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 * systemd-networkd .network DHCP setting UseMTU default has changed
20 from false to true. Meaning, DHCP server advertised MTU setting is
21 now applied by default. This resolves networking issues on low-mtu
24 * journald.conf gained a new boolean setting ReadKMsg= which defaults
25 to on. If turned off kernel log messages will not be read by
26 systemd-journald and not be included in the logs. It also gained a
27 new setting LineMax= for configuring the maximum line length to allow
28 when converting STDOUT/STDERR log streams into individual log
29 records. The new default for this value is 48K, up from the previous
32 * A new setting RuntimeDirectoryPreserve= for units has been added,
33 which allows more detailed control of what to do with a runtime
34 directory configured with RuntimeDirectory= (i.e. a directory below
35 /run or $XDG_RUNTIME_DIR) after a unit is stopped.
37 * The RuntimeDirectory= setting for units gained support for creating
38 deeper subdirectories below /run or $XDG_RUNTIME_DIR, instead of just
39 one top-level directory.
41 * Units gained new options StateDirectory=, CacheDirectory=,
42 LogsDirectory= and ConfigurationDirectory= which are closely related
43 to RuntimeDirectory= but manage per-service directories below
44 /var/lib, /var/cache, /var/log and /etc. By making use of this it is
45 possible to write unit files which when activated automatically gain
46 properly owned service specific directories in these locations, thus
47 making unit files self-contained and increasing compatibility with
48 stateless systems and factory reset where /etc or /var are
49 unpopulated at boot. Matching these new settings there's also
50 StateDirectoryMode=, CacheDirectoryMode=, LogsDirectoryMode=,
51 ConfigurationDirectoryMode= for configuring the access mode of these
54 * Automake support has been removed from this release. systemd is now
57 * systemd-journald will now aggressively cache client metadata during
58 runtime, speeding up log write performance under pressure. This comes
59 at a small price though: as much of the metadata is read
60 asynchronously from /proc/ (and isn't implicitly attached to log
61 datagrams by the kernel, like UID/GID/PID/SELinux are) this means the
62 metadata stored alongside a log entry might be slightly
63 out-of-date. Previously it could only be slightly newer than the log
64 message. The time window is small however, and given that the kernel
65 is unlikely to be improved anytime soon in this regard, this appears
68 * nss-myhostname/systemd-resolved will now by default synthesize an
69 A/AAAA resource record for the "_gateway" hostname, pointing to the
70 current default IP gateway. Previously it did that for the "gateway"
71 name, hampering adoption, as some distributions wanted to leave that
72 host name open for local use. The old behaviour may still be
73 requested at build time.
75 * systemd-networkd's [Address] section in .network files gained a new
76 Scope= setting for configuring the IP address scope. The [Network]
77 section gained a new boolean setting ConfigureWithoutCarrier= that
78 tells systemd-networkd to ignore link sensing when configuring the
79 device. The [DHCP] section gained a new Anonymize= boolean option for
80 turning on a number of options suggested in RFC 7844. A new
81 [RoutingPolicyRule] section has been added for configuring the IP
82 routing policy. The [Route] section has gained support for a new
83 Type= setting which permits configuring
84 blackhole/unreachable/prohibit routes.
86 * The [VRF] section in .netdev files gained a new Table= setting for
87 configuring the routing table to use. The [Tunnel] section gained a
88 new Independent= boolean field for configuring tunnels independent of
89 an underlying network interface. The [Bridge] section gained a new
90 GroupForwardMask= option for configuration of propagation of link
91 local frames between bridge ports.
93 * The WakeOnLan= setting in .link files gained support for a number of
94 new modes. A new TCP6SegmentationOffload= setting has been added for
95 configuring TCP/IPv6 hardware segmentation offload.
97 * The IPv6 RA sender implementation may now optionally send out RDNSS
98 and RDNSSL records for supplying DNS configuration to peers.
100 * systemd-nspawn gained support for a new --system-call-filter= command
101 line option for adding/removing entries in the default system call
102 filter it applies. Moreover systemd-nspawn has been changed to
103 implement a system call whitelist instead of a blacklist.
105 * systemd-run gained support for a new --pipe command line option. If
106 used the STDIN/STDOUT/STDERR file descriptors passed to systemd-run
107 are directly passed on to the activated transient service
108 binary. This allows invoking arbitrary processes as systemd services
109 (for example to take benefit of dependency management, accounting
110 management, resource management or log management that is done
111 automatically for services) — while still allowing them to be
112 integrated in a classic UNIX shell pipeline.
114 * When a service sends RELOAD=1 via sd_notify() and reload propagation
115 using ReloadPropagationTo= is configured, a reload is now propagated
116 to configured units. (Previously this was only done on explicitly
117 requested reloads, using "systemctl reload" or an equivalent
120 * For each service unit a restart counter is now kept: it is increased
121 each time the service is restarted due to Restart=, and may be
122 queried using "systemctl show -p NRestarts …".
124 * New system call filter groups @setuid, @memlock, @signal and
125 @timer have been added, for usage with SystemCallFilter=
126 in unit files and the new --system-call-filter= command line option
127 of systemd-nspawn (see above).
129 * ExecStart= lines in unit files gained two new modifiers: when a
130 command line is prefixed with "!" the command will be executed as
131 configured, except for the credentials applied by
132 setuid()/setgid()/setgroups(). It is very similar to the pre-existing
133 "+", but does still apply namespacing options unlike "+". There's
134 also "!!" now, which is mostly identical, but becomes a NOP on
135 systems that support ambient capabilities. This is useful to write
136 unit files that work with ambient capabilities where possible but
137 automatically fall back to traditional privilege dropping mechanisms
138 on systems where this is not supported.
140 * ListenNetlink= settings in socket units now support RDMA netlink
143 * A new unit file setting LockPersonality= has been added which permits
144 locking down the chosen execution domain ("personality") of a service
147 * A new special target "getty-pre.target" has been added, which is
148 ordered before all text logins, and may be used to order services
149 before, that shall run before these textual logins acquire access to
152 * systemd will now attempt to load the virtio-rng.ko kernel module very
153 early on if a VM environment supporting this is detected. This should
154 improve entropy during early boot in virtualized environments.
156 * A _netdev option is now supported in /etc/crypttab that operates in a
157 similar way as the same option in /etc/fstab: it permits configuring
158 encrypted devices that need to be ordered after the network coming
159 up. Following this logic, two new special targets
160 remote-cryptsetup-pre.target and remote-cryptsetup.target have been
161 added that are to cryptsetup.target what
162 remote-fs.target/remote-fs-pre.target are to local-fs.target.
164 * Service units gained a new UnsetEnvironment= setting which permits
165 unsetting specific environment variables for specific services that
166 are normally passed to it (for example in order to mask out locale
167 settings for specific services that can't deal with it).
169 * Units acquired a new boolean option IPAccounting=. When turned on, IP
170 traffic accounting (packet count as well as byte count) is done for
171 the service, and shown as part of "systemctl status" or "systemd-run
174 * Service units acquired two new options IPAddressAllow= and
175 IPAddressDeny=, taking a list of IPv4 or IPv6 addresses and masks,
176 for configuring a simple IP access control list for all sockets of
177 the unit. These options are available also on .slice and .socket
178 units, permitting flexible access list configuration for individual
179 services as well as groups of services (as defined by a slice unit),
180 including system-wide. Note that IP ACLs configured this way are
181 enforced on every single IPv4 and IPv6 socket created by any process
182 of the service unit, and apply to ingress as well as egress traffic.
