is useful for
3614 invoking it from RPM preinst scriptlets that need to create
3615 users before the first RPM file is installed since these
3616 files might need to be owned by them. A new
3617 %sysusers_create_inline RPM macro has been introduced to do
3618 just that. systemd-sysusers now updates the shadow files as
3619 well as the user/group databases, which should enhance
3620 compatibility with certain tools like grpck.
3622 * A number of bus APIs of PID 1 now optionally consult
3623 PolicyKit to permit access for otherwise unprivileged
3624 clients under certain conditions. Note that this currently
3625 doesn't support interactive authentication yet, but this is
3626 expected to be added eventually, too.
3628 * /etc/machine-info now has new fields for configuring the
3629 deployment environment of the machine, as well as the
3630 location of the machine. hostnamectl has been updated with
3631 new command to update these fields.
3633 * systemd-timesyncd has been updated to automatically acquire
3634 NTP server information from systemd-networkd, which might
3635 have been discovered via DHCP.
3637 * systemd-resolved now includes a caching DNS stub resolver
3638 and a complete LLMNR name resolution implementation. A new
3639 NSS module "nss-resolve" has been added which can be used
3640 instead of glibc's own "nss-dns" to resolve hostnames via
3641 systemd-resolved. Hostnames, addresses and arbitrary RRs may
3642 be resolved via systemd-resolved D-Bus APIs. In contrast to
3643 the glibc internal resolver systemd-resolved is aware of
3644 multi-homed system, and keeps DNS server and caches separate
3645 and per-interface. Queries are sent simultaneously on all
3646 interfaces that have DNS servers configured, in order to
3647 properly handle VPNs and local LANs which might resolve
3648 separate sets of domain names. systemd-resolved may acquire
3649 DNS server information from systemd-networkd automatically,
3650 which in turn might have discovered them via DHCP. A tool
3651 "systemd-resolve-host" has been added that may be used to
3652 query the DNS logic in resolved. systemd-resolved implements
3653 IDNA and automatically uses IDNA or UTF-8 encoding depending
3654 on whether classic DNS or LLMNR is used as transport. In the
3655 next releases we intend to add a DNSSEC and mDNS/DNS-SD
3656 implementation to systemd-resolved.
3658 * A new NSS module nss-mymachines has been added, that
3659 automatically resolves the names of all local registered
3660 containers to their respective IP addresses.
3662 * A new client tool "networkctl" for systemd-networkd has been
3663 added. It currently is entirely passive and will query
3664 networking configuration from udev, rtnetlink and networkd,
3665 and present it to the user in a very friendly
3666 way. Eventually, we hope to extend it to become a full
3667 control utility for networkd.
3669 * .socket units gained a new DeferAcceptSec= setting that
3670 controls the kernels' TCP_DEFER_ACCEPT sockopt for
3671 TCP. Similarly, support for controlling TCP keep-alive
3672 settings has been added (KeepAliveTimeSec=,
3673 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
3674 turning off Nagle's algorithm on TCP has been added
3677 * logind learned a new session type "web", for use in projects
3678 like Cockpit which register web clients as PAM sessions.
3680 * timer units with at least one OnCalendar= setting will now
3681 be started only after time-sync.target has been
3682 reached. This way they will not elapse before the system
3683 clock has been corrected by a local NTP client or
3684 similar. This is particular useful on RTC-less embedded
3685 machines, that come up with an invalid system clock.
3687 * systemd-nspawn's --network-veth= switch should now result in
3688 stable MAC addresses for both the outer and the inner side
3691 * systemd-nspawn gained a new --volatile= switch for running
3692 container instances with /etc or /var unpopulated.
3694 * The kdbus client code has been updated to use the new Linux
3695 3.17 memfd subsystem instead of the old kdbus-specific one.
3697 * systemd-networkd's DHCP client and server now support
3698 FORCERENEW. There are also new configuration options to
3699 configure the vendor client identifier and broadcast mode
3702 * systemd will no longer inform the kernel about the current
3703 timezone, as this is necessarily incorrect and racy as the
3704 kernel has no understanding of DST and similar
3705 concepts. This hence means FAT timestamps will be always
3706 considered UTC, similar to what Android is already
3707 doing. Also, when the RTC is configured to the local time
3708 (rather than UTC) systemd will never synchronize back to it,
3709 as this might confuse Windows at a later boot.
3711 * systemd-analyze gained a new command "verify" for offline
3712 validation of unit files.
3714 * systemd-networkd gained support for a couple of additional
3715 settings for bonding networking setups. Also, the metric for
3716 statically configured routes may now be configured. For
3717 network interfaces where this is appropriate the peer IP
3718 address may now be configured.
3720 * systemd-networkd's DHCP client will no longer request
3721 broadcasting by default, as this tripped up some networks.
3722 For hardware where broadcast is required the feature should
3723 be switched back on using RequestBroadcast=yes.
3725 * systemd-networkd will now set up IPv4LL addresses (when
3726 enabled) even if DHCP is configured successfully.
3728 * udev will now default to respect network device names given
3729 by the kernel when the kernel indicates that these are
3730 predictable. This behavior can be tweaked by changing
3731 NamePolicy= in the relevant .link file.
3733 * A new library systemd-terminal has been added that
3734 implements full TTY stream parsing and rendering. This
3735 library is supposed to be used later on for implementing a
3736 full userspace VT subsystem, replacing the current kernel
3739 * A new tool systemd-journal-upload has been added to push
3740 journal data to a remote system running
3741 systemd-journal-remote.
3743 * journald will no longer forward all local data to another
3744 running syslog daemon. This change has been made because
3745 rsyslog (which appears to be the most commonly used syslog
3746 implementation these days) no longer makes use of this, and
3747 instead pulls the data out of the journal on its own. Since
3748 forwarding the messages to a non-existent syslog server is
3749 more expensive than we assumed we have now turned this
3750 off. If you run a syslog server that is not a recent rsyslog
3751 version, you have to turn this option on again
3752 (ForwardToSyslog= in journald.conf).
3754 * journald now optionally supports the LZ4 compressor for
3755 larger journal fields. This compressor should perform much
3756 better than XZ which was the previous default.
3758 * machinectl now shows the IP addresses of local containers,
3759 if it knows them, plus the interface name of the container.
3761 * A new tool "systemd-escape" has been added that makes it
3762 easy to escape strings to build unit names and similar.
3764 * sd_notify() messages may now include a new ERRNO= field
3765 which is parsed and collected by systemd and shown among the
3766 "systemctl status" output for a service.
3768 * A new component "systemd-firstboot" has been added that
3769 queries the most basic systemd information (timezone,
3770 hostname, root password) interactively on first
3771 boot. Alternatively it may also be used to provision these
3772 things offline on OS images installed into directories.
3774 * The default sysctl.d/ snippets will now set
3776 net.ipv4.conf.default.promote_secondaries=1
3778 This has the benefit of no flushing secondary IP addresses
3779 when primary addresses are removed.
3781 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
3782 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
3783 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
3784 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
3785 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
3786 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
3787 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
3788 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
3789 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
3790 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
3791 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
3792 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
3793 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
3794 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
3795 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
3797 — Berlin, 2014-08-19
3801 * A new tool systemd-sysusers has been added. This tool
3802 creates system users and groups in /etc/passwd and
3803 /etc/group, based on static declarative system user/group
3804 definitions in /usr/lib/sysusers.d/. This is useful to
3805 enable factory resets and volatile systems that boot up with
3806 an empty /etc directory, and thus need system users and
3807 groups created during early boot. systemd now also ships
3808 with two default sysusers.d/ files for the most basic
3809 users and groups systemd and the core operating system
3812 * A new tmpfiles snippet has been added that rebuilds the
3813 essential files in /etc on boot, should they be missing.
3815 * A directive for ensuring automatic clean-up of
3816 /var/cache/man/ has been removed from the default
3817 configuration. This line should now be shipped by the man
3818 implementation. The necessary change has been made to the
3819 man-db implementation. Note that you need to update your man
3820 implementation to one that ships this line, otherwise no
3821 automatic clean-up of /var/cache/man will take place.
3823 * A new condition ConditionNeedsUpdate= has been added that
3824 may conditionalize services to only run when /etc or /var
3825 are "older" than the vendor operating system resources in
3826 /usr. This is useful for reconstructing or updating /etc
3827 after an offline update of /usr or a factory reset, on the
3828 next reboot. Services that want to run once after such an
3829 update or reset should use this condition and order
3830 themselves before the new systemd-update-done.service, which
3831 will mark the two directories as fully updated. A number of
3832 service files have been added making use of this, to rebuild
3833 the udev hardware database, the journald message catalog and
3834 dynamic loader cache (ldconfig). The systemd-sysusers tool
3835 described above also makes use of this now. With this in
3836 place it is now possible to start up a minimal operating
3837 system with /etc empty cleanly. For more information on the
3838 concepts involved see this recent blog story:
3840 http://0pointer.de/blog/projects/stateless.html
3842 * A new system group "input" has been introduced, and all
3843 input device nodes get this group assigned. This is useful
3844 for system-level software to get access to input devices. It
3845 complements what is already done for "audio" and "video".
3847 * systemd-networkd learnt minimal DHCPv4 server support in
3848 addition to the existing DHCPv4 client support. It also
3849 learnt DHCPv6 client and IPv6 Router Solicitation client
3850 support. The DHCPv4 client gained support for static routes
3851 passed in from the server. Note that the [DHCPv4] section
3852 known in older systemd-networkd versions has been renamed to
3853 [DHCP] and is now also used by the DHCPv6 client. Existing
3854 .network files using settings of this section should be
3855 updated, though compatibility is maintained. Optionally, the
3856 client hostname may now be sent to the DHCP server.
3858 * networkd gained support for vxlan virtual networks as well
3859 as tun/tap and dummy devices.
3861 * networkd gained support for automatic allocation of address
3862 ranges for interfaces from a system-wide pool of
3863 addresses. This is useful for dynamically managing a large
3864 number of interfaces with a single network configuration
3865 file. In particular this is useful to easily assign
3866 appropriate IP addresses to the veth links of a large number
3867 of nspawn instances.
3869 * RPM macros for processing sysusers, sysctl and binfmt
3870 drop-in snippets at package installation time have been
3873 * The /etc/os-release file should now be placed in
3874 /usr/lib/os-release. The old location is automatically
3875 created as symlink. /usr/lib is the more appropriate
3876 location of this file, since it shall actually describe the
3877 vendor operating system shipped in /usr, and not the
3878 configuration stored in /etc.
3880 * .mount units gained a new boolean SloppyOptions= setting
3881 that maps to mount(8)'s -s option which enables permissive
3882 parsing of unknown mount options.
3884 * tmpfiles learnt a new "L+" directive which creates a symlink
3885 but (unlike "L") deletes a pre-existing file first, should
3886 it already exist and not already be the correct
3887 symlink. Similarly, "b+", "c+" and "p+" directives have been
3888 added as well, which create block and character devices, as
3889 well as fifos in the filesystem, possibly removing any
3890 pre-existing files of different types.
3892 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
3893 'argument' field (which so far specified the source to
3894 symlink/copy the files from) is now optional. If omitted the
3895 same file os copied from /usr/share/factory/ suffixed by the
3896 full destination path. This is useful for populating /etc
3897 with essential files, by copying them from vendor defaults
3898 shipped in /usr/share/factory/etc.
3900 * A new command "systemctl preset-all" has been added that
3901 applies the service preset settings to all installed unit
3902 files. A new switch --preset-mode= has been added that
3903 controls whether only enable or only disable operations
3906 * A new command "systemctl is-system-running" has been added
3907 that allows checking the overall state of the system, for
3908 example whether it is fully up and running.
3910 * When the system boots up with an empty /etc, the equivalent
3911 to "systemctl preset-all" is executed during early boot, to
3912 make sure all default services are enabled after a factory
3915 * systemd now contains a minimal preset file that enables the
3916 most basic services systemd ships by default.
3918 * Unit files' [Install] section gained a new DefaultInstance=
3919 field for defining the default instance to create if a
3920 template unit is enabled with no instance specified.
3922 * A new passive target cryptsetup-pre.target has been added
3923 that may be used by services that need to make they run and
3924 finish before the first LUKS cryptographic device is set up.
3926 * The /dev/loop-control and /dev/btrfs-control device nodes
3927 are now owned by the "disk" group by default, opening up
3928 access to this group.
3930 * systemd-coredump will now automatically generate a
3931 stack trace of all core dumps taking place on the system,
3932 based on elfutils' libdw library. This stack trace is logged
3935 * systemd-coredump may now optionally store coredumps directly
3936 on disk (in /var/lib/systemd/coredump, possibly compressed),
3937 instead of storing them unconditionally in the journal. This
3938 mode is the new default. A new configuration file
3939 /etc/systemd/coredump.conf has been added to configure this
3940 and other parameters of systemd-coredump.
3942 * coredumpctl gained a new "info" verb to show details about a
3943 specific coredump. A new switch "-1" has also been added
3944 that makes sure to only show information about the most
3945 recent entry instead of all entries. Also, as the tool is
3946 generally useful now the "systemd-" prefix of the binary
3947 name has been removed. Distributions that want to maintain
3948 compatibility with the old name should add a symlink from
3949 the old name to the new name.
3951 * journald's SplitMode= now defaults to "uid". This makes sure
3952 that unprivileged users can access their own coredumps with
3953 coredumpctl without restrictions.
3955 * New kernel command line options "systemd.wants=" (for
3956 pulling an additional unit during boot), "systemd.mask="
3957 (for masking a specific unit for the boot), and
3958 "systemd.debug-shell" (for enabling the debug shell on tty9)
3959 have been added. This is implemented in the new generator
3960 "systemd-debug-generator".
3962 * systemd-nspawn will now by default filter a couple of
3963 syscalls for containers, among them those required for
3964 kernel module loading, direct x86 IO port access, swap
3965 management, and kexec. Most importantly though
3966 open_by_handle_at() is now prohibited for containers,
3967 closing a hole similar to a recently discussed vulnerability
3968 in docker regarding access to files on file hierarchies the
3969 container should normally not have access to. Note that, for
3970 nspawn, we generally make no security claims anyway (and
3971 this is explicitly documented in the man page), so this is
3972 just a fix for one of the most obvious problems.
3974 * A new man page file-hierarchy(7) has been added that
3975 contains a minimized, modernized version of the file system
3976 layout systemd expects, similar in style to the FHS
3977 specification or hier(5). A new tool systemd-path(1) has
3978 been added to query many of these paths for the local
3981 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
3982 longer done. Since the directory now has a per-user size
3983 limit, and is cleaned on logout this appears unnecessary,
3984 in particular since this now brings the lifecycle of this
3985 directory closer in line with how IPC objects are handled.
3987 * systemd.pc now exports a number of additional directories,
3988 including $libdir (which is useful to identify the library
3989 path for the primary architecture of the system), and a
3990 couple of drop-in directories.
3992 * udev's predictable network interface names now use the dev_port
3993 sysfs attribute, introduced in linux 3.15 instead of dev_id to
3994 distinguish between ports of the same PCI function. dev_id should
3995 only be used for ports using the same HW address, hence the need
3998 * machined has been updated to export the OS version of a
3999 container (read from /etc/os-release and
4000 /usr/lib/os-release) on the bus. This is now shown in
4001 "machinectl status" for a machine.
4003 * A new service setting RestartForceExitStatus= has been
4004 added. If configured to a set of exit signals or process
4005 return values, the service will be restarted when the main
4006 daemon process exits with any of them, regardless of the
4009 * systemctl's -H switch for connecting to remote systemd
4010 machines has been extended so that it may be used to
4011 directly connect to a specific container on the
4012 host. "systemctl -H root@foobar:waldi" will now connect as
4013 user "root" to host "foobar", and then proceed directly to
4014 the container named "waldi". Note that currently you have to
4015 authenticate as user "root" for this to work, as entering
4016 containers is a privileged operation.
4018 Contributions from: Andreas Henriksson, Benjamin Steinwender,
4019 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
4020 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
4021 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
4022 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
4023 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
4024 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
4025 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
4026 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
4027 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
4028 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
4029 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
4031 — Berlin, 2014-07-03
4035 * As an experimental feature, udev now tries to lock the
4036 disk device node (flock(LOCK_SH|LOCK_NB)) while it
4037 executes events for the disk or any of its partitions.
4038 Applications like partitioning programs can lock the
4039 disk device node (flock(LOCK_EX)) and claim temporary
4040 device ownership that way; udev will entirely skip all event
4041 handling for this disk and its partitions. If the disk
4042 was opened for writing, the close will trigger a partition
4043 table rescan in udev's "watch" facility, and if needed
4044 synthesize "change" events for the disk and all its partitions.
4045 This is now unconditionally enabled, and if it turns out to
4046 cause major problems, we might turn it on only for specific
4047 devices, or might need to disable it entirely. Device Mapper
4048 devices are excluded from this logic.
4050 * We temporarily dropped the "-l" switch for fsck invocations,
4051 since they collide with the flock() logic above. util-linux
4052 upstream has been changed already to avoid this conflict,
4053 and we will readd "-l" as soon as util-linux with this
4054 change has been released.
4056 * The dependency on libattr has been removed. Since a long
4057 time, the extended attribute calls have moved to glibc, and
4058 libattr is thus unnecessary.
4060 * Virtualization detection works without privileges now. This
4061 means the systemd-detect-virt binary no longer requires
4062 CAP_SYS_PTRACE file capabilities, and our daemons can run
4063 with fewer privileges.