184 * If CPUAccounting= or IPAccounting= is turned on for a unit a new,
185 recognizable log message is generated each time the unit is stopped,
186 containing information about the consumed resources of this
189 * A new setting KeyringMode= has been added to unit files, which may be
190 used to control how the kernel keyring is set up for executed
193 * .timer units now accept calendar specifications in other timezones
194 than UTC or the local timezone.
196 * The tmpfiles snippet var.conf has been changed to create
197 /var/log/btmp with access mode 0660 instead of 0600. It has been
198 owned by the "utmp" group already, and it appears to be generally
199 understood that members of "utmp" can modify/flush the
200 utmp/wtmp/lastlog/btmp databases. Previously this was implemented
201 correctly for all these database excepts btmp, which has been opened
202 up like this now too. Note that while the other databases are
203 world-readable (i.e. 0644), btmp is not and remains more restrictive.
205 Contributions from: Abdó Roig-Maranges, Alan Jenkins, Alexander
206 Kuleshov, Andreas Rammhold, Andrew Jeddeloh, Andrew Soutar, Ansgar
207 Burchardt, b1tninja, bengal, Benjamin Berg, Benjamin Robin, Charles
208 Huber, Christian Hesse, Daniel Berrange, Daniel Mack, Daniel Rusek,
209 dasj19, Davide Cavalca, Dimitri John Ledkov, Diogo Pereira, Djalal
210 Harouni, dkg, dmig, Dmitry Torokhov, ettavolt, Evgeny Vereshchagin,
211 Fabio Kung, Felipe Sateler, Franck Bui, g0tar, Hans de Goede, Harald
212 Hoyer, Insun Pyo, Ivan Kurnosov, Ivan Shapovalov, Jakub Wilk, Jan
213 Synacek, Jason Gunthorpe, Jeremy Bicha, Jérémy Rosen, John Lin,
214 jonasBoss, Jonathan Lebon, Jonathan Teh, Jon Ringle, Jörg Thalheim,
215 Jouke Witteveen, juga0, Justin Michaud, Kai-Heng Feng, Lennart
216 Poettering, Lion Yang, Luca Bruno, Lucas Werkmeister, Lukáš Nykrýn,
217 Marcel Hollerbach, Marcus Lundblad, Martin Pitt, Michael Biebl, Michael
218 Grzeschik, Michal Sekletar, Mike Gilbert, Neil Brown, Nicolas Iooss,
219 Patrik Flykt, pEJipE, Russell Stuart, S. Fan, Shengyao Xue, Stefan
220 Pietsch, Susant Sahani, Tejun Heo, Thomas Miller, Thomas Sailer, Tobias
221 Hunger, Tom Gundersen, Tommi Rantala, Topi Miettinen, Torstein Husebø,
222 userwithuid, Vito Caputo, vliaskov, WaLyong Cho, William Douglas, Xiang
223 Fan, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
229 * Meson is now supported as build system in addition to Automake. It is
230 our plan to remove Automake in one of our next releases, so that
231 Meson becomes our exclusive build system. Hence, please start using
232 the Meson build system in your downstream packaging. There's plenty
233 of documentation around how to use Meson, the extremely brief
236 ./autogen.sh && ./configure && make && sudo make install
240 meson build && ninja -C build && sudo ninja -C build install
242 * Unit files gained support for a new JobRunningTimeoutUSec= setting,
243 which permits configuring a timeout on the time a job is
244 running. This is particularly useful for setting timeouts on jobs for
247 * Unit files gained two new options ConditionUser= and ConditionGroup=
248 for conditionalizing units based on the identity of the user/group
249 running a systemd user instance.
251 * systemd-networkd now understands a new FlowLabel= setting in the
252 [VXLAN] section of .network files, as well as a Priority= in
253 [Bridge], GVRP= + MVRP= + LooseBinding= + ReorderHeader= in [VLAN]
254 and GatewayOnlink= + IPv6Preference= + Protocol= in [Route]. It also
255 gained support for configuration of GENEVE links, and IPv6 address
256 labels. The [Network] section gained the new IPv6ProxyNDP= setting.
258 * .link files now understand a new Port= setting.
260 * systemd-networkd's DHCP support gained support for DHCP option 119
261 (domain search list).
263 * systemd-networkd gained support for serving IPv6 address ranges using
264 the Router Advertisment protocol. The new .network configuration
265 section [IPv6Prefix] may be used to configure the ranges to
266 serve. This is implemented based on a new, minimal, native server
267 implementation of RA.
269 * journalctl's --output= switch gained support for a new parameter
270 "short-iso-precise" for a mode where timestamps are shown as precise
273 * systemd-udevd's "net_id" builtin may now generate stable network
274 interface names from IBM PowerVM VIO devices as well as ACPI platform
277 * MulticastDNS support in systemd-resolved may now be explicitly
278 enabled/disabled using the new MulticastDNS= configuration file
281 * systemd-resolved may now optionally use libidn2 instead of the libidn
282 for processing internationalized domain names. Support for libidn2
283 should be considered experimental and should not be enabled by
286 * "machinectl pull-tar" and related call may now do verification of
287 downloaded images using SUSE-style .sha256 checksum files in addition
288 to the already existing support for validating using Ubuntu-style
291 * sd-bus gained support for a new sd_bus_message_appendv() call which
292 is va_list equivalent of sd_bus_message_append().
294 * sd-boot gained support for validating images using SHIM/MOK.
296 * The SMACK code learnt support for "onlycap".
298 * systemd-mount --umount is now much smarter in figuring out how to
299 properly unmount a device given its mount or device path.
301 * The code to call libnss_dns as a fallback from libnss_resolve when
302 the communication with systemd-resolved fails was removed. This
303 fallback was redundant and interfered with the [!UNAVAIL=return]
304 suffix. See nss-resolve(8) for the recommended configuration.
306 * systemd-logind may now be restarted without losing state. It stores
307 the file descriptors for devices it manages in the system manager
308 using the FDSTORE= mechanism. Please note that further changes in
309 other components may be required to make use of this (for example
310 Xorg has code to listen for stops of systemd-logind and terminate
311 itself when logind is stopped or restarted, in order to avoid using
312 stale file descriptors for graphical devices, which is now
313 counterproductive and must be reverted in order for restarts of
314 systemd-logind to be safe. See
315 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
317 * All kernel install plugins are called with the environment variable
318 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
319 /etc/machine-id. If the file is missing or empty, the variable is
320 empty and BOOT_DIR_ABS is the path of a temporary directory which is
321 removed after all the plugins exit. So, if KERNEL_INSTALL_MACHINE_ID
322 is empty, all plugins should not put anything in BOOT_DIR_ABS.
324 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
325 Kurtz, Alexandros Frantzis, Alexey Brodkin, Alex Lu, Amir Pakdel, Amir
326 Yalon, Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert,
327 Benjamin Robin, Boucman, Charles Plessy, Chris Chiu, Chris Lamb,
328 Christian Brauner, Christian Hesse, Colin Walters, Daniel Drake,
329 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
330 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
331 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
332 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
333 Tierney, George McCollister, Giedrius Statkevičius, Hans de Goede,
334 hecke, Hendrik Westerberg, Hristo Venev, Ian Wienand, Insun Pyo, Ivan
335 Shapovalov, James Cowgill, James Hemsing, Janne Heß, Jan Synacek, Jason
336 Reeder, João Paulo Rechi Vita, John Paul Adrian Glaubitz, Jörg
337 Thalheim, Josef Andersson, Josef Gajdusek, Julian Mehne, Kai Krakow,
338 Krzysztof Jackiewicz, Lars Karlitski, Lennart Poettering, Lluís Gili,
339 Lucas Werkmeister, Lukáš Nykrýn, Łukasz Stelmach, Mantas Mikulėnas,
340 Marcin Bachry, Marcus Cooper, Mark Stosberg, Martin Pitt, Matija Skala,
341 Matt Clarkson, Matthew Garrett, Matthias Greiner, Matthijs van Duin,
342 Max Resch, Michael Biebl, Michal Koutný, Michal Sekletar, Michal
343 Soltys, Michal Suchanek, Mike Gilbert, Nate Clark, Nathaniel R. Lewis,
344 Neil Brown, Nikolai Kondrashov, Pascal S. de Kloe, Pat Riehecky, Patrik
345 Flykt, Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr
346 Szydełko, Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes,
347 Ronny Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan
348 Schweter, Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
349 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
350 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
351 userwithuid, Vito Caputo, Waldemar Brodkorb, WaLyong Cho, Yu, Li-Yu,
352 Yusuke Nojima, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Дамјан
359 * The "hybrid" control group mode has been modified to improve
360 compatibility with "legacy" cgroups-v1 setups. Specifically, the
361 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
362 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
363 cgroups-v1 hierarchy), the only externally visible change being that
364 the cgroups-v2 hierarchy is also mounted, to
365 /sys/fs/cgroup/unified. This should provide a large degree of
366 compatibility with "legacy" cgroups-v1, while taking benefit of the
367 better management capabilities of cgroups-v2.
369 * The default control group setup mode may be selected both a boot-time
370 via a set of kernel command line parameters (specifically:
371 systemd.unified_cgroup_hierarchy= and
372 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
373 default selected on the configure command line
374 (--with-default-hierarchy=). The upstream default is "hybrid"
375 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
376 this will change in a future systemd version to be "unified" (pure
377 cgroups-v2 mode). The third option for the compile time option is
378 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
379 distributions to default to "hybrid" mode for release distributions,
380 starting with v233. We recommend "unified" for development
381 distributions (specifically: distributions such as Fedora's rawhide)
382 as that's where things are headed in the long run. Use "legacy" for
383 greatest stability and compatibility only.
385 * Note one current limitation of "unified" and "hybrid" control group
386 setup modes: the kernel currently does not permit the systemd --user
387 instance (i.e. unprivileged code) to migrate processes between two
388 disconnected cgroup subtrees, even if both are managed and owned by
389 the user. This effectively means "systemd-run --user --scope" doesn't
390 work when invoked from outside of any "systemd --user" service or
391 scope. Specifically, it is not supported from session scopes. We are
392 working on fixing this in a future systemd version. (See #3388 for
393 further details about this.)
395 * DBus policy files are now installed into /usr rather than /etc. Make
396 sure your system has dbus >= 1.9.18 running before upgrading to this
397 version, or override the install path with --with-dbuspolicydir= .
399 * All python scripts shipped with systemd (specifically: the various
400 tests written in Python) now require Python 3.
402 * systemd unit tests can now run standalone (without the source or
403 build directories), and can be installed into /usr/lib/systemd/tests/
404 with 'make install-tests'.
406 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
407 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
410 * Support for the %c, %r, %R specifiers in unit files has been
411 removed. Specifiers are not supposed to be dependent on configuration
412 in the unit file itself (so that they resolve the same regardless
413 where used in the unit files), but these specifiers were influenced
414 by the Slice= option.
416 * The shell invoked by debug-shell.service now defaults to /bin/sh in
417 all cases. If distributions want to use a different shell for this
418 purpose (for example Fedora's /sbin/sushell) they need to specify
419 this explicitly at configure time using --with-debug-shell=.
421 * The confirmation spawn prompt has been reworked to offer the
424 (c)ontinue, proceed without asking anymore
425 (D)ump, show the state of the unit
426 (f)ail, don't execute the command and pretend it failed
428 (i)nfo, show a short summary of the unit
429 (j)obs, show jobs that are in progress
430 (s)kip, don't execute the command and pretend it succeeded
431 (y)es, execute the command
433 The 'n' choice for the confirmation spawn prompt has been removed,
434 because its meaning was confusing.
436 The prompt may now also be redirected to an alternative console by
437 specifying the console as parameter to systemd.confirm_spawn=.
439 * Services of Type=notify require a READY=1 notification to be sent
440 during startup. If no such message is sent, the service now fails,
441 even if the main process exited with a successful exit code.
443 * Services that fail to start up correctly now always have their
444 ExecStopPost= commands executed. Previously, they'd enter "failed"
445 state directly, without executing these commands.
447 * The option MulticastDNS= of network configuration files has acquired
448 an actual implementation. With MulticastDNS=yes a host can resolve
449 names of remote hosts and reply to mDNS A and AAAA requests.
451 * When units are about to be started an additional check is now done to
452 ensure that all dependencies of type BindsTo= (when used in
453 combination with After=) have been started.
455 * systemd-analyze gained a new verb "syscall-filter" which shows which
456 system call groups are defined for the SystemCallFilter= unit file
457 setting, and which system calls they contain.
459 * A new system call filter group "@filesystem" has been added,
460 consisting of various file system related system calls. Group
461 "@reboot" has been added, covering reboot, kexec and shutdown related
462 calls. Finally, group "@swap" has been added covering swap
463 configuration related calls.
465 * A new unit file option RestrictNamespaces= has been added that may be
466 used to restrict access to the various process namespace types the
467 Linux kernel provides. Specifically, it may be used to take away the
468 right for a service unit to create additional file system, network,
469 user, and other namespaces. This sandboxing option is particularly
470 relevant due to the high amount of recently discovered namespacing
471 related vulnerabilities in the kernel.
473 * systemd-udev's .link files gained support for a new AutoNegotiation=
474 setting for configuring Ethernet auto-negotiation.
476 * systemd-networkd's .network files gained support for a new
477 ListenPort= setting in the [DHCP] section to explicitly configure the
478 UDP client port the DHCP client shall listen on.
480 * .network files gained a new Unmanaged= boolean setting for explicitly
481 excluding one or more interfaces from management by systemd-networkd.
483 * The systemd-networkd ProxyARP= option has been renamed to
484 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
485 renamed to ReduceARPProxy=. The old names continue to be available
488 * systemd-networkd gained support for configuring IPv6 Proxy NDP
489 addresses via the new IPv6ProxyNDPAddress= .network file setting.