4065 * systemd-networkd now runs under its own "systemd-network"
4066 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
4067 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
4068 loses the ability to write to files owned by root this way.
4070 * Similarly, systemd-resolved now runs under its own
4071 "systemd-resolve" user with no capabilities remaining.
4073 * Similarly, systemd-bus-proxyd now runs under its own
4074 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
4076 * systemd-networkd gained support for setting up "veth"
4077 virtual Ethernet devices for container connectivity, as well
4078 as GRE and VTI tunnels.
4080 * systemd-networkd will no longer automatically attempt to
4081 manually load kernel modules necessary for certain tunnel
4082 transports. Instead, it is assumed the kernel loads them
4083 automatically when required. This only works correctly on
4084 very new kernels. On older kernels, please consider adding
4085 the kernel modules to /etc/modules-load.d/ as a work-around.
4087 * The resolv.conf file systemd-resolved generates has been
4088 moved to /run/systemd/resolve/. If you have a symlink from
4089 /etc/resolv.conf, it might be necessary to correct it.
4091 * Two new service settings, ProtectHome= and ProtectSystem=,
4092 have been added. When enabled, they will make the user data
4093 (such as /home) inaccessible or read-only and the system
4094 (such as /usr) read-only, for specific services. This allows
4095 very light-weight per-service sandboxing to avoid
4096 modifications of user data or system files from
4097 services. These two new switches have been enabled for all
4098 of systemd's long-running services, where appropriate.
4100 * Socket units gained new SocketUser= and SocketGroup=
4101 settings to set the owner user and group of AF_UNIX sockets
4102 and FIFOs in the file system.
4104 * Socket units gained a new RemoveOnStop= setting. If enabled,
4105 all FIFOS and sockets in the file system will be removed
4106 when the specific socket unit is stopped.
4108 * Socket units gained a new Symlinks= setting. It takes a list
4109 of symlinks to create to file system sockets or FIFOs
4110 created by the specific Unix sockets. This is useful to
4111 manage symlinks to socket nodes with the same life-cycle as
4114 * The /dev/log socket and /dev/initctl FIFO have been moved to
4115 /run, and have been replaced by symlinks. This allows
4116 connecting to these facilities even if PrivateDevices=yes is
4117 used for a service (which makes /dev/log itself unavailable,
4118 but /run is left). This also has the benefit of ensuring
4119 that /dev only contains device nodes, directories and
4120 symlinks, and nothing else.
4122 * sd-daemon gained two new calls sd_pid_notify() and
4123 sd_pid_notifyf(). They are similar to sd_notify() and
4124 sd_notifyf(), but allow overriding of the source PID of
4125 notification messages if permissions permit this. This is
4126 useful to send notify messages on behalf of a different
4127 process (for example, the parent process). The
4128 systemd-notify tool has been updated to make use of this
4129 when sending messages (so that notification messages now
4130 originate from the shell script invoking systemd-notify and
4131 not the systemd-notify process itself. This should minimize
4132 a race where systemd fails to associate notification
4133 messages to services when the originating process already
4136 * A new "on-abnormal" setting for Restart= has been added. If
4137 set, it will result in automatic restarts on all "abnormal"
4138 reasons for a process to exit, which includes unclean
4139 signals, core dumps, timeouts and watchdog timeouts, but
4140 does not include clean and unclean exit codes or clean
4141 signals. Restart=on-abnormal is an alternative for
4142 Restart=on-failure for services that shall be able to
4143 terminate and avoid restarts on certain errors, by
4144 indicating so with an unclean exit code. Restart=on-failure
4145 or Restart=on-abnormal is now the recommended setting for
4146 all long-running services.
4148 * If the InaccessibleDirectories= service setting points to a
4149 mount point (or if there are any submounts contained within
4150 it), it is now attempted to completely unmount it, to make
4151 the file systems truly unavailable for the respective
4154 * The ReadOnlyDirectories= service setting and
4155 systemd-nspawn's --read-only parameter are now recursively
4156 applied to all submounts, too.
4158 * Mount units may now be created transiently via the bus APIs.
4160 * The support for SysV and LSB init scripts has been removed
4161 from the systemd daemon itself. Instead, it is now
4162 implemented as a generator that creates native systemd units
4163 from these scripts when needed. This enables us to remove a
4164 substantial amount of legacy code from PID 1, following the
4165 fact that many distributions only ship a very small number
4166 of LSB/SysV init scripts nowadays.
4168 * Privileged Xen (dom0) domains are not considered
4169 virtualization anymore by the virtualization detection
4170 logic. After all, they generally have unrestricted access to
4171 the hardware and usually are used to manage the unprivileged
4174 * systemd-tmpfiles gained a new "C" line type, for copying
4175 files or entire directories.
4177 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
4178 lines. So far, they have been non-globbing versions of the
4179 latter, and have thus been redundant. In future, it is
4180 recommended to only use "z". "m" has hence been removed
4181 from the documentation, even though it stays supported.
4183 * A tmpfiles snippet to recreate the most basic structure in
4184 /var has been added. This is enough to create the /var/run →
4185 /run symlink and create a couple of structural
4186 directories. This allows systems to boot up with an empty or
4187 volatile /var. Of course, while with this change, the core OS
4188 now is capable with dealing with a volatile /var, not all
4189 user services are ready for it. However, we hope that sooner
4190 or later, many service daemons will be changed upstream so
4191 that they are able to automatically create their necessary
4192 directories in /var at boot, should they be missing. This is
4193 the first step to allow state-less systems that only require
4194 the vendor image for /usr to boot.
4196 * systemd-nspawn has gained a new --tmpfs= switch to mount an
4197 empty tmpfs instance to a specific directory. This is
4198 particularly useful for making use of the automatic
4199 reconstruction of /var (see above), by passing --tmpfs=/var.
4201 * Access modes specified in tmpfiles snippets may now be
4202 prefixed with "~", which indicates that they shall be masked
4203 by whether the existing file or directory is currently
4204 writable, readable or executable at all. Also, if specified,
4205 the sgid/suid/sticky bits will be masked for all
4208 * A new passive target unit "network-pre.target" has been
4209 added which is useful for services that shall run before any
4210 network is configured, for example firewall scripts.
4212 * The "floppy" group that previously owned the /dev/fd*
4213 devices is no longer used. The "disk" group is now used
4214 instead. Distributions should probably deprecate usage of
4217 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
4218 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
4219 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
4220 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
4221 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
4222 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
4225 — Berlin, 2014-06-11
4229 * A new "systemd-timesyncd" daemon has been added for
4230 synchronizing the system clock across the network. It
4231 implements an SNTP client. In contrast to NTP
4232 implementations such as chrony or the NTP reference server,
4233 this only implements a client side, and does not bother with
4234 the full NTP complexity, focusing only on querying time from
4235 one remote server and synchronizing the local clock to
4236 it. Unless you intend to serve NTP to networked clients or
4237 want to connect to local hardware clocks, this simple NTP
4238 client should be more than appropriate for most
4239 installations. The daemon runs with minimal privileges, and
4240 has been hooked up with networkd to only operate when
4241 network connectivity is available. The daemon saves the
4242 current clock to disk every time a new NTP sync has been
4243 acquired, and uses this to possibly correct the system clock
4244 early at bootup, in order to accommodate for systems that
4245 lack an RTC such as the Raspberry Pi and embedded devices,
4246 and to make sure that time monotonically progresses on these
4247 systems, even if it is not always correct. To make use of
4248 this daemon, a new system user and group "systemd-timesync"
4249 needs to be created on installation of systemd.
4251 * The queue "seqnum" interface of libudev has been disabled, as
4252 it was generally incompatible with device namespacing as
4253 sequence numbers of devices go "missing" if the devices are
4254 part of a different namespace.
4256 * "systemctl list-timers" and "systemctl list-sockets" gained
4257 a --recursive switch for showing units of these types also
4258 for all local containers, similar in style to the already
4259 supported --recursive switch for "systemctl list-units".
4261 * A new RebootArgument= setting has been added for service
4262 units, which may be used to specify a kernel reboot argument
4263 to use when triggering reboots with StartLimitAction=.
4265 * A new FailureAction= setting has been added for service
4266 units which may be used to specify an operation to trigger
4267 when a service fails. This works similarly to
4268 StartLimitAction=, but unlike it, controls what is done
4269 immediately rather than only after several attempts to
4270 restart the service in question.
4272 * hostnamed got updated to also expose the kernel name,
4273 release, and version on the bus. This is useful for
4274 executing commands like hostnamectl with the -H switch.
4275 systemd-analyze makes use of this to properly display
4276 details when running non-locally.
4278 * The bootchart tool can now show cgroup information in the
4279 graphs it generates.
4281 * The CFS CPU quota cgroup attribute is now exposed for
4282 services. The new CPUQuota= switch has been added for this
4283 which takes a percentage value. Setting this will have the
4284 result that a service may never get more CPU time than the
4285 specified percentage, even if the machine is otherwise idle.
4287 * systemd-networkd learned IPIP and SIT tunnel support.
4289 * LSB init scripts exposing a dependency on $network will now
4290 get a dependency on network-online.target rather than simply
4291 network.target. This should bring LSB handling closer to
4292 what it was on SysV systems.
4294 * A new fsck.repair= kernel option has been added to control
4295 how fsck shall deal with unclean file systems at boot.
4297 * The (.ini) configuration file parser will now silently
4298 ignore sections whose name begins with "X-". This may be
4299 used to maintain application-specific extension sections in unit
4302 * machined gained a new API to query the IP addresses of
4303 registered containers. "machinectl status" has been updated
4304 to show these addresses in its output.
4306 * A new call sd_uid_get_display() has been added to the
4307 sd-login APIs for querying the "primary" session of a
4308 user. The "primary" session of the user is elected from the
4309 user's sessions and generally a graphical session is
4310 preferred over a text one.
4312 * A minimal systemd-resolved daemon has been added. It
4313 currently simply acts as a companion to systemd-networkd and
4314 manages resolv.conf based on per-interface DNS
4315 configuration, possibly supplied via DHCP. In the long run
4316 we hope to extend this into a local DNSSEC enabled DNS and
4319 * The systemd-networkd-wait-online tool is now enabled by
4320 default. It will delay network-online.target until a network
4321 connection has been configured. The tool primarily integrates
4322 with networkd, but will also make a best effort to make sense
4323 of network configuration performed in some other way.
4325 * Two new service options StartupCPUShares= and
4326 StartupBlockIOWeight= have been added that work similarly to
4327 CPUShares= and BlockIOWeight= however only apply during
4328 system startup. This is useful to prioritize certain services
4329 differently during bootup than during normal runtime.
4331 * hostnamed has been changed to prefer the statically
4332 configured hostname in /etc/hostname (unless set to
4333 'localhost' or empty) over any dynamic one supplied by
4334 dhcp. With this change, the rules for picking the hostname
4335 match more closely the rules of other configuration settings
4336 where the local administrator's configuration in /etc always
4337 overrides any other settings.
4339 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
4340 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
4341 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
4342 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
4343 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
4344 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
4345 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
4346 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
4347 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
4348 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
4349 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
4350 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
4351 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
4352 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
4353 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
4354 Lindskog, WaLyong Cho, Will Woods, Zbigniew
4357 — Beijing, 2014-05-28
4361 * When restoring the screen brightness at boot, stay away from
4362 the darkest setting or from the lowest 5% of the available
4363 range, depending on which is the larger value of both. This
4364 should effectively protect the user from rebooting into a
4365 black screen, should the brightness have been set to minimum
4368 * sd-login gained a new sd_machine_get_class() call to
4369 determine the class ("vm" or "container") of a machine
4370 registered with machined.
4372 * sd-login gained new calls
4373 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
4374 to query the identity of the peer of a local AF_UNIX
4375 connection. They operate similarly to their sd_pid_get_xyz()
4378 * PID 1 will now maintain a system-wide system state engine
4379 with the states "starting", "running", "degraded",
4380 "maintenance", "stopping". These states are bound to system
4381 startup, normal runtime, runtime with at least one failed
4382 service, rescue/emergency mode and system shutdown. This
4383 state is shown in the "systemctl status" output when no unit
4384 name is passed. It is useful to determine system state, in
4385 particularly when doing so for many systems or containers at
4388 * A new command "list-machines" has been added to "systemctl"
4389 that lists all local OS containers and shows their system
4390 state (see above), if systemd runs inside of them.
4392 * systemctl gained a new "-r" switch to recursively enumerate
4393 units on all local containers, when used with the
4394 "list-unit" command (which is the default one that is
4395 executed when no parameters are specified).
4397 * The GPT automatic partition discovery logic will now honour
4398 two GPT partition flags: one may be set on a partition to
4399 cause it to be mounted read-only, and the other may be set
4400 on a partition to ignore it during automatic discovery.
4402 * Two new GPT type UUIDs have been added for automatic root
4403 partition discovery, for 32-bit and 64-bit ARM. This is not
4404 particularly useful for discovering the root directory on
4405 these architectures during bare-metal boots (since UEFI is
4406 not common there), but still very useful to allow booting of
4407 ARM disk images in nspawn with the -i option.
4409 * MAC addresses of interfaces created with nspawn's
4410 --network-interface= switch will now be generated from the
4411 machine name, and thus be stable between multiple invocations
4414 * logind will now automatically remove all IPC objects owned
4415 by a user if she or he fully logs out. This makes sure that
4416 users who are logged out cannot continue to consume IPC
4417 resources. This covers SysV memory, semaphores and message
4418 queues as well as POSIX shared memory and message
4419 queues. Traditionally, SysV and POSIX IPC had no life-cycle
4420 limits. With this functionality, that is corrected. This may
4421 be turned off by using the RemoveIPC= switch of logind.conf.
4423 * The systemd-machine-id-setup and tmpfiles tools gained a
4424 --root= switch to operate on a specific root directory,
4427 * journald can now forward logged messages to the TTYs of all
4428 logged in users ("wall"). This is the default for all
4429 emergency messages now.
4431 * A new tool systemd-journal-remote has been added to stream
4432 journal log messages across the network.
4434 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
4435 controller trees are mounted into it. Note that the
4436 directories mounted beneath it are not read-only. This is a
4437 security measure and is particularly useful because glibc
4438 actually includes a search logic to pick any tmpfs it can
4439 find to implement shm_open() if /dev/shm is not available
4440 (which it might very well be in namespaced setups).
4442 * machinectl gained a new "poweroff" command to cleanly power
4443 down a local OS container.
4445 * The PrivateDevices= unit file setting will now also drop the
4446 CAP_MKNOD capability from the capability bound set, and
4447 imply DevicePolicy=closed.
4449 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
4450 comprehensively on all long-running systemd services where
4451 this is appropriate.
4453 * systemd-udevd will now run in a disassociated mount
4454 namespace. To mount directories from udev rules, make sure to
4455 pull in mount units via SYSTEMD_WANTS properties.
4457 * The kdbus support gained support for uploading policy into
4458 the kernel. sd-bus gained support for creating "monitoring"
4459 connections that can eavesdrop into all bus communication
4460 for debugging purposes.
4462 * Timestamps may now be specified in seconds since the UNIX
4463 epoch Jan 1st, 1970 by specifying "@" followed by the value
4466 * Native tcpwrap support in systemd has been removed. tcpwrap
4467 is old code, not really maintained anymore and has serious
4468 shortcomings, and better options such as firewalls
4469 exist. For setups that require tcpwrap usage, please
4470 consider invoking your socket-activated service via tcpd,
4471 like on traditional inetd.
4473 * A new system.conf configuration option
4474 DefaultTimerAccuracySec= has been added that controls the
4475 default AccuracySec= setting of .timer units.
4477 * Timer units gained a new WakeSystem= switch. If enabled,
4478 timers configured this way will cause the system to resume
4479 from system suspend (if the system supports that, which most
4482 * Timer units gained a new Persistent= switch. If enabled,
4483 timers configured this way will save to disk when they have
4484 been last triggered. This information is then used on next
4485 reboot to possible execute overdue timer events, that
4486 could not take place because the system was powered off.
4487 This enables simple anacron-like behaviour for timer units.
4489 * systemctl's "list-timers" will now also list the time a
4490 timer unit was last triggered in addition to the next time
4491 it will be triggered.
4493 * systemd-networkd will now assign predictable IPv4LL
4494 addresses to its local interfaces.
4496 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
4497 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
4498 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
4499 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
4500 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
4501 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
4502 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
4503 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
4506 — Berlin, 2014-03-25
4510 * A new unit file setting RestrictAddressFamilies= has been
4511 added to restrict which socket address families unit
4512 processes gain access to. This takes address family names
4513 like "AF_INET" or "AF_UNIX", and is useful to minimize the
4514 attack surface of services via exotic protocol stacks. This
4515 is built on seccomp system call filters.
4517 * Two new unit file settings RuntimeDirectory= and
4518 RuntimeDirectoryMode= have been added that may be used to
4519 manage a per-daemon runtime directories below /run. This is
4520 an alternative for setting up directory permissions with
4521 tmpfiles snippets, and has the advantage that the runtime
4522 directory's lifetime is bound to the daemon runtime and that
4523 the daemon starts up with an empty directory each time. This
4524 is particularly useful when writing services that drop
4525 privileges using the User= or Group= setting.
4527 * The DeviceAllow= unit setting now supports globbing for
4528 matching against device group names.
4530 * The systemd configuration file system.conf gained new
4531 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
4532 DefaultMemoryAccounting= to globally turn on/off accounting
4533 for specific resources (cgroups) for all units. These
4534 settings may still be overridden individually in each unit
4537 * systemd-gpt-auto-generator is now able to discover /srv and
4538 root partitions in addition to /home and swap partitions. It
4539 also supports LUKS-encrypted partitions now. With this in
4540 place, automatic discovery of partitions to mount following
4541 the Discoverable Partitions Specification
4542 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
4543 is now a lot more complete. This allows booting without
4544 /etc/fstab and without root= on the kernel command line on
4545 systems prepared appropriately.
4547 * systemd-nspawn gained a new --image= switch which allows
4548 booting up disk images and Linux installations on any block
4549 device that follow the Discoverable Partitions Specification
4550 (see above). This means that installations made with
4551 appropriately updated installers may now be started and
4552 deployed using container managers, completely
4553 unmodified. (We hope that libvirt-lxc will add support for
4554 this feature soon, too.)
4556 * systemd-nspawn gained a new --network-macvlan= setting to
4557 set up a private macvlan interface for the
4558 container. Similarly, systemd-networkd gained a new
4559 Kind=macvlan setting in .netdev files.
4561 * systemd-networkd now supports configuring local addresses
4564 * A new tool systemd-network-wait-online has been added to
4565 synchronously wait for network connectivity using
4568 * The sd-bus.h bus API gained a new sd_bus_track object for
4569 tracking the life-cycle of bus peers. Note that sd-bus.h is
4570 still not a public API though (unless you specify
4571 --enable-kdbus on the configure command line, which however
4572 voids your warranty and you get no API stability guarantee).
4574 * The $XDG_RUNTIME_DIR runtime directories for each user are
4575 now individual tmpfs instances, which has the benefit of
4576 introducing separate pools for each user, with individual
4577 size limits, and thus making sure that unprivileged clients
4578 can no longer negatively impact the system or other users by
4579 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
4580 RuntimeDirectorySize= has been introduced that allows
4581 controlling the default size limit for all users. It
4582 defaults to 10% of the available physical memory. This is no
4583 replacement for quotas on tmpfs though (which the kernel
4584 still does not support), as /dev/shm and /tmp are still
4585 shared resources used by both the system and unprivileged
4588 * logind will now automatically turn off automatic suspending
4589 on laptop lid close when more than one display is
4590 connected. This was previously expected to be implemented
4591 individually in desktop environments (such as GNOME),
4592 however has been added to logind now, in order to fix a
4593 boot-time race where a desktop environment might not have
4594 been started yet and thus not been able to take an inhibitor
4595 lock at the time where logind already suspends the system
4596 due to a closed lid.
4598 * logind will now wait at least 30s after each system
4599 suspend/resume cycle, and 3min after system boot before
4600 suspending the system due to a closed laptop lid. This
4601 should give USB docking stations and similar enough time to
4602 be probed and configured after system resume and boot in
4603 order to then act as suspend blocker.
4605 * systemd-run gained a new --property= setting which allows
4606 initialization of resource control properties (and others)
4607 for the created scope or service unit. Example: "systemd-run
4608 --property=BlockIOWeight=10 updatedb" may be used to run
4609 updatedb at a low block IO scheduling weight.
4611 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
4612 now also work in --scope mode.
4614 * When systemd is compiled with kdbus support, basic support
4615 for enforced policies is now in place. (Note that enabling
4616 kdbus still voids your warranty and no API compatibility
4619 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
4620 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
4621 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
4622 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
4623 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
4624 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
4625 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
4626 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
4627 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
4628 Zbigniew Jędrzejewski-Szmek
4630 — Berlin, 2014-03-12
4634 * systemd will now relabel /dev after loading the SMACK policy
4635 according to SMACK rules.
4637 * A new unit file option AppArmorProfile= has been added to
4638 set the AppArmor profile for the processes of a unit.
4640 * A new condition check ConditionArchitecture= has been added
4641 to conditionalize units based on the system architecture, as
4642 reported by uname()'s "machine" field.
4644 * systemd-networkd now supports matching on the system
4645 virtualization, architecture, kernel command line, host name
4648 * logind is now a lot more aggressive when suspending the
4649 machine due to a closed laptop lid. Instead of acting only
4650 on the lid close action, it will continuously watch the lid
4651 status and act on it. This is useful for laptops where the
4652 power button is on the outside of the chassis so that it can
4653 be reached without opening the lid (such as the Lenovo
4654 Yoga). On those machines, logind will now immediately
4655 re-suspend the machine if the power button has been
4656 accidentally pressed while the laptop was suspended and in a
4657 backpack or similar.
4659 * logind will now watch SW_DOCK switches and inhibit reaction
4660 to the lid switch if it is pressed. This means that logind
4661 will not suspend the machine anymore if the lid is closed
4662 and the system is docked, if the laptop supports SW_DOCK
4663 notifications via the input layer. Note that ACPI docking
4664 stations do not generate this currently. Also note that this
4665 logic is usually not fully sufficient and Desktop
4666 Environments should take a lid switch inhibitor lock when an
4667 external display is connected, as systemd will not watch
4670 * nspawn will now make use of the devices cgroup controller by
4671 default, and only permit creation of and access to the usual
4672 API device nodes like /dev/null or /dev/random, as well as
4673 access to (but not creation of) the pty devices.
4675 * We will now ship a default .network file for
4676 systemd-networkd that automatically configures DHCP for
4677 network interfaces created by nspawn's --network-veth or
4678 --network-bridge= switches.
4680 * systemd will now understand the usual M, K, G, T suffixes
4681 according to SI conventions (i.e. to the base 1000) when
4682 referring to throughput and hardware metrics. It will stay
4683 with IEC conventions (i.e. to the base 1024) for software
4684 metrics, according to what is customary according to
4685 Wikipedia. We explicitly document which base applies for
4686 each configuration option.
4688 * The DeviceAllow= setting in unit files now supports a syntax
4689 to whitelist an entire group of devices node majors at once,
4690 based on the /proc/devices listing. For example, with the
4691 string "char-pts", it is now possible to whitelist all
4692 current and future pseudo-TTYs at once.
4694 * sd-event learned a new "post" event source. Event sources of
4695 this type are triggered by the dispatching of any event
4696 source of a type that is not "post". This is useful for
4697 implementing clean-up and check event sources that are
4698 triggered by other work being done in the program.
4700 * systemd-networkd is no longer statically enabled, but uses
4701 the usual [Install] sections so that it can be
4702 enabled/disabled using systemctl. It still is enabled by
4705 * When creating a veth interface pair with systemd-nspawn, the
4706 host side will now be prefixed with "vb-" if
4707 --network-bridge= is used, and with "ve-" if --network-veth
4708 is used. This way, it is easy to distinguish these cases on
4709 the host, for example to apply different configuration to
4710 them with systemd-networkd.
4712 * The compatibility libraries for libsystemd-journal.so,
4713 libsystem-id128.so, libsystemd-login.so and
4714 libsystemd-daemon.so do not make use of IFUNC
4715 anymore. Instead, we now build libsystemd.so multiple times
4716 under these alternative names. This means that the footprint
4717 is drastically increased, but given that these are
4718 transitional compatibility libraries, this should not matter
4719 much. This change has been made necessary to support the ARM
4720 platform for these compatibility libraries, as the ARM
4721 toolchain is not really at the same level as the toolchain
4722 for other architectures like x86 and does not support
4723 IFUNC. Please make sure to use --enable-compat-libs only
4724 during a transitional period!
4726 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
4727 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
4728 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
4729 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
4730 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
4731 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
4732 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
4733 Zbigniew Jędrzejewski-Szmek
4735 — Berlin, 2014-02-24
4739 * A new component "systemd-networkd" has been added that can
4740 be used to configure local network interfaces statically or
4741 via DHCP. It is capable of bringing up bridges, VLANs, and
4742 bonding. Currently, no hook-ups for interactive network
4743 configuration are provided. Use this for your initrd,
4744 container, embedded, or server setup if you need a simple,
4745 yet powerful, network configuration solution. This
4746 configuration subsystem is quite nifty, as it allows wildcard
4747 hotplug matching in interfaces. For example, with a single
4748 configuration snippet, you can configure that all Ethernet
4749 interfaces showing up are automatically added to a bridge,
4750 or similar. It supports link-sensing and more.
4752 * A new tool "systemd-socket-proxyd" has been added which can
4753 act as a bidirectional proxy for TCP sockets. This is
4754 useful for adding socket activation support to services that
4755 do not actually support socket activation, including virtual
4756 machines and the like.
4758 * Add a new tool to save/restore rfkill state on
4761 * Save/restore state of keyboard backlights in addition to
4762 display backlights on shutdown/boot.
4764 * udev learned a new SECLABEL{} construct to label device
4765 nodes with a specific security label when they appear. For
4766 now, only SECLABEL{selinux} is supported, but the syntax is
4767 prepared for additional security frameworks.
4769 * udev gained a new scheme to configure link-level attributes
4770 from files in /etc/systemd/network/*.link. These files can
4771 match against MAC address, device path, driver name and type,
4772 and will apply attributes like the naming policy, link speed,
4773 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
4774 address assignment policy (randomized, ...).
4776 * The configuration of network interface naming rules for
4777 "permanent interface names" has changed: a new NamePolicy=
4778 setting in the [Link] section of .link files determines the
4779 priority of possible naming schemes (onboard, slot, MAC,
4780 path). The default value of this setting is determined by
4781 /usr/lib/net/links/99-default.link. Old
4782 80-net-name-slot.rules udev configuration file has been
4783 removed, so local configuration overriding this file should
4784 be adapted to override 99-default.link instead.
4786 * When the User= switch is used in a unit file, also
4787 initialize $SHELL= based on the user database entry.
4789 * systemd no longer depends on libdbus. All communication is
4790 now done with sd-bus, systemd's low-level bus library
4793 * kdbus support has been added to PID 1 itself. When kdbus is
4794 enabled, this causes PID 1 to set up the system bus and
4795 enable support for a new ".busname" unit type that
4796 encapsulates bus name activation on kdbus. It works a little
4797 bit like ".socket" units, except for bus names. A new
4798 generator has been added that converts classic dbus1 service
4799 activation files automatically into native systemd .busname
4802 * sd-bus: add a light-weight vtable implementation that allows
4803 defining objects on the bus with a simple static const
4804 vtable array of its methods, signals and properties.
4806 * systemd will not generate or install static dbus
4807 introspection data anymore to /usr/share/dbus-1/interfaces,
4808 as the precise format of these files is unclear, and
4809 nothing makes use of it.
4811 * A proxy daemon is now provided to proxy clients connecting
4812 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
4813 compatibility with classic D-Bus.
4815 * A bus driver implementation has been added that supports the
4816 classic D-Bus bus driver calls on kdbus, also for
4817 compatibility purposes.
4819 * A new API "sd-event.h" has been added that implements a
4820 minimal event loop API built around epoll. It provides a
4821 couple of features that direct epoll usage is lacking:
4822 prioritization of events, scales to large numbers of timer
4823 events, per-event timer slack (accuracy), system-wide
4824 coalescing of timer events, exit handlers, watchdog
4825 supervision support using systemd's sd_notify() API, child
4828 * A new API "sd-rntl.h" has been added that provides an API
4829 around the route netlink interface of the kernel, similar in
4830 style to "sd-bus.h".
4832 * A new API "sd-dhcp-client.h" has been added that provides a
4833 small DHCPv4 client-side implementation. This is used by
4836 * There is a new kernel command line option
4837 "systemd.restore_state=0|1". When set to "0", none of the
4838 systemd tools will restore saved runtime state to hardware
4839 devices. More specifically, the rfkill and backlight states
4842 * The FsckPassNo= compatibility option in mount/service units
4843 has been removed. The fstab generator will now add the
4844 necessary dependencies automatically, and does not require
4845 PID1's support for that anymore.
4847 * journalctl gained a new switch, --list-boots, that lists
4848 recent boots with their times and boot IDs.
4850 * The various tools like systemctl, loginctl, timedatectl,
4851 busctl, systemd-run, ... have gained a new switch "-M" to
4852 connect to a specific, local OS container (as direct
4853 connection, without requiring SSH). This works on any
4854 container that is registered with machined, such as those
4855 created by libvirt-lxc or nspawn.
4857 * systemd-run and systemd-analyze also gained support for "-H"
4858 to connect to remote hosts via SSH. This is particularly
4859 useful for systemd-run because it enables queuing of jobs
4860 onto remote systems.
4862 * machinectl gained a new command "login" to open a getty
4863 login in any local container. This works with any container
4864 that is registered with machined (such as those created by
4865 libvirt-lxc or nspawn), and which runs systemd inside.
4867 * machinectl gained a new "reboot" command that may be used to
4868 trigger a reboot on a specific container that is registered
4869 with machined. This works on any container that runs an init
4870 system of some kind.
4872 * systemctl gained a new "list-timers" command to print a nice
4873 listing of installed timer units with the times they elapse
4876 * Alternative reboot() parameters may now be specified on the
4877 "systemctl reboot" command line and are passed to the
4878 reboot() system call.
4880 * systemctl gained a new --job-mode= switch to configure the
4881 mode to queue a job with. This is a more generic version of
4882 --fail, --irreversible, and --ignore-dependencies, which are
4883 still available but not advertised anymore.
4885 * /etc/systemd/system.conf gained new settings to configure
4886 various default timeouts of units, as well as the default
4887 start limit interval and burst. These may still be overridden
4890 * PID1 will now export on the bus profile data of the security
4891 policy upload process (such as the SELinux policy upload to
4894 * journald: when forwarding logs to the console, include
4895 timestamps (following the setting in
4896 /sys/module/printk/parameters/time).
4898 * OnCalendar= in timer units now understands the special
4899 strings "yearly" and "annually". (Both are equivalent)
4901 * The accuracy of timer units is now configurable with the new
4902 AccuracySec= setting. It defaults to 1min.
4904 * A new dependency type JoinsNamespaceOf= has been added that
4905 allows running two services within the same /tmp and network
4906 namespace, if PrivateNetwork= or PrivateTmp= are used.
4908 * A new command "cat" has been added to systemctl. It outputs
4909 the original unit file of a unit, and concatenates the
4910 contents of additional "drop-in" unit file snippets, so that
4911 the full configuration is shown.
4913 * systemctl now supports globbing on the various "list-xyz"
4914 commands, like "list-units" or "list-sockets", as well as on
4915 those commands which take multiple unit names.
4917 * journalctl's --unit= switch gained support for globbing.
4919 * All systemd daemons now make use of the watchdog logic so
4920 that systemd automatically notices when they hang.
4922 * If the $container_ttys environment variable is set,
4923 getty-generator will automatically spawn a getty for each
4924 listed tty. This is useful for container managers to request
4925 login gettys to be spawned on as many ttys as needed.
4927 * %h, %s, %U specifier support is not available anymore when
4928 used in unit files for PID 1. This is because NSS calls are
4929 not safe from PID 1. They stay available for --user
4930 instances of systemd, and as special case for the root user.
4932 * loginctl gained a new "--no-legend" switch to turn off output
4935 * The "sd-login.h" API gained three new calls:
4936 sd_session_is_remote(), sd_session_get_remote_user(),
4937 sd_session_get_remote_host() to query information about
4940 * The udev hardware database now also carries vendor/product
4941 information of SDIO devices.
4943 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
4944 determine whether watchdog notifications are requested by
4947 * Socket-activated per-connection services now include a
4948 short description of the connection parameters in the
4951 * tmpfiles gained a new "--boot" option. When this is not used,
4952 only lines where the command character is not suffixed with
4953 "!" are executed. When this option is specified, those
4954 options are executed too. This partitions tmpfiles
4955 directives into those that can be safely executed at any
4956 time, and those which should be run only at boot (for
4957 example, a line that creates /run/nologin).
4959 * A new API "sd-resolve.h" has been added which provides a simple
4960 asynchronous wrapper around glibc NSS host name resolution
4961 calls, such as getaddrinfo(). In contrast to glibc's
4962 getaddrinfo_a(), it does not use signals. In contrast to most
4963 other asynchronous name resolution libraries, this one does
4964 not reimplement DNS, but reuses NSS, so that alternate
4965 host name resolution systems continue to work, such as mDNS,
4966 LDAP, etc. This API is based on libasyncns, but it has been
4967 cleaned up for inclusion in systemd.
4969 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
4970 "sd-daemon.h" are no longer found in individual libraries
4971 libsystemd-journal.so, libsystemd-login.so,
4972 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
4973 merged them into a single library, libsystemd.so, which
4974 provides all symbols. The reason for this is cyclic
4975 dependencies, as these libraries tend to use each other's
4976 symbols. So far, we have managed to workaround that by linking
4977 a copy of a good part of our code into each of these
4978 libraries again and again, which, however, makes certain
4979 things hard to do, like sharing static variables. Also, it
4980 substantially increases footprint. With this change, there
4981 is only one library for the basic APIs systemd
4982 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
4983 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
4984 library as well, however are subject to the --enable-kdbus
4985 switch (see below). Note that "sd-dhcp-client.h" is not part
4986 of this library (this is because it only consumes, never
4987 provides, services of/to other APIs). To make the transition
4988 easy from the separate libraries to the unified one, we
4989 provide the --enable-compat-libs compile-time switch which
4990 will generate stub libraries that are compatible with the
4991 old ones but redirect all calls to the new one.