491 * systemd-networkd's bonding device support gained support for two new
492 configuration options ActiveSlave= and PrimarySlave=.
494 * The various options in the [Match] section of .network files gained
495 support for negative matching.
497 * New systemd-specific mount options are now understood in /etc/fstab:
499 x-systemd.mount-timeout= may be used to configure the maximum
500 permitted runtime of the mount command.
502 x-systemd.device-bound may be set to bind a mount point to its
503 backing device unit, in order to automatically remove a mount point
504 if its backing device is unplugged. This option may also be
505 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
506 on the block device, which is now automatically set for all CDROM
507 drives, so that mounted CDs are automatically unmounted when they are
508 removed from the drive.
510 x-systemd.after= and x-systemd.before= may be used to explicitly
511 order a mount after or before another unit or mount point.
513 * Enqueued start jobs for device units are now automatically garbage
514 collected if there are no jobs waiting for them anymore.
516 * systemctl list-jobs gained two new switches: with --after, for every
517 queued job the jobs it's waiting for are shown; with --before the
518 jobs which it's blocking are shown.
520 * systemd-nspawn gained support for ephemeral boots from disk images
521 (or in other words: --ephemeral and --image= may now be
522 combined). Moreover, ephemeral boots are now supported for normal
523 directories, even if the backing file system is not btrfs. Of course,
524 if the file system does not support file system snapshots or
525 reflinks, the initial copy operation will be relatively expensive, but
526 this should still be suitable for many use cases.
528 * Calendar time specifications in .timer units now support
529 specifications relative to the end of a month by using "~" instead of
530 "-" as separator between month and day. For example, "*-02~03" means
531 "the third last day in February". In addition a new syntax for
532 repeated events has been added using the "/" character. For example,
533 "9..17/2:00" means "every two hours from 9am to 5pm".
535 * systemd-socket-proxyd gained a new parameter --connections-max= for
536 configuring the maximum number of concurrent connections.
538 * sd-id128 gained a new API for generating unique IDs for the host in a
539 way that does not leak the machine ID. Specifically,
540 sd_id128_get_machine_app_specific() derives an ID based on the
541 machine ID a in well-defined, non-reversible, stable way. This is
542 useful whenever an identifier for the host is needed but where the
543 identifier shall not be useful to identify the system beyond the
544 scope of the application itself. (Internally this uses HMAC-SHA256 as
545 keyed hash function using the machine ID as input.)
547 * NotifyAccess= gained a new supported value "exec". When set
548 notifications are accepted from all processes systemd itself invoked,
549 including all control processes.
551 * .nspawn files gained support for defining overlay mounts using the
552 Overlay= and OverlayReadOnly= options. Previously this functionality
553 was only available on the systemd-nspawn command line.
555 * systemd-nspawn's --bind= and --overlay= options gained support for
556 bind/overlay mounts whose source lies within the container tree by
557 prefixing the source path with "+".
559 * systemd-nspawn's --bind= and --overlay= options gained support for
560 automatically allocating a temporary source directory in /var/tmp
561 that is removed when the container dies. Specifically, if the source
562 directory is specified as empty string this mechanism is selected. An
563 example usage is --overlay=+/var::/var, which creates an overlay
564 mount based on the original /var contained in the image, overlayed
565 with a temporary directory in the host's /var/tmp. This way changes
566 to /var are automatically flushed when the container shuts down.
568 * systemd-nspawn --image= option does now permit raw file system block
569 devices (in addition to images containing partition tables, as
572 * The disk image dissection logic in systemd-nspawn gained support for
573 automatically setting up LUKS encrypted as well as Verity protected
574 partitions. When a container is booted from an encrypted image the
575 passphrase is queried at start-up time. When a container with Verity
576 data is started, the root hash is search in a ".roothash" file
577 accompanying the disk image (alternatively, pass the root hash via
578 the new --root-hash= command line option).
580 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
581 be used to dissect disk images the same way as systemd-nspawn does
582 it, following the Bootable Partition Specification. It may even be
583 used to mount disk images with complex partition setups (including
584 LUKS and Verity partitions) to a local host directory, in order to
585 inspect them. This tool is not considered public API (yet), and is
586 thus not installed into /usr/bin. Please do not rely on its
587 existence, since it might go away or be changed in later systemd
590 * A new generator "systemd-verity-generator" has been added, similar in
591 style to "systemd-cryptsetup-generator", permitting automatic setup of
592 Verity root partitions when systemd boots up. In order to make use of
593 this your partition setup should follow the Discoverable Partitions
594 Specification, and the GPT partition ID of the root file system
595 partition should be identical to the upper 128bit of the Verity root
596 hash. The GPT partition ID of the Verity partition protecting it
597 should be the lower 128bit of the Verity root hash. If the partition
598 image follows this model it is sufficient to specify a single
599 "roothash=" kernel command line argument to both configure which root
600 image and verity partition to use as well as the root hash for
601 it. Note that systemd-nspawn's Verity support follows the same
602 semantics, meaning that disk images with proper Verity data in place
603 may be booted in containers with systemd-nspawn as well as on
604 physical systems via the verity generator. Also note that the "mkosi"
605 tool available at https://github.com/systemd/mkosi has been updated
606 to generate Verity protected disk images following this scheme. In
607 fact, it has been updated to generate disk images that optionally
608 implement a complete UEFI SecureBoot trust chain, involving a signed
609 kernel and initrd image that incorporates such a root hash as well as
610 a Verity-enabled root partition.
612 * The hardware database (hwdb) udev supports has been updated to carry
613 accelerometer quirks.
615 * All system services are now run with a fresh kernel keyring set up
616 for them. The invocation ID is stored by default in it, thus
617 providing a safe, non-overridable way to determine the invocation
620 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
621 options for bind mounting arbitrary paths in a service-specific
622 way. When these options are used, arbitrary host or service files and
623 directories may be mounted to arbitrary locations in the service's
626 * Documentation has been added that lists all of systemd's low-level
627 environment variables:
629 https://github.com/systemd/systemd/blob/master/ENVIRONMENT.md
631 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
632 whether a specific socket file descriptor matches a specified socket
635 * systemd-firstboot has been updated to check for the
636 systemd.firstboot= kernel command line option. It accepts a boolean
637 and when set to false the first boot questions are skipped.
639 * systemd-fstab-generator has been updated to check for the
640 systemd.volatile= kernel command line option, which either takes an
641 optional boolean parameter or the special value "state". If used the
642 system may be booted in a "volatile" boot mode. Specifically,
643 "systemd.volatile" is used, the root directory will be mounted as
644 tmpfs, and only /usr is mounted from the actual root file system. If
645 "systemd.volatile=state" is used, the root directory will be mounted
646 as usual, but /var is mounted as tmpfs. This concept provides similar
647 functionality as systemd-nspawn's --volatile= option, but provides it
648 on physical boots. Use this option for implementing stateless
649 systems, or testing systems with all state and/or configuration reset
650 to the defaults. (Note though that many distributions are not
651 prepared to boot up without a populated /etc or /var, though.)
653 * systemd-gpt-auto-generator gained support for LUKS encrypted root
654 partitions. Previously it only supported LUKS encrypted partitions
655 for all other uses, except for the root partition itself.