4993 * All of the kdbus logic and the new APIs "sd-bus.h",
4994 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
4995 and "sd-utf8.h" are compile-time optional via the
4996 "--enable-kdbus" switch, and they are not compiled in by
4997 default. To make use of kdbus, you have to explicitly enable
4998 the switch. Note however, that neither the kernel nor the
4999 userspace API for all of this is considered stable yet. We
5000 want to maintain the freedom to still change the APIs for
5001 now. By specifying this build-time switch, you acknowledge
5002 that you are aware of the instability of the current
5005 * Also, note that while kdbus is pretty much complete,
5006 it lacks one thing: proper policy support. This means you
5007 can build a fully working system with all features; however,
5008 it will be highly insecure. Policy support will be added in
5009 one of the next releases, at the same time that we will
5010 declare the APIs stable.
5012 * When the kernel command line argument "kdbus" is specified,
5013 systemd will automatically load the kdbus.ko kernel module. At
5014 this stage of development, it is only useful for testing kdbus
5015 and should not be used in production. Note: if "--enable-kdbus"
5016 is specified, and the kdbus.ko kernel module is available, and
5017 "kdbus" is added to the kernel command line, the entire system
5018 runs with kdbus instead of dbus-daemon, with the above mentioned
5019 problem of missing the system policy enforcement. Also a future
5020 version of kdbus.ko or a newer systemd will not be compatible with
5021 each other, and will unlikely be able to boot the machine if only
5022 one of them is updated.
5024 * systemctl gained a new "import-environment" command which
5025 uploads the caller's environment (or parts thereof) into the
5026 service manager so that it is inherited by services started
5027 by the manager. This is useful to upload variables like
5028 $DISPLAY into the user service manager.
5030 * A new PrivateDevices= switch has been added to service units
5031 which allows running a service with a namespaced /dev
5032 directory that does not contain any device nodes for
5033 physical devices. More specifically, it only includes devices
5034 such as /dev/null, /dev/urandom, and /dev/zero which are API
5037 * logind has been extended to support behaviour like VT
5038 switching on seats that do not support a VT. This makes
5039 multi-session available on seats that are not the first seat
5040 (seat0), and on systems where kernel support for VTs has
5041 been disabled at compile-time.
5043 * If a process holds a delay lock for system sleep or shutdown
5044 and fails to release it in time, we will now log its
5045 identity. This makes it easier to identify processes that
5046 cause slow suspends or power-offs.
5048 * When parsing /etc/crypttab, support for a new key-slot=
5049 option as supported by Debian is added. It allows indicating
5050 which LUKS slot to use on disk, speeding up key loading.
5052 * The sd_journald_sendv() API call has been checked and
5053 officially declared to be async-signal-safe so that it may
5054 be invoked from signal handlers for logging purposes.
5056 * Boot-time status output is now enabled automatically after a
5057 short timeout if boot does not progress, in order to give
5058 the user an indication what she or he is waiting for.
5060 * The boot-time output has been improved to show how much time
5061 remains until jobs expire.
5063 * The KillMode= switch in service units gained a new possible
5064 value "mixed". If set, and the unit is shut down, then the
5065 initial SIGTERM signal is sent only to the main daemon
5066 process, while the following SIGKILL signal is sent to
5067 all remaining processes of the service.
5069 * When a scope unit is registered, a new property "Controller"
5070 may be set. If set to a valid bus name, systemd will send a
5071 RequestStop() signal to this name when it would like to shut
5072 down the scope. This may be used to hook manager logic into
5073 the shutdown logic of scope units. Also, scope units may now
5074 be put in a special "abandoned" state, in which case the
5075 manager process which created them takes no further
5076 responsibilities for it.
5078 * When reading unit files, systemd will now verify
5079 the access mode of these files, and warn about certain
5080 suspicious combinations. This has been added to make it
5081 easier to track down packaging bugs where unit files are
5082 marked executable or world-writable.
5084 * systemd-nspawn gained a new "--setenv=" switch to set
5085 container-wide environment variables. The similar option in
5086 systemd-activate was renamed from "--environment=" to
5087 "--setenv=" for consistency.
5089 * systemd-nspawn has been updated to create a new kdbus domain
5090 for each container that is invoked, thus allowing each
5091 container to have its own set of system and user buses,
5092 independent of the host.
5094 * systemd-nspawn gained a new --drop-capability= switch to run
5095 the container with less capabilities than the default. Both
5096 --drop-capability= and --capability= now take the special
5097 string "all" for dropping or keeping all capabilities.
5099 * systemd-nspawn gained new switches for executing containers
5100 with specific SELinux labels set.
5102 * systemd-nspawn gained a new --quiet switch to not generate
5103 any additional output but the container's own console
5106 * systemd-nspawn gained a new --share-system switch to run a
5107 container without PID namespacing enabled.
5109 * systemd-nspawn gained a new --register= switch to control
5110 whether the container is registered with systemd-machined or
5111 not. This is useful for containers that do not run full
5112 OS images, but only specific apps.
5114 * systemd-nspawn gained a new --keep-unit which may be used
5115 when invoked as the only program from a service unit, and
5116 results in registration of the unit service itself in
5117 systemd-machined, instead of a newly opened scope unit.
5119 * systemd-nspawn gained a new --network-interface= switch for
5120 moving arbitrary interfaces to the container. The new
5121 --network-veth switch creates a virtual Ethernet connection
5122 between host and container. The new --network-bridge=
5123 switch then allows assigning the host side of this virtual
5124 Ethernet connection to a bridge device.
5126 * systemd-nspawn gained a new --personality= switch for
5127 setting the kernel personality for the container. This is
5128 useful when running a 32-bit container on a 64-bit host. A
5129 similar option Personality= is now also available for service
5132 * logind will now also track a "Desktop" identifier for each
5133 session which encodes the desktop environment of it. This is
5134 useful for desktop environments that want to identify
5135 multiple running sessions of itself easily.
5137 * A new SELinuxContext= setting for service units has been
5138 added that allows setting a specific SELinux execution
5139 context for a service.
5141 * Most systemd client tools will now honour $SYSTEMD_LESS for
5142 settings of the "less" pager. By default, these tools will
5143 override $LESS to allow certain operations to work, such as
5144 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
5145 influence this logic.
5147 * systemd's "seccomp" hook-up has been changed to make use of
5148 the libseccomp library instead of using its own
5149 implementation. This has benefits for portability among
5152 * For usage together with SystemCallFilter=, a new
5153 SystemCallErrorNumber= setting has been introduced that
5154 allows configuration of a system error number to be returned
5155 on filtered system calls, instead of immediately killing the
5156 process. Also, SystemCallArchitectures= has been added to
5157 limit access to system calls of a particular architecture
5158 (in order to turn off support for unused secondary
5159 architectures). There is also a global
5160 SystemCallArchitectures= setting in system.conf now to turn
5161 off support for non-native system calls system-wide.
5163 * systemd requires a kernel with a working name_to_handle_at(),
5164 please see the kernel config requirements in the README file.
5166 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
5167 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
5168 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
5169 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
5170 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
5171 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
5172 Elia Pinto, Florian Weimer, George McCollister, Goffredo
5173 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
5174 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
5175 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
5176 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
5177 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
5178 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
5179 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
5180 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
5181 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
5182 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
5183 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
5184 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
5185 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
5186 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
5187 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
5188 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
5189 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
5191 — Berlin, 2014-02-20
5195 * logind has gained support for facilitating privileged input
5196 and drm device access for unprivileged clients. This work is
5197 useful to allow Wayland display servers (and similar
5198 programs, such as kmscon) to run under the user's ID and
5199 access input and drm devices which are normally
5200 protected. When this is used (and the kernel is new enough)
5201 logind will "mute" IO on the file descriptors passed to
5202 Wayland as long as it is in the background and "unmute" it
5203 if it returns into the foreground. This allows secure
5204 session switching without allowing background sessions to
5205 eavesdrop on input and display data. This also introduces
5206 session switching support if VT support is turned off in the
5207 kernel, and on seats that are not seat0.
5209 * A new kernel command line option luks.options= is understood
5210 now which allows specifying LUKS options for usage for LUKS
5211 encrypted partitions specified with luks.uuid=.
5213 * tmpfiles.d(5) snippets may now use specifier expansion in
5214 path names. More specifically %m, %b, %H, %v, are now
5215 replaced by the local machine id, boot id, hostname, and
5216 kernel version number.
5218 * A new tmpfiles.d(5) command "m" has been introduced which
5219 may be used to change the owner/group/access mode of a file
5220 or directory if it exists, but do nothing if it does not.
5222 * This release removes high-level support for the
5223 MemorySoftLimit= cgroup setting. The underlying kernel
5224 cgroup attribute memory.soft_limit= is currently badly
5225 designed and likely to be removed from the kernel API in its
5226 current form, hence we should not expose it for now.
5228 * The memory.use_hierarchy cgroup attribute is now enabled for
5229 all cgroups systemd creates in the memory cgroup
5230 hierarchy. This option is likely to be come the built-in
5231 default in the kernel anyway, and the non-hierarchical mode
5232 never made much sense in the intrinsically hierarchical
5235 * A new field _SYSTEMD_SLICE= is logged along with all journal
5236 messages containing the slice a message was generated
5237 from. This is useful to allow easy per-customer filtering of
5238 logs among other things.
5240 * systemd-journald will no longer adjust the group of journal
5241 files it creates to the "systemd-journal" group. Instead we
5242 rely on the journal directory to be owned by the
5243 "systemd-journal" group, and its setgid bit set, so that the
5244 kernel file system layer will automatically enforce that
5245 journal files inherit this group assignment. The reason for
5246 this change is that we cannot allow NSS look-ups from
5247 journald which would be necessary to resolve
5248 "systemd-journal" to a numeric GID, because this might
5249 create deadlocks if NSS involves synchronous queries to
5250 other daemons (such as nscd, or sssd) which in turn are
5251 logging clients of journald and might block on it, which
5252 would then dead lock. A tmpfiles.d(5) snippet included in
5253 systemd will make sure the setgid bit and group are
5254 properly set on the journal directory if it exists on every
5255 boot. However, we recommend adjusting it manually after
5256 upgrades too (or from RPM scriptlets), so that the change is
5257 not delayed until next reboot.
5259 * Backlight and random seed files in /var/lib/ have moved into
5260 the /var/lib/systemd/ directory, in order to centralize all
5261 systemd generated files in one directory.
5263 * Boot time performance measurements (as displayed by
5264 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
5265 performance information if that's available to determine how
5266 much time BIOS and boot loader initialization required. With
5267 a sufficiently new BIOS you hence no longer need to boot
5268 with Gummiboot to get access to such information.
5270 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
5271 Cristian Rodríguez, Dave Reisner, David Herrmann, David
5272 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
5273 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
5274 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
5275 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
5276 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
5278 — Berlin, 2013-10-02
5282 * The Restart= option for services now understands a new
5283 on-watchdog setting, which will restart the service
5284 automatically if the service stops sending out watchdog keep
5285 alive messages (as configured with WatchdogSec=).
5287 * The getty generator (which is responsible for bringing up a
5288 getty on configured serial consoles) will no longer only
5289 start a getty on the primary kernel console but on all
5290 others, too. This makes the order in which console= is
5291 specified on the kernel command line less important.
5293 * libsystemd-logind gained a new sd_session_get_vt() call to
5294 retrieve the VT number of a session.
5296 * If the option "tries=0" is set for an entry of /etc/crypttab
5297 its passphrase is queried indefinitely instead of any
5298 maximum number of tries.
5300 * If a service with a configure PID file terminates its PID
5301 file will now be removed automatically if it still exists
5302 afterwards. This should put an end to stale PID files.
5304 * systemd-run will now also take relative binary path names
5305 for execution and no longer insists on absolute paths.
5307 * InaccessibleDirectories= and ReadOnlyDirectories= now take
5308 paths that are optionally prefixed with "-" to indicate that
5309 it should not be considered a failure if they do not exist.
5311 * journalctl -o (and similar commands) now understands a new
5312 output mode "short-precise", it is similar to "short" but
5313 shows timestamps with usec accuracy.
5315 * The option "discard" (as known from Debian) is now
5316 synonymous to "allow-discards" in /etc/crypttab. In fact,
5317 "discard" is preferred now (since it is easier to remember
5320 * Some licensing clean-ups were made, so that more code is now
5321 LGPL-2.1 licensed than before.
5323 * A minimal tool to save/restore the display backlight
5324 brightness across reboots has been added. It will store the
5325 backlight setting as late as possible at shutdown, and
5326 restore it as early as possible during reboot.
5328 * A logic to automatically discover and enable home and swap
5329 partitions on GPT disks has been added. With this in place
5330 /etc/fstab becomes optional for many setups as systemd can
5331 discover certain partitions located on the root disk
5332 automatically. Home partitions are recognized under their
5333 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
5334 partitions are recognized under their GPT type ID
5335 0657fd6da4ab43c484e50933c84b4f4f.
5337 * systemd will no longer pass any environment from the kernel
5338 or initrd to system services. If you want to set an
5339 environment for all services, do so via the kernel command
5340 line systemd.setenv= assignment.
5342 * The systemd-sysctl tool no longer natively reads the file
5343 /etc/sysctl.conf. If desired, the file should be symlinked
5344 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
5345 legacy support by a symlink rather than built-in code, it
5346 also makes the otherwise hidden order of application of the
5347 different files visible. (Note that this partly reverts to a
5348 pre-198 application order of sysctl knobs!)
5350 * The "systemctl set-log-level" and "systemctl dump" commands
5351 have been moved to systemd-analyze.
5353 * systemd-run learned the new --remain-after-exit switch,
5354 which causes the scope unit not to be cleaned up
5355 automatically after the process terminated.
5357 * tmpfiles learned a new --exclude-prefix= switch to exclude
5358 certain paths from operation.
5360 * journald will now automatically flush all messages to disk
5361 as soon as a message at the log level CRIT, ALERT or EMERG
5364 Contributions from: Andrew Cook, Brandon Philips, Christian
5365 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
5366 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
5367 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
5368 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
5369 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
5370 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
5371 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
5372 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
5373 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
5374 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
5375 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
5376 William Giokas, Zbigniew Jędrzejewski-Szmek
5378 — Berlin, 2013-09-13
5382 * The documentation has been updated to cover the various new
5383 concepts introduced with 205.
5385 * Unit files now understand the new %v specifier which
5386 resolves to the kernel version string as returned by "uname
5389 * systemctl now supports filtering the unit list output by
5390 load state, active state and sub state, using the new
5393 * "systemctl status" will now show the results of the
5394 condition checks (like ConditionPathExists= and similar) of
5395 the last start attempts of the unit. They are also logged to
5398 * "journalctl -b" may now be used to look for boot output of a
5399 specific boot. Try "journalctl -b -1" for the previous boot,
5400 but the syntax is substantially more powerful.
5402 * "journalctl --show-cursor" has been added which prints the
5403 cursor string the last shown log line. This may then be used
5404 with the new "journalctl --after-cursor=" switch to continue
5405 browsing logs from that point on.
5407 * "journalctl --force" may now be used to force regeneration
5410 * Creation of "dead" device nodes has been moved from udev
5411 into kmod and tmpfiles. Previously, udev would read the kmod
5412 databases to pre-generate dead device nodes based on meta
5413 information contained in kernel modules, so that these would
5414 be auto-loaded on access rather then at boot. As this
5415 does not really have much to do with the exposing actual
5416 kernel devices to userspace this has always been slightly
5417 alien in the udev codebase. Following the new scheme kmod
5418 will now generate a runtime snippet for tmpfiles from the
5419 module meta information and it now is tmpfiles' job to the
5420 create the nodes. This also allows overriding access and
5421 other parameters for the nodes using the usual tmpfiles
5422 facilities. As side effect this allows us to remove the
5423 CAP_SYS_MKNOD capability bit from udevd entirely.
5425 * logind's device ACLs may now be applied to these "dead"
5426 devices nodes too, thus finally allowing managed access to
5427 devices such as /dev/snd/sequencer without loading the
5428 backing module right-away.
5430 * A new RPM macro has been added that may be used to apply
5431 tmpfiles configuration during package installation.
5433 * systemd-detect-virt and ConditionVirtualization= now can
5434 detect User-Mode-Linux machines (UML).
5436 * journald will now implicitly log the effective capabilities
5437 set of processes in the message metadata.
5439 * systemd-cryptsetup has gained support for TrueCrypt volumes.
5441 * The initrd interface has been simplified (more specifically,
5442 support for passing performance data via environment
5443 variables and fsck results via files in /run has been
5444 removed). These features were non-essential, and are
5445 nowadays available in a much nicer way by having systemd in
5446 the initrd serialize its state and have the hosts systemd
5447 deserialize it again.
5449 * The udev "keymap" data files and tools to apply keyboard
5450 specific mappings of scan to key codes, and force-release
5451 scan code lists have been entirely replaced by a udev
5452 "keyboard" builtin and a hwdb data file.
5454 * systemd will now honour the kernel's "quiet" command line
5455 argument also during late shutdown, resulting in a
5456 completely silent shutdown when used.
5458 * There's now an option to control the SO_REUSEPORT socket
5459 option in .socket units.
5461 * Instance units will now automatically get a per-template
5462 subslice of system.slice unless something else is explicitly
5463 configured. For example, instances of sshd@.service will now
5464 implicitly be placed in system-sshd.slice rather than
5465 system.slice as before.
5467 * Test coverage support may now be enabled at build time.
5469 Contributions from: Dave Reisner, Frederic Crozat, Harald
5470 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
5471 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
5472 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
5473 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
5474 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
5475 Giokas, Zbigniew Jędrzejewski-Szmek
5477 — Berlin, 2013-07-23
5481 * Two new unit types have been introduced:
5483 Scope units are very similar to service units, however, are
5484 created out of pre-existing processes — instead of PID 1
5485 forking off the processes. By using scope units it is
5486 possible for system services and applications to group their
5487 own child processes (worker processes) in a powerful way
5488 which then maybe used to organize them, or kill them
5489 together, or apply resource limits on them.
5491 Slice units may be used to partition system resources in an
5492 hierarchical fashion and then assign other units to them. By
5493 default there are now three slices: system.slice (for all
5494 system services), user.slice (for all user sessions),
5495 machine.slice (for VMs and containers).
5497 Slices and scopes have been introduced primarily in
5498 context of the work to move cgroup handling to a
5499 single-writer scheme, where only PID 1
5500 creates/removes/manages cgroups.
5502 * There's a new concept of "transient" units. In contrast to
5503 normal units these units are created via an API at runtime,
5504 not from configuration from disk. More specifically this
5505 means it is now possible to run arbitrary programs as
5506 independent services, with all execution parameters passed
5507 in via bus APIs rather than read from disk. Transient units
5508 make systemd substantially more dynamic then it ever was,
5509 and useful as a general batch manager.
5511 * logind has been updated to make use of scope and slice units
5512 for managing user sessions. As a user logs in he will get
5513 his own private slice unit, to which all sessions are added
5514 as scope units. We also added support for automatically
5515 adding an instance of user@.service for the user into the
5516 slice. Effectively logind will no longer create cgroup
5517 hierarchies on its own now, it will defer entirely to PID 1
5518 for this by means of scope, service and slice units. Since
5519 user sessions this way become entities managed by PID 1
5520 the output of "systemctl" is now a lot more comprehensive.
5522 * A new mini-daemon "systemd-machined" has been added which
5523 may be used by virtualization managers to register local
5524 VMs/containers. nspawn has been updated accordingly, and
5525 libvirt will be updated shortly. machined will collect a bit
5526 of meta information about the VMs/containers, and assign
5527 them their own scope unit (see above). The collected
5528 meta-data is then made available via the "machinectl" tool,
5529 and exposed in "ps" and similar tools. machined/machinectl
5530 is compile-time optional.
5532 * As discussed earlier, the low-level cgroup configuration
5533 options ControlGroup=, ControlGroupModify=,
5534 ControlGroupPersistent=, ControlGroupAttribute= have been
5535 removed. Please use high-level attribute settings instead as
5536 well as slice units.
5538 * A new bus call SetUnitProperties() has been added to alter
5539 various runtime parameters of a unit. This is primarily
5540 useful to alter cgroup parameters dynamically in a nice way,
5541 but will be extended later on to make more properties
5542 modifiable at runtime. systemctl gained a new set-properties
5543 command that wraps this call.
5545 * A new tool "systemd-run" has been added which can be used to
5546 run arbitrary command lines as transient services or scopes,
5547 while configuring a number of settings via the command
5548 line. This tool is currently very basic, however already
5549 very useful. We plan to extend this tool to even allow
5550 queuing of execution jobs with time triggers from the
5551 command line, similar in fashion to "at".
5553 * nspawn will now inform the user explicitly that kernels with
5554 audit enabled break containers, and suggest the user to turn
5557 * Support for detecting the IMA and AppArmor security
5558 frameworks with ConditionSecurity= has been added.
5560 * journalctl gained a new "-k" switch for showing only kernel
5561 messages, mimicking dmesg output; in addition to "--user"
5562 and "--system" switches for showing only user's own logs
5565 * systemd-delta can now show information about drop-in
5566 snippets extending unit files.
5568 * libsystemd-bus has been substantially updated but is still
5569 not available as public API.
5571 * systemd will now look for the "debug" argument on the kernel
5572 command line and enable debug logging, similar to what
5573 "systemd.log_level=debug" already did before.
5575 * "systemctl set-default", "systemctl get-default" has been
5576 added to configure the default.target symlink, which
5577 controls what to boot into by default.
5579 * "systemctl set-log-level" has been added as a convenient
5580 way to raise and lower systemd logging threshold.
5582 * "systemd-analyze plot" will now show the time the various
5583 generators needed for execution, as well as information
5584 about the unit file loading.
5586 * libsystemd-journal gained a new sd_journal_open_files() call
5587 for opening specific journal files. journactl also gained a
5588 new switch to expose this new functionality. Previously we
5589 only supported opening all files from a directory, or all
5590 files from the system, as opening individual files only is
5591 racy due to journal file rotation.
5593 * systemd gained the new DefaultEnvironment= setting in
5594 /etc/systemd/system.conf to set environment variables for
5597 * If a privileged process logs a journal message with the
5598 OBJECT_PID= field set, then journald will automatically
5599 augment this with additional OBJECT_UID=, OBJECT_GID=,
5600 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
5601 system services want to log events about specific client
5602 processes. journactl/systemctl has been updated to make use
5603 of this information if all log messages regarding a specific
5606 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
5607 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
5608 Reisner, David Coppa, David King, David Strauss, Eelco
5609 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
5610 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
5611 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
5612 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
5613 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
5614 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
5615 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
5616 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
5617 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
5618 Łukasz Stelmach, 장동준
5622 * The Python bindings gained some minimal support for the APIs
5623 exposed by libsystemd-logind.
5625 * ConditionSecurity= gained support for detecting SMACK. Since
5626 this condition already supports SELinux and AppArmor we only
5627 miss IMA for this. Patches welcome!
5629 Contributions from: Karol Lewandowski, Lennart Poettering,
5630 Zbigniew Jędrzejewski-Szmek
5634 * systemd-nspawn will now create /etc/resolv.conf if
5635 necessary, before bind-mounting the host's file onto it.
5637 * systemd-nspawn will now store meta information about a
5638 container on the container's cgroup as extended attribute
5639 fields, including the root directory.
5641 * The cgroup hierarchy has been reworked in many ways. All
5642 objects any of the components systemd creates in the cgroup
5643 tree are now suffixed. More specifically, user sessions are
5644 now placed in cgroups suffixed with ".session", users in
5645 cgroups suffixed with ".user", and nspawn containers in
5646 cgroups suffixed with ".nspawn". Furthermore, all cgroup
5647 names are now escaped in a simple scheme to avoid collision
5648 of userspace object names with kernel filenames. This work
5649 is preparation for making these objects relocatable in the
5650 cgroup tree, in order to allow easy resource partitioning of
5651 these objects without causing naming conflicts.
5653 * systemctl list-dependencies gained the new switches
5654 --plain, --reverse, --after and --before.
5656 * systemd-inhibit now shows the process name of processes that
5657 have taken an inhibitor lock.
5659 * nss-myhostname will now also resolve "localhost"
5660 implicitly. This makes /etc/hosts an optional file and
5661 nicely handles that on IPv6 ::1 maps to both "localhost" and
5664 * libsystemd-logind.so gained a new call
5665 sd_get_machine_names() to enumerate running containers and
5666 VMs (currently only supported by very new libvirt and
5667 nspawn). sd_login_monitor can now be used to watch
5668 VMs/containers coming and going.
5670 * .include is not allowed recursively anymore, and only in
5671 unit files. Usually it is better to use drop-in snippets in
5672 .d/*.conf anyway, as introduced with systemd 198.
5674 * systemd-analyze gained a new "critical-chain" command that
5675 determines the slowest chain of units run during system
5676 boot-up. It is very useful for tracking down where
5677 optimizing boot time is the most beneficial.
5679 * systemd will no longer allow manipulating service paths in
5680 the name=systemd:/system cgroup tree using ControlGroup= in
5681 units. (But is still fine with it in all other dirs.)
5683 * There's a new systemd-nspawn@.service service file that may
5684 be used to easily run nspawn containers as system
5685 services. With the container's root directory in
5686 /var/lib/container/foobar it is now sufficient to run
5687 "systemctl start systemd-nspawn@foobar.service" to boot it.
5689 * systemd-cgls gained a new parameter "--machine" to list only
5690 the processes within a certain container.
5692 * ConditionSecurity= now can check for "apparmor". We still
5693 are lacking checks for SMACK and IMA for this condition
5694 check though. Patches welcome!
5696 * A new configuration file /etc/systemd/sleep.conf has been
5697 added that may be used to configure which kernel operation
5698 systemd is supposed to execute when "suspend", "hibernate"
5699 or "hybrid-sleep" is requested. This makes the new kernel
5700 "freeze" state accessible to the user.
5702 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
5703 the passed argument if applicable.
5705 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
5706 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
5707 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
5708 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
5709 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
5710 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
5711 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
5716 * The output of 'systemctl list-jobs' got some polishing. The
5717 '--type=' argument may now be passed more than once. A new
5718 command 'systemctl list-sockets' has been added which shows
5719 a list of kernel sockets systemd is listening on with the
5720 socket units they belong to, plus the units these socket
5723 * The experimental libsystemd-bus library got substantial
5724 updates to work in conjunction with the (also experimental)
5725 kdbus kernel project. It works well enough to exchange
5726 messages with some sophistication. Note that kdbus is not
5727 ready yet, and the library is mostly an elaborate test case
5728 for now, and not installable.
5730 * systemd gained a new unit 'systemd-static-nodes.service'
5731 that generates static device nodes earlier during boot, and
5732 can run in conjunction with udev.
5734 * libsystemd-login gained a new call sd_pid_get_user_unit()
5735 to retrieve the user systemd unit a process is running
5736 in. This is useful for systems where systemd is used as
5739 * systemd-nspawn now places all containers in the new /machine
5740 top-level cgroup directory in the name=systemd
5741 hierarchy. libvirt will soon do the same, so that we get a
5742 uniform separation of /system, /user and /machine for system
5743 services, user processes and containers/virtual
5744 machines. This new cgroup hierarchy is also useful to stick
5745 stable names to specific container instances, which can be
5746 recognized later this way (this name may be controlled
5747 via systemd-nspawn's new -M switch). libsystemd-login also
5748 gained a new call sd_pid_get_machine_name() to retrieve the
5749 name of the container/VM a specific process belongs to.
5751 * bootchart can now store its data in the journal.
5753 * libsystemd-journal gained a new call
5754 sd_journal_add_conjunction() for AND expressions to the
5755 matching logic. This can be used to express more complex
5756 logical expressions.
5758 * journactl can now take multiple --unit= and --user-unit=
5761 * The cryptsetup logic now understands the "luks.key=" kernel
5762 command line switch for specifying a file to read the
5763 decryption key from. Also, if a configured key file is not
5764 found the tool will now automatically fall back to prompting
5767 * Python systemd.journal module was updated to wrap recently
5768 added functions from libsystemd-journal. The interface was
5769 changed to bring the low level interface in s.j._Reader
5770 closer to the C API, and the high level interface in
5771 s.j.Reader was updated to wrap and convert all data about
5774 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
5775 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
5776 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
5777 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
5778 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
5779 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
5783 * journalctl --update-catalog now understands a new --root=
5784 option to operate on catalogs found in a different root
5787 * During shutdown after systemd has terminated all running
5788 services a final killing loop kills all remaining left-over
5789 processes. We will now print the name of these processes
5790 when we send SIGKILL to them, since this usually indicates a
5793 * If /etc/crypttab refers to password files stored on
5794 configured mount points automatic dependencies will now be
5795 generated to ensure the specific mount is established first
5796 before the key file is attempted to be read.
5798 * 'systemctl status' will now show information about the
5799 network sockets a socket unit is listening on.
5801 * 'systemctl status' will also shown information about any
5802 drop-in configuration file for units. (Drop-In configuration
5803 files in this context are files such as
5804 /etc/systemd/systemd/foobar.service.d/*.conf)
5806 * systemd-cgtop now optionally shows summed up CPU times of
5807 cgroups. Press '%' while running cgtop to switch between
5808 percentage and absolute mode. This is useful to determine
5809 which cgroups use up the most CPU time over the entire
5810 runtime of the system. systemd-cgtop has also been updated
5811 to be 'pipeable' for processing with further shell tools.
5813 * 'hostnamectl set-hostname' will now allow setting of FQDN
5816 * The formatting and parsing of time span values has been
5817 changed. The parser now understands fractional expressions
5818 such as "5.5h". The formatter will now output fractional
5819 expressions for all time spans under 1min, i.e. "5.123456s"
5820 rather than "5s 123ms 456us". For time spans under 1s
5821 millisecond values are shown, for those under 1ms
5822 microsecond values are shown. This should greatly improve
5823 all time-related output of systemd.
5825 * libsystemd-login and libsystemd-journal gained new
5826 functions for querying the poll() events mask and poll()
5827 timeout value for integration into arbitrary event
5830 * localectl gained the ability to list available X11 keymaps
5831 (models, layouts, variants, options).
5833 * 'systemd-analyze dot' gained the ability to filter for
5834 specific units via shell-style globs, to create smaller,
5835 more useful graphs. I.e. it is now possible to create simple
5836 graphs of all the dependencies between only target units, or
5837 of all units that Avahi has dependencies with.
5839 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
5840 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
5841 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
5842 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
5843 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
5844 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
5845 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
5849 * The boot-time readahead implementation for rotating media
5850 will now read the read-ahead data in multiple passes which
5851 consist of all read requests made in equidistant time
5852 intervals. This means instead of strictly reading read-ahead
5853 data in its physical order on disk we now try to find a
5854 middle ground between physical and access time order.
5856 * /etc/os-release files gained a new BUILD_ID= field for usage
5857 on operating systems that provide continuous builds of OS
5860 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
5861 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
5862 William Douglas, Zbigniew Jędrzejewski-Szmek
5866 * systemd-python gained an API exposing libsystemd-daemon.
5868 * The SMACK setup logic gained support for uploading CIPSO
5871 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
5872 ReadOnlyDirectories= and InaccessibleDirectories= has
5873 changed. The private /tmp and /var/tmp directories are now
5874 shared by all processes of a service (which means
5875 ExecStartPre= may now leave data in /tmp that ExecStart= of
5876 the same service can still access). When a service is
5877 stopped its temporary directories are immediately deleted
5878 (normal clean-up with tmpfiles is still done in addition to
5881 * By default, systemd will now set a couple of sysctl
5882 variables in the kernel: the safe sysrq options are turned
5883 on, IP route verification is turned on, and source routing
5884 disabled. The recently added hardlink and softlink
5885 protection of the kernel is turned on. These settings should
5886 be reasonably safe, and good defaults for all new systems.
5888 * The predictable network naming logic may now be turned off
5889 with a new kernel command line switch: net.ifnames=0.
5891 * A new libsystemd-bus module has been added that implements a
5892 pretty complete D-Bus client library. For details see:
5894 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
5896 * journald will now explicitly flush the journal files to disk
5897 at the latest 5min after each write. The file will then also
5898 be marked offline until the next write. This should increase
5899 reliability in case of a crash. The synchronization delay
5900 can be configured via SyncIntervalSec= in journald.conf.
5902 * There's a new remote-fs-setup.target unit that can be used
5903 to pull in specific services when at least one remote file
5904 system is to be mounted.
5906 * There are new targets timers.target and paths.target as
5907 canonical targets to pull user timer and path units in
5908 from. This complements sockets.target with a similar
5909 purpose for socket units.
5911 * libudev gained a new call udev_device_set_attribute_value()
5912 to set sysfs attributes of a device.
5914 * The udev daemon now sets the default number of worker
5915 processes executed in parallel based on the number of available
5916 CPUs instead of the amount of available RAM. This is supposed
5917 to provide a more reliable default and limit a too aggressive
5918 parallelism for setups with 1000s of devices connected.
5920 Contributions from: Auke Kok, Colin Walters, Cristian
5921 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
5922 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
5923 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
5924 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
5925 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
5926 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
5927 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
5928 Zbigniew Jędrzejewski-Szmek
5932 * Configuration of unit files may now be extended via drop-in
5933 files without having to edit/override the unit files
5934 themselves. More specifically, if the administrator wants to
5935 change one value for a service file foobar.service he can
5936 now do so by dropping in a configuration snippet into
5937 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
5938 will load all these snippets and apply them on top of the
5939 main unit configuration file, possibly extending or
5940 overriding its settings. Using these drop-in snippets is
5941 generally nicer than the two earlier options for changing
5942 unit files locally: copying the files from
5943 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
5944 them there; or creating a new file in /etc/systemd/system/
5945 that incorporates the original one via ".include". Drop-in
5946 snippets into these .d/ directories can be placed in any
5947 directory systemd looks for units in, and the usual
5948 overriding semantics between /usr/lib, /etc and /run apply
5951 * Most unit file settings which take lists of items can now be
5952 reset by assigning the empty string to them. For example,
5953 normally, settings such as Environment=FOO=BAR append a new
5954 environment variable assignment to the environment block,
5955 each time they are used. By assigning Environment= the empty
5956 string the environment block can be reset to empty. This is
5957 particularly useful with the .d/*.conf drop-in snippets
5958 mentioned above, since this adds the ability to reset list
5959 settings from vendor unit files via these drop-ins.
5961 * systemctl gained a new "list-dependencies" command for
5962 listing the dependencies of a unit recursively.