657 * Socket units gained support for listening on AF_VSOCK sockets for
658 communication in virtualized QEMU environments.
660 * The "configure" script gained a new option --with-fallback-hostname=
661 for specifying the fallback hostname to use if none is configured in
662 /etc/hostname. For example, by specifying
663 --with-fallback-hostname=fedora it is possible to default to a
664 hostname of "fedora" on pristine installations.
666 * systemd-cgls gained support for a new --unit= switch for listing only
667 the control groups of a specific unit. Similar --user-unit= has been
668 added for listing only the control groups of a specific user unit.
670 * systemd-mount gained a new --umount switch for unmounting a mount or
671 automount point (and all mount/automount points below it).
673 * systemd will now refuse full configuration reloads (via systemctl
674 daemon-reload and related calls) unless at least 16MiB of free space
675 are available in /run. This is a safety precaution in order to ensure
676 that generators can safely operate after the reload completed.
678 * A new unit file option RootImage= has been added, which has a similar
679 effect as RootDirectory= but mounts the service's root directory from
680 a disk image instead of plain directory. This logic reuses the same
681 image dissection and mount logic that systemd-nspawn already uses,
682 and hence supports any disk images systemd-nspawn supports, including
683 those following the Discoverable Partition Specification, as well as
684 Verity enabled images. This option enables systemd to run system
685 services directly off disk images acting as resource bundles,
686 possibly even including full integrity data.
688 * A new MountAPIVFS= unit file option has been added, taking a boolean
689 argument. If enabled /proc, /sys and /dev (collectively called the
690 "API VFS") will be mounted for the service. This is only relevant if
691 RootDirectory= or RootImage= is used for the service, as these mounts
692 are of course in place in the host mount namespace anyway.
694 * systemd-nspawn gained support for a new --pivot-root= switch. If
695 specified the root directory within the container image is pivoted to
696 the specified mount point, while the original root disk is moved to a
697 different place. This option enables booting of ostree images
698 directly with systemd-nspawn.
700 * The systemd build scripts will no longer complain if the NTP server
701 addresses are not changed from the defaults. Google now supports
702 these NTP servers officially. We still recommend downstreams to
703 properly register an NTP pool with the NTP pool project though.
705 * coredumpctl gained a new "--reverse" option for printing the list
706 of coredumps in reverse order.
708 * coredumpctl will now show additional information about truncated and
709 inaccessible coredumps, as well as coredumps that are still being
710 processed. It also gained a new --quiet switch for suppressing
711 additional informational message in its output.
713 * coredumpctl gained support for only showing coredumps newer and/or
714 older than specific timestamps, using the new --since= and --until=
715 options, reminiscent of journalctl's options by the same name.
717 * The systemd-coredump logic has been improved so that it may be reused
718 to collect backtraces in non-compiled languages, for example in
719 scripting languages such as Python.
721 * machinectl will now show the UID shift of local containers, if user
722 namespacing is enabled for them.
724 * systemd will now optionally run "environment generator" binaries at
725 configuration load time. They may be used to add environment
726 variables to the environment block passed to services invoked. One
727 user environment generator is shipped by default that sets up
728 environment variables based on files dropped into /etc/environment.d
729 and ~/.config/environment.d/.
731 * systemd-resolved now includes the new, recently published 2017 DNSSEC
734 * hostnamed has been updated to report a new chassis type of
735 "convertible" to cover "foldable" laptops that can both act as a
736 tablet and as a laptop, such as various Lenovo Yoga devices.
738 Contributions from: Adrián López, Alexander Galanin, Alexander
739 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
740 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
741 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
742 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
743 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
744 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
745 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
746 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
747 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
748 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
749 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
750 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
751 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
752 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
753 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
754 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
755 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
756 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
757 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
758 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
759 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
760 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
761 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
762 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
763 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
764 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
765 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
772 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
773 RestrictAddressFamilies= enabled. These sandboxing options should
774 generally be compatible with the various external udev call-out
775 binaries we are aware of, however there may be exceptions, in
776 particular when exotic languages for these call-outs are used. In
777 this case, consider turning off these settings locally.
779 * The new RemoveIPC= option can be used to remove IPC objects owned by
780 the user or group of a service when that service exits.
782 * The new ProtectKernelModules= option can be used to disable explicit
783 load and unload operations of kernel modules by a service. In
784 addition access to /usr/lib/modules is removed if this option is set.
786 * ProtectSystem= option gained a new value "strict", which causes the
787 whole file system tree with the exception of /dev, /proc, and /sys,
788 to be remounted read-only for a service.
790 * The new ProtectKernelTunables= option can be used to disable
791 modification of configuration files in /sys and /proc by a service.
792 Various directories and files are remounted read-only, so access is
793 restricted even if the file permissions would allow it.
795 * The new ProtectControlGroups= option can be used to disable write
796 access by a service to /sys/fs/cgroup.
798 * Various systemd services have been hardened with
799 ProtectKernelTunables=yes, ProtectControlGroups=yes,
800 RestrictAddressFamilies=.
802 * Support for dynamically creating users for the lifetime of a service
803 has been added. If DynamicUser=yes is specified, user and group IDs
804 will be allocated from the range 61184..65519 for the lifetime of the
805 service. They can be resolved using the new nss-systemd.so NSS
806 module. The module must be enabled in /etc/nsswitch.conf. Services
807 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
808 any resources allocated by the service will be cleaned up when the
809 service exits. They also have ProtectHome=read-only and
810 ProtectSystem=strict enabled, so they are not able to make any
811 permanent modifications to the system.
813 * The nss-systemd module also always resolves root and nobody, making
814 it possible to have no /etc/passwd or /etc/group files in minimal
815 container or chroot environments.
817 * Services may be started with their own user namespace using the new
818 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
819 under which the service is running are mapped. All other users are
822 * Support for the cgroup namespace has been added to systemd-nspawn. If
823 supported by kernel, the container system started by systemd-nspawn
824 will have its own view of the cgroup hierarchy. This new behaviour
825 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
827 * The new MemorySwapMax= option can be used to limit the maximum swap
828 usage under the unified cgroup hierarchy.
830 * Support for the CPU controller in the unified cgroup hierarchy has
831 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
832 options. This controller requires out-of-tree patches for the kernel
833 and the support is provisional.
835 * Mount and automount units may now be created transiently
836 (i.e. dynamically at runtime via the bus API, instead of requiring
837 unit files in the file system).
839 * systemd-mount is a new tool which may mount file systems – much like
840 mount(8), optionally pulling in additional dependencies through
841 transient .mount and .automount units. For example, this tool
842 automatically runs fsck on a backing block device before mounting,
843 and allows the automount logic to be used dynamically from the
844 command line for establishing mount points. This tool is particularly
845 useful when dealing with removable media, as it will ensure fsck is
846 run – if necessary – before the first access and that the file system
847 is quickly unmounted after each access by utilizing the automount
848 logic. This maximizes the chance that the file system on the
849 removable media stays in a clean state, and if it isn't in a clean
850 state is fixed automatically.