5964 * Inhibitors are now honored and listed by "systemctl
5965 suspend", "systemctl poweroff" (and similar) too, not only
5966 GNOME. These commands will also list active sessions by
5969 * Resource limits (as exposed by the various control group
5970 controllers) can now be controlled dynamically at runtime
5971 for all units. More specifically, you can now use a command
5972 like "systemctl set-cgroup-attr foobar.service cpu.shares
5973 2000" to alter the CPU shares a specific service gets. These
5974 settings are stored persistently on disk, and thus allow the
5975 administrator to easily adjust the resource usage of
5976 services with a few simple commands. This dynamic resource
5977 management logic is also available to other programs via the
5978 bus. Almost any kernel cgroup attribute and controller is
5981 * systemd-vconsole-setup will now copy all font settings to
5982 all allocated VTs, where it previously applied them only to
5985 * libsystemd-login gained the new sd_session_get_tty() API
5988 * This release drops support for a few legacy or
5989 distribution-specific LSB facility names when parsing init
5990 scripts: $x-display-manager, $mail-transfer-agent,
5991 $mail-transport-agent, $mail-transfer-agent, $smtp,
5992 $null. Also, the mail-transfer-agent.target unit backing
5993 this has been removed. Distributions which want to retain
5994 compatibility with this should carry the burden for
5995 supporting this themselves and patch support for these back
5996 in, if they really need to. Also, the facilities $syslog and
5997 $local_fs are now ignored, since systemd does not support
5998 early-boot LSB init scripts anymore, and these facilities
5999 are implied anyway for normal services. syslog.target has
6002 * There are new bus calls on PID1's Manager object for
6003 cancelling jobs, and removing snapshot units. Previously,
6004 both calls were only available on the Job and Snapshot
6007 * systemd-journal-gatewayd gained SSL support.
6009 * The various "environment" files, such as /etc/locale.conf
6010 now support continuation lines with a backslash ("\") as
6011 last character in the line, similarly in style (but different)
6012 to how this is supported in shells.
6014 * For normal user processes the _SYSTEMD_USER_UNIT= field is
6015 now implicitly appended to every log entry logged. systemctl
6016 has been updated to filter by this field when operating on a
6017 user systemd instance.
6019 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
6020 CAP_AUDIT_CONTROL capabilities to the capabilities set for
6021 the container. This makes it easier to boot unmodified
6022 Fedora systems in a container, which however still requires
6023 audit=0 to be passed on the kernel command line. Auditing in
6024 kernel and userspace is unfortunately still too broken in
6025 context of containers, hence we recommend compiling it out
6026 of the kernel or using audit=0. Hopefully this will be fixed
6027 one day for good in the kernel.
6029 * nspawn gained the new --bind= and --bind-ro= parameters to
6030 bind mount specific directories from the host into the
6033 * nspawn will now mount its own devpts file system instance
6034 into the container, in order not to leak pty devices from
6035 the host into the container.
6037 * systemd will now read the firmware boot time performance
6038 information from the EFI variables, if the used boot loader
6039 supports this, and takes it into account for boot performance
6040 analysis via "systemd-analyze". This is currently supported
6041 only in conjunction with Gummiboot, but could be supported
6042 by other boot loaders too. For details see:
6044 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
6046 * A new generator has been added that automatically mounts the
6047 EFI System Partition (ESP) to /boot, if that directory
6048 exists, is empty, and no other file system has been
6049 configured to be mounted there.
6051 * logind will now send out PrepareForSleep(false) out
6052 unconditionally, after coming back from suspend. This may be
6053 used by applications as asynchronous notification for
6054 system resume events.
6056 * "systemctl unlock-sessions" has been added, that allows
6057 unlocking the screens of all user sessions at once, similar
6058 to how "systemctl lock-sessions" already locked all users
6059 sessions. This is backed by a new D-Bus call UnlockSessions().
6061 * "loginctl seat-status" will now show the master device of a
6062 seat. (i.e. the device of a seat that needs to be around for
6063 the seat to be considered available, usually the graphics
6066 * tmpfiles gained a new "X" line type, that allows
6067 configuration of files and directories (with wildcards) that
6068 shall be excluded from automatic cleanup ("aging").
6070 * udev default rules set the device node permissions now only
6071 at "add" events, and do not change them any longer with a
6072 later "change" event.
6074 * The log messages for lid events and power/sleep keypresses
6075 now carry a message ID.
6077 * We now have a substantially larger unit test suite, but this
6078 continues to be work in progress.
6080 * udevadm hwdb gained a new --root= parameter to change the
6081 root directory to operate relative to.
6083 * logind will now issue a background sync() request to the kernel
6084 early at shutdown, so that dirty buffers are flushed to disk early
6085 instead of at the last moment, in order to optimize shutdown
6088 * A new bootctl tool has been added that is an interface for
6089 certain boot loader operations. This is currently a preview
6090 and is likely to be extended into a small mechanism daemon
6091 like timedated, localed, hostnamed, and can be used by
6092 graphical UIs to enumerate available boot options, and
6093 request boot into firmware operations.
6095 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
6096 the rest of the package. It also has been updated to work
6097 correctly in initrds.
6099 * Policykit previously has been runtime optional, and is now
6100 also compile time optional via a configure switch.
6102 * systemd-analyze has been reimplemented in C. Also "systemctl
6103 dot" has moved into systemd-analyze.
6105 * "systemctl status" with no further parameters will now print
6106 the status of all active or failed units.
6108 * Operations such as "systemctl start" can now be executed
6109 with a new mode "--irreversible" which may be used to queue
6110 operations that cannot accidentally be reversed by a later
6111 job queuing. This is by default used to make shutdown
6112 requests more robust.
6114 * The Python API of systemd now gained a new module for
6115 reading journal files.
6117 * A new tool kernel-install has been added that can install
6118 kernel images according to the Boot Loader Specification:
6120 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
6122 * Boot time console output has been improved to provide
6123 animated boot time output for hanging jobs.
6125 * A new tool systemd-activate has been added which can be used
6126 to test socket activation with, directly from the command
6127 line. This should make it much easier to test and debug
6128 socket activation in daemons.
6130 * journalctl gained a new "--reverse" (or -r) option to show
6131 journal output in reverse order (i.e. newest line first).
6133 * journalctl gained a new "--pager-end" (or -e) option to jump
6134 to immediately jump to the end of the journal in the
6135 pager. This is only supported in conjunction with "less".
6137 * journalctl gained a new "--user-unit=" option, that works
6138 similarly to "--unit=" but filters for user units rather than
6141 * A number of unit files to ease adoption of systemd in
6142 initrds has been added. This moves some minimal logic from
6143 the various initrd implementations into systemd proper.
6145 * The journal files are now owned by a new group
6146 "systemd-journal", which exists specifically to allow access
6147 to the journal, and nothing else. Previously, we used the
6148 "adm" group for that, which however possibly covers more
6149 than just journal/log file access. This new group is now
6150 already used by systemd-journal-gatewayd to ensure this
6151 daemon gets access to the journal files and as little else
6152 as possible. Note that "make install" will also set FS ACLs
6153 up for /var/log/journal to give "adm" and "wheel" read
6154 access to it, in addition to "systemd-journal" which owns
6155 the journal files. We recommend that packaging scripts also
6156 add read access to "adm" + "wheel" to /var/log/journal, and
6157 all existing/future journal files. To normal users and
6158 administrators little changes, however packagers need to
6159 ensure to create the "systemd-journal" system group at
6160 package installation time.
6162 * The systemd-journal-gatewayd now runs as unprivileged user
6163 systemd-journal-gateway:systemd-journal-gateway. Packaging
6164 scripts need to create these system user/group at
6167 * timedated now exposes a new boolean property CanNTP that
6168 indicates whether a local NTP service is available or not.
6170 * systemd-detect-virt will now also detect xen PVs
6172 * The pstore file system is now mounted by default, if it is
6175 * In addition to the SELinux and IMA policies we will now also
6176 load SMACK policies at early boot.
6178 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
6179 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
6180 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
6181 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
6182 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
6183 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
6184 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
6185 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
6186 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
6187 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
6188 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
6189 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
6190 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
6191 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
6195 * Timer units now support calendar time events in addition to
6196 monotonic time events. That means you can now trigger a unit
6197 based on a calendar time specification such as "Thu,Fri
6198 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
6199 or fifth day of any month of the year 2013, given that it is
6200 a thursday or friday. This brings timer event support
6201 considerably closer to cron's capabilities. For details on
6202 the supported calendar time specification language see
6205 * udev now supports a number of different naming policies for
6206 network interfaces for predictable names, and a combination
6207 of these policies is now the default. Please see this wiki
6208 document for details:
6210 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
6212 * Auke Kok's bootchart implementation has been added to the
6213 systemd tree. It is an optional component that can graph the
6214 boot in quite some detail. It is one of the best bootchart
6215 implementations around and minimal in its code and
6218 * nss-myhostname has been integrated into the systemd source
6219 tree. nss-myhostname guarantees that the local hostname
6220 always stays resolvable via NSS. It has been a weak
6221 requirement of systemd-hostnamed since a long time, and
6222 since its code is actually trivial we decided to just
6223 include it in systemd's source tree. It can be turned off
6224 with a configure switch.
6226 * The read-ahead logic is now capable of properly detecting
6227 whether a btrfs file system is on SSD or rotating media, in
6228 order to optimize the read-ahead scheme. Previously, it was
6229 only capable of detecting this on traditional file systems
6232 * In udev, additional device properties are now read from the
6233 IAB in addition to the OUI database. Also, Bluetooth company
6234 identities are attached to the devices as well.
6236 * In service files %U may be used as specifier that is
6237 replaced by the configured user name of the service.
6239 * nspawn may now be invoked without a controlling TTY. This
6240 makes it suitable for invocation as its own service. This
6241 may be used to set up a simple containerized server system
6242 using only core OS tools.
6244 * systemd and nspawn can now accept socket file descriptors
6245 when they are started for socket activation. This enables
6246 implementation of socket activated nspawn
6247 containers. i.e. think about autospawning an entire OS image
6248 when the first SSH or HTTP connection is received. We expect
6249 that similar functionality will also be added to libvirt-lxc
6252 * journalctl will now suppress ANSI color codes when
6253 presenting log data.
6255 * systemctl will no longer show control group information for
6256 a unit if the control group is empty anyway.
6258 * logind can now automatically suspend/hibernate/shutdown the
6261 * /etc/machine-info and hostnamed now also expose the chassis
6262 type of the system. This can be used to determine whether
6263 the local system is a laptop, desktop, handset or
6264 tablet. This information may either be configured by the
6265 user/vendor or is automatically determined from ACPI and DMI
6266 information if possible.
6268 * A number of PolicyKit actions are now bound together with
6269 "imply" rules. This should simplify creating UIs because
6270 many actions will now authenticate similar ones as well.
6272 * Unit files learnt a new condition ConditionACPower= which
6273 may be used to conditionalize a unit depending on whether an
6274 AC power source is connected or not, of whether the system
6275 is running on battery power.
6277 * systemctl gained a new "is-failed" verb that may be used in
6278 shell scripts and suchlike to check whether a specific unit
6279 is in the "failed" state.
6281 * The EnvironmentFile= setting in unit files now supports file
6282 globbing, and can hence be used to easily read a number of
6283 environment files at once.
6285 * systemd will no longer detect and recognize specific
6286 distributions. All distribution-specific #ifdeffery has been
6287 removed, systemd is now fully generic and
6288 distribution-agnostic. Effectively, not too much is lost as
6289 a lot of the code is still accessible via explicit configure
6290 switches. However, support for some distribution specific
6291 legacy configuration file formats has been dropped. We
6292 recommend distributions to simply adopt the configuration
6293 files everybody else uses now and convert the old
6294 configuration from packaging scripts. Most distributions
6295 already did that. If that's not possible or desirable,
6296 distributions are welcome to forward port the specific
6297 pieces of code locally from the git history.
6299 * When logging a message about a unit systemd will now always
6300 log the unit name in the message meta data.
6302 * localectl will now also discover system locale data that is
6303 not stored in locale archives, but directly unpacked.
6305 * logind will no longer unconditionally use framebuffer
6306 devices as seat masters, i.e. as devices that are required
6307 to be existing before a seat is considered preset. Instead,
6308 it will now look for all devices that are tagged as
6309 "seat-master" in udev. By default, framebuffer devices will
6310 be marked as such, but depending on local systems, other
6311 devices might be marked as well. This may be used to
6312 integrate graphics cards using closed source drivers (such
6313 as NVidia ones) more nicely into logind. Note however, that
6314 we recommend using the open source NVidia drivers instead,
6315 and no udev rules for the closed-source drivers will be
6316 shipped from us upstream.
6318 Contributions from: Adam Williamson, Alessandro Crismani, Auke
6319 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
6320 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
6321 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
6322 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
6323 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
6324 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
6325 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
6326 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
6327 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
6328 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
6329 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
6334 * udev gained support for loading additional device properties
6335 from an indexed database that is keyed by vendor/product IDs
6336 and similar device identifiers. For the beginning this
6337 "hwdb" is populated with data from the well-known PCI and
6338 USB database, but also includes PNP, ACPI and OID data. In
6339 the longer run this indexed database shall grow into
6340 becoming the one central database for non-essential
6341 userspace device metadata. Previously, data from the PCI/USB
6342 database was only attached to select devices, since the
6343 lookup was a relatively expensive operation due to O(n) time
6344 complexity (with n being the number of entries in the
6345 database). Since this is now O(1), we decided to add in this
6346 data for all devices where this is available, by
6347 default. Note that the indexed database needs to be rebuilt
6348 when new data files are installed. To achieve this you need
6349 to update your packaging scripts to invoke "udevadm hwdb
6350 --update" after installation of hwdb data files. For
6351 RPM-based distributions we introduced the new
6352 %udev_hwdb_update macro for this purpose.
6354 * The Journal gained support for the "Message Catalog", an
6355 indexed database to link up additional information with
6356 journal entries. For further details please check:
6358 https://www.freedesktop.org/wiki/Software/systemd/catalog
6360 The indexed message catalog database also needs to be
6361 rebuilt after installation of message catalog files. Use
6362 "journalctl --update-catalog" for this. For RPM-based
6363 distributions we introduced the %journal_catalog_update
6364 macro for this purpose.
6366 * The Python Journal bindings gained support for the standard
6367 Python logging framework.
6369 * The Journal API gained new functions for checking whether
6370 the underlying file system of a journal file is capable of
6371 properly reporting file change notifications, or whether
6372 applications that want to reflect journal changes "live"
6373 need to recheck journal files continuously in appropriate
6376 * It is now possible to set the "age" field for tmpfiles
6377 entries to 0, indicating that files matching this entry
6378 shall always be removed when the directories are cleaned up.
6380 * coredumpctl gained a new "gdb" verb which invokes gdb
6381 right-away on the selected coredump.
6383 * There's now support for "hybrid sleep" on kernels that
6384 support this, in addition to "suspend" and "hibernate". Use
6385 "systemctl hybrid-sleep" to make use of this.
6387 * logind's HandleSuspendKey= setting (and related settings)
6388 now gained support for a new "lock" setting to simply
6389 request the screen lock on all local sessions, instead of
6390 actually executing a suspend or hibernation.
6392 * systemd will now mount the EFI variables file system by
6395 * Socket units now gained support for configuration of the
6396 SMACK security label.
6398 * timedatectl will now output the time of the last and next
6399 daylight saving change.
6401 * We dropped support for various legacy and distro-specific
6402 concepts, such as insserv, early-boot SysV services
6403 (i.e. those for non-standard runlevels such as 'b' or 'S')
6404 or ArchLinux /etc/rc.conf support. We recommend the
6405 distributions who still need support this to either continue
6406 to maintain the necessary patches downstream, or find a
6407 different solution. (Talk to us if you have questions!)
6409 * Various systemd components will now bypass PolicyKit checks
6410 for root and otherwise handle properly if PolicyKit is not
6411 found to be around. This should fix most issues for
6412 PolicyKit-less systems. Quite frankly this should have been
6413 this way since day one. It is absolutely our intention to
6414 make systemd work fine on PolicyKit-less systems, and we
6415 consider it a bug if something does not work as it should if
6416 PolicyKit is not around.
6418 * For embedded systems it is now possible to build udev and
6419 systemd without blkid and/or kmod support.
6421 * "systemctl switch-root" is now capable of switching root
6422 more than once. I.e. in addition to transitions from the
6423 initrd to the host OS it is now possible to transition to
6424 further OS images from the host. This is useful to implement
6425 offline updating tools.
6427 * Various other additions have been made to the RPM macros
6428 shipped with systemd. Use %udev_rules_update() after
6429 installing new udev rules files. %_udevhwdbdir,
6430 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
6431 %_sysctldir are now available which resolve to the right
6432 directories for packages to place various data files in.
6434 * journalctl gained the new --full switch (in addition to
6435 --all, to disable ellipsation for long messages.
6437 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
6438 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
6439 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
6440 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
6441 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
6442 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
6443 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
6444 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
6445 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
6449 * journalctl gained new --since= and --until= switches to
6450 filter by time. It also now supports nice filtering for
6451 units via --unit=/-u.
6453 * Type=oneshot services may use ExecReload= and do the
6456 * The journal daemon now supports time-based rotation and
6457 vacuuming, in addition to the usual disk-space based
6460 * The journal will now index the available field values for
6461 each field name. This enables clients to show pretty drop
6462 downs of available match values when filtering. The bash
6463 completion of journalctl has been updated
6464 accordingly. journalctl gained a new switch -F to list all
6465 values a certain field takes in the journal database.