852 * LazyUnmount=yes option for mount units has been added to expose the
853 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
856 * /efi will be used as the mount point of the EFI boot partition, if
857 the directory is present, and the mount point was not configured
858 through other means (e.g. fstab). If /efi directory does not exist,
859 /boot will be used as before. This makes it easier to automatically
860 mount the EFI partition on systems where /boot is used for something
863 * When operating on GPT disk images for containers, systemd-nspawn will
864 now mount the ESP to /boot or /efi according to the same rules as PID
865 1 running on a host. This allows tools like "bootctl" to operate
866 correctly within such containers, in order to make container images
867 bootable on physical systems.
869 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
871 * Two new user session targets have been added to support running
872 graphical sessions under the systemd --user instance:
873 graphical-session.target and graphical-session-pre.target. See
874 systemd.special(7) for a description of how those targets should be
877 * The vconsole initialization code has been significantly reworked to
878 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
879 support unicode keymaps. Font and keymap configuration will now be
880 copied to all allocated virtual consoles.
882 * FreeBSD's bhyve virtualization is now detected.
884 * Information recorded in the journal for core dumps now includes the
885 contents of /proc/mountinfo and the command line of the process at
886 the top of the process hierarchy (which is usually the init process
889 * systemd-journal-gatewayd learned the --directory= option to serve
890 files from the specified location.
892 * journalctl --root=… can be used to peruse the journal in the
893 /var/log/ directories inside of a container tree. This is similar to
894 the existing --machine= option, but does not require the container to
897 * The hardware database has been extended to support
898 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
901 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
902 specify the click rate for mice which include a horizontal wheel with
903 a click rate that is different than the one for the vertical wheel.
905 * systemd-run gained a new --wait option that makes service execution
906 synchronous. (Specifically, the command will not return until the
907 specified service binary exited.)
909 * systemctl gained a new --wait option that causes the start command to
910 wait until the units being started have terminated again.
912 * A new journal output mode "short-full" has been added which displays
913 timestamps with abbreviated English day names and adds a timezone
914 suffix. Those timestamps include more information than the default
915 "short" output mode, and can be passed directly to journalctl's
916 --since= and --until= options.
918 * /etc/resolv.conf will be bind-mounted into containers started by
919 systemd-nspawn, if possible, so any changes to resolv.conf contents
920 are automatically propagated to the container.
922 * The number of instances for socket-activated services originating
923 from a single IP address can be limited with
924 MaxConnectionsPerSource=, extending the existing setting of
927 * systemd-networkd gained support for vcan ("Virtual CAN") interface
930 * .netdev and .network configuration can now be extended through
933 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
934 Segmentation Offload, Generic Receive Offload, Large Receive Offload
935 can be enabled and disabled using the new UDPSegmentationOffload=,
936 TCPSegmentationOffload=, GenericSegmentationOffload=,
937 GenericReceiveOffload=, LargeReceiveOffload= options in the
938 [Link] section of .link files.
940 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
941 Port VLAN ID can be configured for bridge devices using the new STP=,
942 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
943 section of .netdev files.
945 * The route table to which routes received over DHCP or RA should be
946 added can be configured with the new RouteTable= option in the [DHCP]
947 and [IPv6AcceptRA] sections of .network files.
949 * The Address Resolution Protocol can be disabled on links managed by
950 systemd-networkd using the ARP=no setting in the [Link] section of
953 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
954 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
955 encode information about the result and exit codes of the current
956 service runtime cycle.
958 * systemd-sysctl will now configure kernel parameters in the order
959 they occur in the configuration files. This matches what sysctl
960 has been traditionally doing.
962 * kernel-install "plugins" that are executed to perform various
963 tasks after a new kernel is added and before an old one is removed
964 can now return a special value to terminate the procedure and
965 prevent any later plugins from running.
967 * Journald's SplitMode=login setting has been deprecated. It has been
968 removed from documentation, and its use is discouraged. In a future
969 release it will be completely removed, and made equivalent to current
970 default of SplitMode=uid.
972 * Storage=both option setting in /etc/systemd/coredump.conf has been
973 removed. With fast LZ4 compression storing the core dump twice is not
976 * The --share-system systemd-nspawn option has been replaced with an
977 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
978 this functionality is discouraged. In addition the variables
979 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
980 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
981 individual namespaces.
983 * "machinectl list" now shows the IP address of running containers in
984 the output, as well as OS release information.
986 * "loginctl list" now shows the TTY of each session in the output.
988 * sd-bus gained new API calls sd_bus_track_set_recursive(),
989 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
990 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
991 tracking objects in a "recursive" mode, where a single client can be
992 counted multiple times, if it takes multiple references.
994 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
995 sd_bus_get_exit_on_disconnect(). They may be used to to make a
996 process using sd-bus automatically exit if the bus connection is
999 * Bus clients of the service manager may now "pin" loaded units into
1000 memory, by taking an explicit reference on them. This is useful to
1001 ensure the client can retrieve runtime data about the service even
1002 after the service completed execution. Taking such a reference is
1003 available only for privileged clients and should be helpful to watch
1004 running services in a race-free manner, and in particular collect
1005 information about exit statuses and results.
1007 * The nss-resolve module has been changed to strictly return UNAVAIL
1008 when communication via D-Bus with resolved failed, and NOTFOUND when
1009 a lookup completed but was negative. This means it is now possible to
1010 neatly configure fallbacks using nsswitch.conf result checking
1011 expressions. Taking benefit of this, the new recommended
1012 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
1014 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
1016 * A new setting CtrlAltDelBurstAction= has been added to
1017 /etc/systemd/system.conf which may be used to configure the precise
1018 behaviour if the user on the console presses Ctrl-Alt-Del more often
1019 than 7 times in 2s. Previously this would unconditionally result in
1020 an expedited, immediate reboot. With this new setting the precise
1021 operation may be configured in more detail, and also turned off
1024 * In .netdev files two new settings RemoteChecksumTx= and
1025 RemoteChecksumRx= are now understood that permit configuring the
1026 remote checksumming logic for VXLAN networks.
1028 * The service manager learnt a new "invocation ID" concept for invoked
1029 services. Each runtime cycle of a service will get a new invocation
1030 ID (a 128bit random UUID) assigned that identifies the current
1031 run of the service uniquely and globally. A new invocation ID
1032 is generated each time a service starts up. The journal will store
1033 the invocation ID of a service along with any logged messages, thus
1034 making the invocation ID useful for matching the online runtime of a
1035 service with the offline log data it generated in a safe way without
1036 relying on synchronized timestamps. In many ways this new service
1037 invocation ID concept is similar to the kernel's boot ID concept that
1038 uniquely and globally identifies the runtime of each boot. The
1039 invocation ID of a service is passed to the service itself via an
1040 environment variable ($INVOCATION_ID). A new bus call
1041 GetUnitByInvocationID() has been added that is similar to GetUnit()
1042 but instead of retrieving the bus path for a unit by its name
1043 retrieves it by its invocation ID. The returned path is valid only as
1044 long as the passed invocation ID is current.
1046 * systemd-resolved gained a new "DNSStubListener" setting in
1047 resolved.conf. It either takes a boolean value or the special values
1048 "udp" and "tcp", and configures whether to enable the stub DNS
1049 listener on 127.0.0.53:53.
1051 * IP addresses configured via networkd may now carry additional
1052 configuration settings supported by the kernel. New options include:
1053 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
1054 PrefixRoute=, AutoJoin=.
1056 * The PAM configuration fragment file for "user@.service" shipped with
1057 systemd (i.e. the --user instance of systemd) has been stripped to
1058 the minimum necessary to make the system boot. Previously, it
1059 contained Fedora-specific stanzas that did not apply to other
1060 distributions. It is expected that downstream distributions add
1061 additional configuration lines, matching their needs to this file,
1062 using it only as rough template of what systemd itself needs. Note
1063 that this reduced fragment does not even include an invocation of
1064 pam_limits which most distributions probably want to add, even though
1065 systemd itself does not need it. (There's also the new build time
1066 option --with-pamconfdir=no to disable installation of the PAM
1069 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
1070 capability is now also dropped from its set (in addition to
1071 CAP_SYS_MKNOD as before).
1073 * In service unit files it is now possible to connect a specific named
1074 file descriptor with stdin/stdout/stdout of an executed service. The
1075 name may be specified in matching .socket units using the
1076 FileDescriptorName= setting.
1078 * A number of journal settings may now be configured on the kernel
1079 command line. Specifically, the following options are now understood:
1080 systemd.journald.max_level_console=,
1081 systemd.journald.max_level_store=,
1082 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
1083 systemd.journald.max_level_wall=.
1085 * "systemctl is-enabled --full" will now show by which symlinks a unit
1086 file is enabled in the unit dependency tree.
1088 * Support for VeraCrypt encrypted partitions has been added to the
1089 "cryptsetup" logic and /etc/crypttab.
1091 * systemd-detect-virt gained support for a new --private-users switch
1092 that checks whether the invoking processes are running inside a user
1093 namespace. Similar, a new special value "private-users" for the
1094 existing ConditionVirtualization= setting has been added, permitting
1095 skipping of specific units in user namespace environments.
1097 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
1098 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
1099 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
1100 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
1101 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
1102 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
1103 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
1104 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
1105 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
1106 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
1107 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
1108 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
1109 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
1110 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
1111 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
1112 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
1113 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
1114 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
1115 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
1116 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
1117 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
1118 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
1119 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
1120 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
1121 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
1122 Jędrzejewski-Szmek, Zeal Jagannatha
1124 — Santa Fe, 2016-11-03
1128 * In service units the various ExecXYZ= settings have been extended
1129 with an additional special character as first argument of the
1130 assigned value: if the character '+' is used the specified command
1131 line it will be run with full privileges, regardless of User=,
1132 Group=, CapabilityBoundingSet= and similar options. The effect is
1133 similar to the existing PermissionsStartOnly= option, but allows
1134 configuration of this concept for each executed command line
1137 * Services may now alter the service watchdog timeout at runtime by
1138 sending a WATCHDOG_USEC= message via sd_notify().
1140 * MemoryLimit= and related unit settings now optionally take percentage
1141 specifications. The percentage is taken relative to the amount of
1142 physical memory in the system (or in case of containers, the assigned
1143 amount of memory). This allows scaling service resources neatly with
1144 the amount of RAM available on the system. Similarly, systemd-logind's
1145 RuntimeDirectorySize= option now also optionally takes percentage
1148 * In similar fashion TasksMax= takes percentage values now, too. The
1149 value is taken relative to the configured maximum number of processes
1150 on the system. The per-service task maximum has been changed to 15%
1151 using this functionality. (Effectively this is an increase of 512 →
1152 4915 for service units, given the kernel's default pid_max setting.)
1154 * Calendar time specifications in .timer units now understand a ".."
1155 syntax for time ranges. Example: "4..7:10" may now be used for
1156 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
1159 * The InaccessableDirectories=, ReadOnlyDirectories= and
1160 ReadWriteDirectories= unit file settings have been renamed to
1161 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
1162 applied to all kinds of file nodes, and not just directories, with
1163 the exception of symlinks. Specifically these settings may now be
1164 used on block and character device nodes, UNIX sockets and FIFOS as
1165 well as regular files. The old names of these settings remain
1166 available for compatibility.
1168 * systemd will now log about all service processes it kills forcibly
1169 (using SIGKILL) because they remained after the clean shutdown phase
1170 of the service completed. This should help identifying services that
1171 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
1172 systemd-logind's configuration a similar log message is generated for
1173 processes killed at the end of each session due to this setting.
1175 * systemd will now set the $JOURNAL_STREAM environment variable for all
1176 services whose stdout/stderr are connected to the Journal (which
1177 effectively means by default: all services). The variable contains
1178 the device and inode number of the file descriptor used for
1179 stdout/stderr. This may be used by invoked programs to detect whether
1180 their stdout/stderr is connected to the Journal, in which case they
1181 can switch over to direct Journal communication, thus being able to
1182 pass extended, structured metadata along with their log messages. As
1183 one example, this is now used by glib's logging primitives.
1185 * When using systemd's default tmp.mount unit for /tmp, the mount point
1186 will now be established with the "nosuid" and "nodev" options. This
1187 avoids privilege escalation attacks that put traps and exploits into
1188 /tmp. However, this might cause problems if you e. g. put container
1189 images or overlays into /tmp; if you need this, override tmp.mount's
1190 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
1193 * systemd now supports the "memory" cgroup controller also on
1196 * The systemd-cgtop tool now optionally takes a control group path as
1197 command line argument. If specified, the control group list shown is
1198 limited to subgroups of that group.
1200 * The SystemCallFilter= unit file setting gained support for
1201 pre-defined, named system call filter sets. For example
1202 SystemCallFilter=@clock is now an effective way to make all clock
1203 changing-related system calls unavailable to a service. A number of
1204 similar pre-defined groups are defined. Writing system call filters
1205 for system services is simplified substantially with this new
1206 concept. Accordingly, all of systemd's own, long-running services now
1207 enable system call filtering based on this, by default.
1209 * A new service setting MemoryDenyWriteExecute= has been added, taking
1210 a boolean value. If turned on, a service may no longer create memory
1211 mappings that are writable and executable at the same time. This
1212 enhances security for services where this is enabled as it becomes
1213 harder to dynamically write and then execute memory in exploited
1214 service processes. This option has been enabled for all of systemd's
1215 own long-running services.
1217 * A new RestrictRealtime= service setting has been added, taking a
1218 boolean argument. If set the service's processes may no longer
1219 acquire realtime scheduling. This improves security as realtime
1220 scheduling may otherwise be used to easily freeze the system.
1222 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
1223 value. This may be used for requesting that the system manager inside
1224 of the container reports start-up completion to nspawn which then
1225 propagates this notification further to the service manager
1226 supervising nspawn itself. A related option NotifyReady= in .nspawn
1227 files has been added too. This functionality allows ordering of the
1228 start-up of multiple containers using the usual systemd ordering
1231 * machinectl gained a new command "stop" that is an alias for
1234 * systemd-resolved gained support for contacting DNS servers on
1235 link-local IPv6 addresses.