6467 * More service events are now written as structured messages
6468 to the journal, and made recognizable via message IDs.
6470 * The timedated, localed and hostnamed mini-services which
6471 previously only provided support for changing time, locale
6472 and hostname settings from graphical DEs such as GNOME now
6473 also have a minimal (but very useful) text-based client
6474 utility each. This is probably the nicest way to changing
6475 these settings from the command line now, especially since
6476 it lists available options and is fully integrated with bash
6479 * There's now a new tool "systemd-coredumpctl" to list and
6480 extract coredumps from the journal.
6482 * We now install a README each in /var/log/ and
6483 /etc/rc.d/init.d explaining where the system logs and init
6484 scripts went. This hopefully should help folks who go to
6485 that dirs and look into the otherwise now empty void and
6486 scratch their heads.
6488 * When user-services are invoked (by systemd --user) the
6489 $MANAGERPID env var is set to the PID of systemd.
6491 * SIGRTMIN+24 when sent to a --user instance will now result
6492 in immediate termination of systemd.
6494 * gatewayd received numerous feature additions such as a
6495 "follow" mode, for live syncing and filtering.
6497 * browse.html now allows filtering and showing detailed
6498 information on specific entries. Keyboard navigation and
6499 mouse screen support has been added.
6501 * gatewayd/journalctl now supports HTML5/JSON
6502 Server-Sent-Events as output.
6504 * The SysV init script compatibility logic will now
6505 heuristically determine whether a script supports the
6506 "reload" verb, and only then make this available as
6509 * "systemctl status --follow" has been removed, use "journalctl
6512 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
6513 have been removed since they are hardly useful to be
6516 * And I'd like to take the opportunity to specifically mention
6517 Zbigniew for his great contributions. Zbigniew, you rock!
6519 Contributions from: Andrew Eikum, Christian Hesse, Colin
6520 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
6521 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
6522 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
6523 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
6524 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
6525 Jędrzejewski-Szmek, Сковорода Никита Андреевич
6529 * If /etc/vconsole.conf is non-existent or empty we will no
6530 longer load any console font or key map at boot by
6531 default. Instead the kernel defaults will be left
6532 intact. This is definitely the right thing to do, as no
6533 configuration should mean no configuration, and hard-coding
6534 font names that are different on all archs is probably a bad
6535 idea. Also, the kernel default key map and font should be
6536 good enough for most cases anyway, and mostly identical to
6537 the userspace fonts/key maps we previously overloaded them
6538 with. If distributions want to continue to default to a
6539 non-kernel font or key map they should ship a default
6540 /etc/vconsole.conf with the appropriate contents.
6542 Contributions from: Colin Walters, Daniel J Walsh, Dave
6543 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
6544 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6548 * journalctl gained a new --cursor= switch to show entries
6549 starting from the specified location in the journal.
6551 * We now enforce a size limit on journal entry fields exported
6552 with "-o json" in journalctl. Fields larger than 4K will be
6553 assigned null. This can be turned off with --all.
6555 * An (optional) journal gateway daemon is now available as
6556 "systemd-journal-gatewayd.service". This service provides
6557 access to the journal via HTTP and JSON. This functionality
6558 will be used to implement live log synchronization in both
6559 pull and push modes, but has various other users too, such
6560 as easy log access for debugging of embedded devices. Right
6561 now it is already useful to retrieve the journal via HTTP:
6563 # systemctl start systemd-journal-gatewayd.service
6564 # wget http://localhost:19531/entries
6566 This will download the journal contents in a
6567 /var/log/messages compatible format. The same as JSON:
6569 # curl -H"Accept: application/json" http://localhost:19531/entries
6571 This service is also accessible via a web browser where a
6572 single static HTML5 app is served that uses the JSON logic
6573 to enable the user to do some basic browsing of the
6574 journal. This will be extended later on. Here's an example
6575 screenshot of this app in its current state:
6577 http://0pointer.de/public/journal-gatewayd
6579 Contributions from: Kay Sievers, Lennart Poettering, Robert
6580 Milasan, Tom Gundersen
6584 * The bash completion logic is now available for journalctl
6587 * We do not mount the "cpuset" controller anymore together with
6588 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
6589 started if no parameters are assigned to it. "cpuset" hence
6590 broke code that assumed it could create "cpu" groups and
6593 * journalctl -f will now subscribe to terminal size changes,
6594 and line break accordingly.
6596 Contributions from: Dave Reisner, Kay Sievers, Lennart
6597 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
6601 * nspawn will now create a symlink /etc/localtime in the
6602 container environment, copying the host's timezone
6603 setting. Previously this has been done via a bind mount, but
6604 since symlinks cannot be bind mounted this has now been
6605 changed to create/update the appropriate symlink.
6607 * journalctl -n's line number argument is now optional, and
6608 will default to 10 if omitted.
6610 * journald will now log the maximum size the journal files may
6611 take up on disk. This is particularly useful if the default
6612 built-in logic of determining this parameter from the file
6613 system size is used. Use "systemctl status
6614 systemd-journald.service" to see this information.
6616 * The multi-seat X wrapper tool has been stripped down. As X
6617 is now capable of enumerating graphics devices via udev in a
6618 seat-aware way the wrapper is not strictly necessary
6619 anymore. A stripped down temporary stop-gap is still shipped
6620 until the upstream display managers have been updated to
6621 fully support the new X logic. Expect this wrapper to be
6622 removed entirely in one of the next releases.
6624 * HandleSleepKey= in logind.conf has been split up into
6625 HandleSuspendKey= and HandleHibernateKey=. The old setting
6626 is not available anymore. X11 and the kernel are
6627 distinguishing between these keys and we should too. This
6628 also means the inhibition lock for these keys has been split
6631 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
6632 Poettering, Lukas Nykryn, Václav Pavlín
6636 * Whenever a unit changes state we will now log this to the
6637 journal and show along the unit's own log output in
6640 * ConditionPathIsMountPoint= can now properly detect bind
6641 mount points too. (Previously, a bind mount of one file
6642 system to another place in the same file system could not be
6643 detected as mount, since they shared struct stat's st_dev
6646 * We will now mount the cgroup controllers cpu, cpuacct,
6647 cpuset and the controllers net_cls, net_prio together by
6650 * nspawn containers will now have a virtualized boot
6651 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
6652 over with a randomized ID at container initialization). This
6653 has the effect of making "journalctl -b" do the right thing
6656 * The JSON output journal serialization has been updated not
6657 to generate "endless" list objects anymore, but rather one
6658 JSON object per line. This is more in line how most JSON
6659 parsers expect JSON objects. The new output mode
6660 "json-pretty" has been added to provide similar output, but
6661 neatly aligned for readability by humans.
6663 * We dropped all explicit sync() invocations in the shutdown
6664 code. The kernel does this implicitly anyway in the kernel
6665 reboot() syscall. halt(8)'s -n option is now a compatibility
6668 * We now support virtualized reboot() in containers, as
6669 supported by newer kernels. We will fall back to exit() if
6670 CAP_SYS_REBOOT is not available to the container. Also,
6671 nspawn makes use of this now and will actually reboot the
6672 container if the containerized OS asks for that.
6674 * journalctl will only show local log output by default
6675 now. Use --merge (-m) to show remote log output, too.
6677 * libsystemd-journal gained the new sd_journal_get_usage()
6678 call to determine the current disk usage of all journal
6679 files. This is exposed in the new "journalctl --disk-usage"
6682 * journald gained a new configuration setting SplitMode= in
6683 journald.conf which may be used to control how user journals
6684 are split off. See journald.conf(5) for details.
6686 * A new condition type ConditionFileNotEmpty= has been added.
6688 * tmpfiles' "w" lines now support file globbing, to write
6689 multiple files at once.
6691 * We added Python bindings for the journal submission
6692 APIs. More Python APIs for a number of selected APIs will
6693 likely follow. Note that we intend to add native bindings
6694 only for the Python language, as we consider it common
6695 enough to deserve bindings shipped within systemd. There are
6696 various projects outside of systemd that provide bindings
6697 for languages such as PHP or Lua.
6699 * Many conditions will now resolve specifiers such as %i. In
6700 addition, PathChanged= and related directives of .path units
6701 now support specifiers as well.
6703 * There's now a new RPM macro definition for the system preset
6706 * journald will now warn if it ca not forward a message to the
6707 syslog daemon because its socket is full.
6709 * timedated will no longer write or process /etc/timezone,
6710 except on Debian. As we do not support late mounted /usr
6711 anymore /etc/localtime always being a symlink is now safe,
6712 and hence the information in /etc/timezone is not necessary
6715 * logind will now always reserve one VT for a text getty (VT6
6716 by default). Previously if more than 6 X sessions where
6717 started they took up all the VTs with auto-spawned gettys,
6718 so that no text gettys were available anymore.
6720 * udev will now automatically inform the btrfs kernel logic
6721 about btrfs RAID components showing up. This should make
6722 simple hotplug based btrfs RAID assembly work.
6724 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
6725 (but not for its children which will stay at the kernel
6726 default). This should allow setups with a lot more listening
6729 * systemd will now always pass the configured timezone to the
6730 kernel at boot. timedated will do the same when the timezone
6733 * logind's inhibition logic has been updated. By default,
6734 logind will now handle the lid switch, the power and sleep
6735 keys all the time, even in graphical sessions. If DEs want
6736 to handle these events on their own they should take the new
6737 handle-power-key, handle-sleep-key and handle-lid-switch
6738 inhibitors during their runtime. A simple way to achieve
6739 that is to invoke the DE wrapped in an invocation of:
6741 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
6743 * Access to unit operations is now checked via SELinux taking
6744 the unit file label and client process label into account.
6746 * systemd will now notify the administrator in the journal
6747 when he over-mounts a non-empty directory.
6749 * There are new specifiers that are resolved in unit files,
6750 for the host name (%H), the machine ID (%m) and the boot ID
6753 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
6754 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
6755 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
6756 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
6757 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
6758 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
6759 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
6763 * Support for reading structured kernel messages from
6764 /dev/kmsg has now been added and is enabled by default.
6766 * Support for reading kernel messages from /proc/kmsg has now
6767 been removed. If you want kernel messages in the journal
6768 make sure to run a recent kernel (>= 3.5) that supports
6769 reading structured messages from /dev/kmsg (see
6770 above). /proc/kmsg is now exclusive property of classic
6771 syslog daemons again.
6773 * The libudev API gained the new
6774 udev_device_new_from_device_id() call.
6776 * The logic for file system namespace (ReadOnlyDirectory=,
6777 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
6778 require pivot_root() anymore. This means fewer temporary
6779 directories are created below /tmp for this feature.
6781 * nspawn containers will now see and receive all submounts
6782 made on the host OS below the root file system of the
6785 * Forward Secure Sealing is now supported for Journal files,
6786 which provide cryptographical sealing of journal files so
6787 that attackers cannot alter log history anymore without this
6788 being detectable. Lennart will soon post a blog story about
6789 this explaining it in more detail.
6791 * There are two new service settings RestartPreventExitStatus=
6792 and SuccessExitStatus= which allow configuration of exit
6793 status (exit code or signal) which will be excepted from the
6794 restart logic, resp. consider successful.
6796 * journalctl gained the new --verify switch that can be used
6797 to check the integrity of the structure of journal files and
6798 (if Forward Secure Sealing is enabled) the contents of
6801 * nspawn containers will now be run with /dev/stdin, /dev/fd/
6802 and similar symlinks pre-created. This makes running shells
6803 as container init process a lot more fun.
6805 * The fstab support can now handle PARTUUID= and PARTLABEL=
6808 * A new ConditionHost= condition has been added to match
6809 against the hostname (with globs) and machine ID. This is
6810 useful for clusters where a single OS image is used to
6811 provision a large number of hosts which shall run slightly
6812 different sets of services.
6814 * Services which hit the restart limit will now be placed in a
6817 Contributions from: Bertram Poettering, Dave Reisner, Huang
6818 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
6819 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
6823 * When running in --user mode systemd will now become a
6824 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
6825 tree a lot more organized.
6827 * A new PartOf= unit dependency type has been introduced that
6828 may be used to group services in a natural way.
6830 * "systemctl enable" may now be used to enable instances of
6833 * journalctl now prints error log levels in red, and
6834 warning/notice log levels in bright white. It also supports
6835 filtering by log level now.
6837 * cgtop gained a new -n switch (similar to top), to configure
6838 the maximum number of iterations to run for. It also gained
6839 -b, to run in batch mode (accepting no input).
6841 * The suffix ".service" may now be omitted on most systemctl
6842 command lines involving service unit names.
6844 * There's a new bus call in logind to lock all sessions, as
6845 well as a loginctl verb for it "lock-sessions".
6847 * libsystemd-logind.so gained a new call sd_journal_perror()
6848 that works similar to libc perror() but logs to the journal
6849 and encodes structured information about the error number.
6851 * /etc/crypttab entries now understand the new keyfile-size=
6854 * shutdown(8) now can send a (configurable) wall message when
6855 a shutdown is cancelled.
6857 * The mount propagation mode for the root file system will now
6858 default to "shared", which is useful to make containers work
6859 nicely out-of-the-box so that they receive new mounts from
6860 the host. This can be undone locally by running "mount
6861 --make-rprivate /" if needed.
6863 * The prefdm.service file has been removed. Distributions
6864 should maintain this unit downstream if they intend to keep
6865 it around. However, we recommend writing normal unit files
6866 for display managers instead.
6868 * Since systemd is a crucial part of the OS we will now
6869 default to a number of compiler switches that improve
6870 security (hardening) such as read-only relocations, stack
6871 protection, and suchlike.
6873 * The TimeoutSec= setting for services is now split into
6874 TimeoutStartSec= and TimeoutStopSec= to allow configuration
6875 of individual time outs for the start and the stop phase of
6878 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
6879 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
6880 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
6881 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
6882 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
6883 Gundersen, Zbigniew Jędrzejewski-Szmek
6887 * The journal and id128 C APIs are now fully documented as man
6890 * Extra safety checks have been added when transitioning from
6891 the initial RAM disk to the main system to avoid accidental
6894 * /etc/crypttab entries now understand the new keyfile-offset=
6897 * systemctl -t can now be used to filter by unit load state.
6899 * The journal C API gained the new sd_journal_wait() call to
6900 make writing synchronous journal clients easier.
6902 * journalctl gained the new -D switch to show journals from a
6905 * journalctl now displays a special marker between log
6906 messages of two different boots.
6908 * The journal is now explicitly flushed to /var via a service
6909 systemd-journal-flush.service, rather than implicitly simply
6910 by seeing /var/log/journal to be writable.
6912 * journalctl (and the journal C APIs) can now match for much
6913 more complex expressions, with alternatives and
6916 * When transitioning from the initial RAM disk to the main
6917 system we will now kill all processes in a killing spree to
6918 ensure no processes stay around by accident.
6920 * Three new specifiers may be used in unit files: %u, %h, %s
6921 resolve to the user name, user home directory resp. user
6922 shell. This is useful for running systemd user instances.
6924 * We now automatically rotate journal files if their data
6925 object hash table gets a fill level > 75%. We also size the
6926 hash table based on the configured maximum file size. This
6927 together should lower hash collisions drastically and thus
6928 speed things up a bit.
6930 * journalctl gained the new "--header" switch to introspect
6931 header data of journal files.
6933 * A new setting SystemCallFilters= has been added to services
6934 which may be used to apply blacklists or whitelists to
6935 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
6937 * nspawn gained a new --link-journal= switch (and quicker: -j)
6938 to link the container journal with the host. This makes it
6939 very easy to centralize log viewing on the host for all
6940 guests while still keeping the journal files separated.
6942 * Many bugfixes and optimizations
6944 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
6945 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
6946 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
6951 * Several tools now understand kernel command line arguments,
6952 which are only read when run in an initial RAM disk. They
6953 usually follow closely their normal counterparts, but are
6956 * There's a new tool to analyze the readahead files that are
6957 automatically generated at boot. Use:
6959 /usr/lib/systemd/systemd-readahead analyze /.readahead
6961 * We now provide an early debug shell on tty9 if this enabled. Use:
6963 systemctl enable debug-shell.service
6965 * All plymouth related units have been moved into the Plymouth
6966 package. Please make sure to upgrade your Plymouth version
6969 * systemd-tmpfiles now supports getting passed the basename of
6970 a configuration file only, in which case it will look for it
6971 in all appropriate directories automatically.
6973 * udevadm info now takes a /dev or /sys path as argument, and
6974 does the right thing. Example:
6976 udevadm info /dev/sda
6977 udevadm info /sys/class/block/sda
6979 * systemctl now prints a warning if a unit is stopped but a
6980 unit that might trigger it continues to run. Example: a
6981 service is stopped but the socket that activates it is left
6984 * "systemctl status" will now mention if the log output was
6985 shortened due to rotation since a service has been started.
6987 * The journal API now exposes functions to determine the
6988 "cutoff" times due to rotation.
6990 * journald now understands SIGUSR1 and SIGUSR2 for triggering
6991 immediately flushing of runtime logs to /var if possible,
6992 resp. for triggering immediate rotation of the journal
6995 * It is now considered an error if a service is attempted to
6996 be stopped that is not loaded.