1237 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
1238 its caches. A method call for requesting the same operation has been
1239 added to the bus API too, and is made available via "systemd-resolve
1242 * systemd-resolve gained a new --status switch. If passed a brief
1243 summary of the used DNS configuration with per-interface information
1246 * resolved.conf gained a new Cache= boolean option, defaulting to
1247 on. If turned off local DNS caching is disabled. This comes with a
1248 performance penalty in particular when DNSSEC is enabled. Note that
1249 resolved disables its internal caching implicitly anyway, when the
1250 configured DNS server is on a host-local IP address such as ::1 or
1251 127.0.0.1, thus automatically avoiding double local caching.
1253 * systemd-resolved now listens on the local IP address 127.0.0.53:53
1254 for DNS requests. This improves compatibility with local programs
1255 that do not use the libc NSS or systemd-resolved's bus APIs for name
1256 resolution. This minimal DNS service is only available to local
1257 programs and does not implement the full DNS protocol, but enough to
1258 cover local DNS clients. A new, static resolv.conf file, listing just
1259 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
1260 now recommended to make /etc/resolv.conf a symlink to this file in
1261 order to route all DNS lookups to systemd-resolved, regardless if
1262 done via NSS, the bus API or raw DNS packets. Note that this local
1263 DNS service is not as fully featured as the libc NSS or
1264 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
1265 used to deliver link-local address information (as this implies
1266 sending a local interface index along), LLMNR/mDNS support via this
1267 interface is severely restricted. It is thus strongly recommended for
1268 all applications to use the libc NSS API or native systemd-resolved
1271 * systemd-networkd's bridge support learned a new setting
1272 VLANFiltering= for controlling VLAN filtering. Moreover a new section
1273 in .network files has been added for configuring VLAN bridging in
1274 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
1276 * systemd-networkd's IPv6 Router Advertisement code now makes use of
1277 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
1278 now be acquired without relying on DHCPv6. Two new options
1279 UseDomains= and UseDNS= have been added to configure this behaviour.
1281 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
1282 renamed IPv6AcceptRA=, without altering its behaviour. The old
1283 setting name remains available for compatibility reasons.
1285 * The systemd-networkd VTI/VTI6 tunneling support gained new options
1286 Key=, InputKey= and OutputKey=.
1288 * systemd-networkd gained support for VRF ("Virtual Routing Function")
1289 interface configuration.
1291 * "systemctl edit" may now be used to create new unit files by
1292 specifying the --force switch.
1294 * sd-event gained a new function sd_event_get_iteration() for
1295 requesting the current iteration counter of the event loop. It starts
1296 at zero and is increased by one with each event loop iteration.
1298 * A new rpm macro %systemd_ordering is provided by the macros.systemd
1299 file. It can be used in lieu of %systemd_requires in packages which
1300 don't use any systemd functionality and are intended to be installed
1301 in minimal containers without systemd present. This macro provides
1302 ordering dependencies to ensure that if the package is installed in
1303 the same rpm transaction as systemd, systemd will be installed before
1304 the scriptlets for the package are executed, allowing unit presets
1307 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
1308 been added to simplify packaging of generators.
1310 * The os-release file gained VERSION_CODENAME field for the
1311 distribution nickname (e.g. VERSION_CODENAME=woody).
1313 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
1314 can be set to disable parsing of metadata and the creation
1315 of persistent symlinks for that device.
1317 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
1318 to make them available to logged-in users has been reverted.
1320 * Much of the common code of the various systemd components is now
1321 built into an internal shared library libsystemd-shared-231.so
1322 (incorporating the systemd version number in the name, to be updated
1323 with future releases) that the components link to. This should
1324 decrease systemd footprint both in memory during runtime and on
1325 disk. Note that the shared library is not for public use, and is
1326 neither API not ABI stable, but is likely to change with every new
1327 released update. Packagers need to make sure that binaries
1328 linking to libsystemd-shared.so are updated in step with the
1331 * Configuration for "mkosi" is now part of the systemd
1332 repository. mkosi is a tool to easily build legacy-free OS images,
1333 and is available on github: https://github.com/systemd/mkosi. If
1334 "mkosi" is invoked in the build tree a new raw OS image is generated
1335 incorporating the systemd sources currently being worked on and a
1336 clean, fresh distribution installation. The generated OS image may be
1337 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
1338 UEFI PC. This functionality is particularly useful to easily test
1339 local changes made to systemd in a pristine, defined environment. See
1340 HACKING for details.
1342 * configure learned the --with-support-url= option to specify the
1343 distribution's bugtracker.
1345 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
1346 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
1347 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
1348 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
1349 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
1350 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
1351 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
1352 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
1353 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
1354 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
1355 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
1356 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
1357 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
1358 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
1359 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
1360 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
1361 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
1362 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
1363 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
1365 — Berlin, 2016-07-25
1369 * DNSSEC is now turned on by default in systemd-resolved (in
1370 "allow-downgrade" mode), but may be turned off during compile time by
1371 passing "--with-default-dnssec=no" to "configure" (and of course,
1372 during runtime with DNSSEC= in resolved.conf). We recommend
1373 downstreams to leave this on at least during development cycles and
1374 report any issues with the DNSSEC logic upstream. We are very
1375 interested in collecting feedback about the DNSSEC validator and its
1376 limitations in the wild. Note however, that DNSSEC support is
1377 probably nothing downstreams should turn on in stable distros just
1378 yet, as it might create incompatibilities with a few DNS servers and
1379 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
1380 automatically whenever we detect such incompatible setups, but there
1381 might be systems we do not cover yet. Hence: please help us testing
1382 the DNSSEC code, leave this on where you can, report back, but then
1383 again don't consider turning this on in your stable, LTS or
1384 production release just yet. (Note that you have to enable
1385 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
1386 and its DNSSEC mode for host name resolution from local
1389 * systemd-resolve conveniently resolves DANE records with the --tlsa
1390 option and OPENPGPKEY records with the --openpgp option. It also
1391 supports dumping raw DNS record data via the new --raw= switch.
1393 * systemd-logind will now by default terminate user processes that are
1394 part of the user session scope unit (session-XX.scope) when the user
1395 logs out. This behavior is controlled by the KillUserProcesses=
1396 setting in logind.conf, and the previous default of "no" is now
1397 changed to "yes". This means that user sessions will be properly
1398 cleaned up after, but additional steps are necessary to allow
1399 intentionally long-running processes to survive logout.
1401 While the user is logged in at least once, user@.service is running,
1402 and any service that should survive the end of any individual login
1403 session can be started at a user service or scope using systemd-run.
1404 systemd-run(1) man page has been extended with an example which shows
1405 how to run screen in a scope unit underneath user@.service. The same
1406 command works for tmux.
1408 After the user logs out of all sessions, user@.service will be
1409 terminated too, by default, unless the user has "lingering" enabled.
1410 To effectively allow users to run long-term tasks even if they are
1411 logged out, lingering must be enabled for them. See loginctl(1) for
1412 details. The default polkit policy was modified to allow users to
1413 set lingering for themselves without authentication.
1415 Previous defaults can be restored at compile time by the
1416 --without-kill-user-processes option to "configure".
1418 * systemd-logind gained new configuration settings SessionsMax= and
1419 InhibitorsMax=, both with a default of 8192. It will not register new
1420 user sessions or inhibitors above this limit.
1422 