6998 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
7000 * systemd-analyze now supports Python 3
7002 * tmpfiles now supports cleaning up directories via aging
7003 where the first level dirs are always kept around but
7004 directories beneath it automatically aged. This is enabled
7005 by prefixing the age field with '~'.
7007 * Seat objects now expose CanGraphical, CanTTY properties
7008 which is required to deal with very fast bootups where the
7009 display manager might be running before the graphics drivers
7010 completed initialization.
7012 * Seat objects now expose a State property.
7014 * We now include RPM macros for service enabling/disabling
7015 based on the preset logic. We recommend RPM based
7016 distributions to make use of these macros if possible. This
7017 makes it simpler to reuse RPM spec files across
7020 * We now make sure that the collected systemd unit name is
7021 always valid when services log to the journal via
7024 * There's a new man page kernel-command-line(7) detailing all
7025 command line options we understand.
7027 * The fstab generator may now be disabled at boot by passing
7028 fstab=0 on the kernel command line.
7030 * A new kernel command line option modules-load= is now understood
7031 to load a specific kernel module statically, early at boot.
7033 * Unit names specified on the systemctl command line are now
7034 automatically escaped as needed. Also, if file system or
7035 device paths are specified they are automatically turned
7036 into the appropriate mount or device unit names. Example:
7038 systemctl status /home
7039 systemctl status /dev/sda
7041 * The SysVConsole= configuration option has been removed from
7042 system.conf parsing.
7044 * The SysV search path is no longer exported on the D-Bus
7047 * The Names= option has been removed from unit file parsing.
7049 * There's a new man page bootup(7) detailing the boot process.
7051 * Every unit and every generator we ship with systemd now
7052 comes with full documentation. The self-explanatory boot is
7055 * A couple of services gained "systemd-" prefixes in their
7056 name if they wrap systemd code, rather than only external
7057 code. Among them fsck@.service which is now
7058 systemd-fsck@.service.
7060 * The HaveWatchdog property has been removed from the D-Bus
7063 * systemd.confirm_spawn= on the kernel command line should now
7066 * There's a new man page crypttab(5) which details all options
7067 we actually understand.
7069 * systemd-nspawn gained a new --capability= switch to pass
7070 additional capabilities to the container.
7072 * timedated will now read known NTP implementation unit names
7073 from /usr/lib/systemd/ntp-units.d/*.list,
7074 systemd-timedated-ntp.target has been removed.
7076 * journalctl gained a new switch "-b" that lists log data of
7077 the current boot only.
7079 * The notify socket is in the abstract namespace again, in
7080 order to support daemons which chroot() at start-up.
7082 * There is a new Storage= configuration option for journald
7083 which allows configuration of where log data should go. This
7084 also provides a way to disable journal logging entirely, so
7085 that data collected is only forwarded to the console, the
7086 kernel log buffer or another syslog implementation.
7088 * Many bugfixes and optimizations
7090 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
7091 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
7092 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
7093 Shawn Landden, Tom Gundersen
7097 * "systemctl help <unit>" now shows the man page if one is
7100 * Several new man pages have been added.
7102 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
7103 MaxLevelConsole= can now be specified in
7104 journald.conf. These options allow reducing the amount of
7105 data stored on disk or forwarded by the log level.
7107 * TimerSlackNSec= can now be specified in system.conf for
7108 PID1. This allows system-wide power savings.
7110 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
7111 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
7116 * logind is now capable of (optionally) handling power and
7117 sleep keys as well as the lid switch.
7119 * journalctl now understands the syntax "journalctl
7120 /usr/bin/avahi-daemon" to get all log output of a specific
7123 * CapabilityBoundingSet= in system.conf now also influences
7124 the capability bound set of usermode helpers of the kernel.
7126 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
7127 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
7128 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
7129 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
7133 * Note that we skipped 139 releases here in order to set the
7134 new version to something that is greater than both udev's
7135 and systemd's most recent version number.
7137 * udev: all udev sources are merged into the systemd source tree now.
7138 All future udev development will happen in the systemd tree. It
7139 is still fully supported to use the udev daemon and tools without
7140 systemd running, like in initramfs or other init systems. Building
7141 udev though, will require the *build* of the systemd tree, but
7142 udev can be properly *run* without systemd.
7144 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
7145 should be used to create dead device nodes as workarounds for broken
7148 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
7149 no longer supported. udev_monitor_new_from_netlink() needs to be
7150 used to subscribe to events.
7152 * udev: when udevd is started by systemd, processes which are left
7153 behind by forking them off of udev rules, are unconditionally cleaned
7154 up and killed now after the event handling has finished. Services or
7155 daemons must be started as systemd services. Services can be
7156 pulled-in by udev to get started, but they can no longer be directly
7157 forked by udev rules.
7159 * udev: the daemon binary is called systemd-udevd now and installed
7160 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
7161 to adapt to that, create symlink, or rename the binary after building
7164 * libudev no longer provides these symbols:
7165 udev_monitor_from_socket()
7166 udev_queue_get_failed_list_entry()
7167 udev_get_{dev,sys,run}_path()
7168 The versions number was bumped and symbol versioning introduced.
7170 * systemd-loginctl and systemd-journalctl have been renamed
7171 to loginctl and journalctl to match systemctl.
7173 * The config files: /etc/systemd/systemd-logind.conf and
7174 /etc/systemd/systemd-journald.conf have been renamed to
7175 logind.conf and journald.conf. Package updates should rename
7176 the files to the new names on upgrade.
7178 * For almost all files the license is now LGPL2.1+, changed
7179 from the previous GPL2.0+. Exceptions are some minor stuff
7180 of udev (which will be changed to LGPL2.1 eventually, too),
7181 and the MIT licensed sd-daemon.[ch] library that is suitable
7182 to be used as drop-in files.
7184 * systemd and logind now handle system sleep states, in
7185 particular suspending and hibernating.
7187 * logind now implements a sleep/shutdown/idle inhibiting logic
7188 suitable for a variety of uses. Soonishly Lennart will blog
7189 about this in more detail.
7191 * var-run.mount and var-lock.mount are no longer provided
7192 (which previously bind mounted these directories to their new
7193 places). Distributions which have not converted these
7194 directories to symlinks should consider stealing these files
7195 from git history and add them downstream.
7197 * We introduced the Documentation= field for units and added
7198 this to all our shipped units. This is useful to make it
7199 easier to explore the boot and the purpose of the various
7202 * All smaller setup units (such as
7203 systemd-vconsole-setup.service) now detect properly if they
7204 are run in a container and are skipped when
7205 appropriate. This guarantees an entirely noise-free boot in
7206 Linux container environments such as systemd-nspawn.
7208 * A framework for implementing offline system updates is now
7209 integrated, for details see:
7210 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
7212 * A new service type Type=idle is available now which helps us
7213 avoiding ugly interleaving of getty output and boot status
7216 * There's now a system-wide CapabilityBoundingSet= option to
7217 globally reduce the set of capabilities for the
7218 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
7219 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
7220 even CAP_NET_ADMIN system-wide for secure systems.
7222 * There are now system-wide DefaultLimitXXX= options to
7223 globally change the defaults of the various resource limits
7224 for all units started by PID 1.
7226 * Harald Hoyer's systemd test suite has been integrated into
7227 systemd which allows easy testing of systemd builds in qemu
7228 and nspawn. (This is really awesome! Ask us for details!)
7230 * The fstab parser is now implemented as generator, not inside
7233 * systemctl will now warn you if .mount units generated from
7234 /etc/fstab are out of date due to changes in fstab that
7235 have not been read by systemd yet.
7237 * systemd is now suitable for usage in initrds. Dracut has
7238 already been updated to make use of this. With this in place
7239 initrds get a slight bit faster but primarily are much
7240 easier to introspect and debug since "systemctl status" in
7241 the host system can be used to introspect initrd services,
7242 and the journal from the initrd is kept around too.
7244 * systemd-delta has been added, a tool to explore differences
7245 between user/admin configuration and vendor defaults.
7247 * PrivateTmp= now affects both /tmp and /var/tmp.
7249 * Boot time status messages are now much prettier and feature
7250 proper english language. Booting up systemd has never been
7253 * Read-ahead pack files now include the inode number of all
7254 files to pre-cache. When the inode changes the pre-caching
7255 is not attempted. This should be nicer to deal with updated
7256 packages which might result in changes of read-ahead
7259 * We now temporaritly lower the kernel's read_ahead_kb variable
7260 when collecting read-ahead data to ensure the kernel's
7261 built-in read-ahead does not add noise to our measurements
7262 of necessary blocks to pre-cache.
7264 * There's now RequiresMountsFor= to add automatic dependencies
7265 for all mounts necessary for a specific file system path.
7267 * MountAuto= and SwapAuto= have been removed from
7268 system.conf. Mounting file systems at boot has to take place
7271 * nspawn now learned a new switch --uuid= to set the machine
7272 ID on the command line.
7274 * nspawn now learned the -b switch to automatically search
7277 * vt102 is now the default TERM for serial TTYs, upgraded from
7280 * systemd-logind now works on VT-less systems.
7282 * The build tree has been reorganized. The individual
7283 components now have directories of their own.
7285 * A new condition type ConditionPathIsReadWrite= is now available.
7287 * nspawn learned the new -C switch to create cgroups for the
7288 container in other hierarchies.
7290 * We now have support for hardware watchdogs, configurable in
7293 * The scheduled shutdown logic now has a public API.
7295 * We now mount /tmp as tmpfs by default, but this can be
7296 masked and /etc/fstab can override it.
7298 * Since udisks does not make use of /media anymore we are not
7299 mounting a tmpfs on it anymore.
7301 * journalctl gained a new --local switch to only interleave
7302 locally generated journal files.
7304 * We can now load the IMA policy at boot automatically.
7306 * The GTK tools have been split off into a systemd-ui.
7308 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
7309 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
7310 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
7311 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
7312 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
7313 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
7314 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
7315 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
7316 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
7321 * This is mostly a bugfix release
7323 * Support optional initialization of the machine ID from the
7324 KVM or container configured UUID.
7326 * Support immediate reboots with "systemctl reboot -ff"
7328 * Show /etc/os-release data in systemd-analyze output
7330 * Many bugfixes for the journal, including endianness fixes and
7331 ensuring that disk space enforcement works
7333 * sd-login.h is C++ compatible again
7335 * Extend the /etc/os-release format on request of the Debian
7338 * We now refuse non-UTF8 strings used in various configuration
7339 and unit files. This is done to ensure we do not pass invalid
7340 data over D-Bus or expose it elsewhere.
7342 * Register Mimo USB Screens as suitable for automatic seat
7345 * Read SELinux client context from journal clients in a race
7348 * Reorder configuration file lookup order. /etc now always
7349 overrides /run in order to allow the administrator to always
7350 and unconditionally override vendor-supplied or
7351 automatically generated data.
7353 * The various user visible bits of the journal now have man
7354 pages. We still lack man pages for the journal API calls
7357 * We now ship all man pages in HTML format again in the
7360 Contributions from: Dave Reisner, Dirk Eibach, Frederic
7361 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
7362 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
7367 * This is mostly a bugfix release
7369 * systems lacking /etc/os-release are no longer supported.
7371 * Various functionality updates to libsystemd-login.so
7373 * Track class of PAM logins to distinguish greeters from
7376 Contributions from: Kay Sievers, Lennart Poettering, Michael
7381 * This is an important bugfix release for v41.
7383 * Building man pages is now optional which should be useful
7384 for those building systemd from git but unwilling to install
7387 * Watchdog support for supervising services is now usable. In
7388 a future release support for hardware watchdogs
7389 (i.e. /dev/watchdog) will be added building on this.
7391 * Service start rate limiting is now configurable and can be
7392 turned off per service. When a start rate limit is hit a
7393 reboot can automatically be triggered.
7395 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
7397 Contributions from: Benjamin Franzke, Bill Nottingham,
7398 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
7399 Schmidt, Michał Górny, Piotr Drąg
7403 * The systemd binary is installed /usr/lib/systemd/systemd now;
7404 An existing /sbin/init symlink needs to be adapted with the
7407 * The code that loads kernel modules has been ported to invoke
7408 libkmod directly, instead of modprobe. This means we do not
7409 support systems with module-init-tools anymore.
7411 * Watchdog support is now already useful, but still not
7414 * A new kernel command line option systemd.setenv= is
7415 understood to set system wide environment variables
7416 dynamically at boot.
7418 * We now limit the set of capabilities of systemd-journald.
7420 * We now set SIGPIPE to ignore by default, since it only is
7421 useful in shell pipelines, and has little use in general
7422 code. This can be disabled with IgnoreSIPIPE=no in unit
7425 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
7426 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
7431 * This is mostly a bugfix release
7433 * We now expose the reason why a service failed in the
7434 "Result" D-Bus property.
7436 * Rudimentary service watchdog support (will be completed over
7437 the next few releases.)
7439 * When systemd forks off in order execute some service we will
7440 now immediately changes its argv[0] to reflect which process
7441 it will execute. This is useful to minimize the time window
7442 with a generic argv[0], which makes bootcharts more useful
7444 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
7445 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
7446 Mike Kazantsev, Ray Strode
7450 * This is mostly a test release, but incorporates many
7453 * New systemd-cgtop tool to show control groups by their
7456 * Linking against libacl for ACLs is optional again. If
7457 disabled, support tracking device access for active logins
7458 goes becomes unavailable, and so does access to the user
7459 journals by the respective users.
7461 * If a group "adm" exists, journal files are automatically
7462 owned by them, thus allow members of this group full access
7463 to the system journal as well as all user journals.
7465 * The journal now stores the SELinux context of the logging
7466 client for all entries.
7468 * Add C++ inclusion guards to all public headers
7470 * New output mode "cat" in the journal to print only text
7471 messages, without any meta data like date or time.
7473 * Include tiny X server wrapper as a temporary stop-gap to
7474 teach XOrg udev display enumeration. This is used by display
7475 managers such as gdm, and will go away as soon as XOrg
7476 learned native udev hotplugging for display devices.
7478 * Add new systemd-cat tool for executing arbitrary programs
7479 with STDERR/STDOUT connected to the journal. Can also act as
7480 BSD logger replacement, and does so by default.
7482 * Optionally store all locally generated coredumps in the
7483 journal along with meta data.
7485 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
7486 writing short strings to files (for usage for /sys), and for
7487 creating symlinks, character and block device nodes.
7489 * New unit file option ControlGroupPersistent= to make cgroups
7490 persistent, following the mechanisms outlined in
7491 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
7493 * Support multiple local RTCs in a sane way
7495 * No longer monopolize IO when replaying readahead data on
7496 rotating disks, since we might starve non-file-system IO to
7497 death, since fanotify() will not see accesses done by blkid,
7500 * Do not show kernel threads in systemd-cgls anymore, unless
7501 requested with new -k switch.
7503 Contributions from: Dan Horák, Kay Sievers, Lennart
7504 Poettering, Michal Schmidt
7508 * This is mostly a test release, but incorporates many
7511 * The git repository moved to:
7512 git://anongit.freedesktop.org/systemd/systemd
7513 ssh://git.freedesktop.org/git/systemd/systemd
7515 * First release with the journal
7516 http://0pointer.de/blog/projects/the-journal.html
7518 * The journal replaces both systemd-kmsg-syslogd and
7519 systemd-stdout-bridge.
7521 * New sd_pid_get_unit() API call in libsystemd-logind
7523 * Many systemadm clean-ups
7525 * Introduce remote-fs-pre.target which is ordered before all
7526 remote mounts and may be used to start services before all
7529 * Added Mageia support
7531 * Add bash completion for systemd-loginctl
7533 * Actively monitor PID file creation for daemons which exit in
7534 the parent process before having finished writing the PID
7535 file in the daemon process. Daemons which do this need to be
7536 fixed (i.e. PID file creation must have finished before the
7537 parent exits), but we now react a bit more gracefully to them.
7539 * Add colourful boot output, mimicking the well-known output
7540 of existing distributions.
7542 * New option PassCredentials= for socket units, for
7543 compatibility with a recent kernel ABI breakage.
7545 * /etc/rc.local is now hooked in via a generator binary, and
7546 thus will no longer act as synchronization point during
7549 * systemctl list-unit-files now supports --root=.
7551 * systemd-tmpfiles now understands two new commands: z, Z for
7552 relabelling files according to the SELinux database. This is
7553 useful to apply SELinux labels to specific files in /sys,
7556 * Output of SysV services is now forwarded to both the console
7557 and the journal by default, not only just the console.
7559 * New man pages for all APIs from libsystemd-login.
7561 * The build tree got reorganized and the build system is a
7562 lot more modular allowing embedded setups to specifically
7563 select the components of systemd they are interested in.
7565 * Support for Linux systems lacking the kernel VT subsystem is
7568 * configure's --with-rootdir= got renamed to
7569 --with-rootprefix= to follow the naming used by udev and
7572 * Unless specified otherwise we will now install to /usr instead
7573 of /usr/local by default.
7575 * Processes with '@' in argv[0][0] are now excluded from the
7576 final shut-down killing spree, following the logic explained
7578 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
7580 * All processes remaining in a service cgroup when we enter
7581 the START or START_PRE states are now killed with
7582 SIGKILL. That means it is no longer possible to spawn
7583 background processes from ExecStart= lines (which was never
7584 supported anyway, and bad style).
7586 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
7587 reloading of units together.
7589 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
7590 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
7591 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
7592 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
7593 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek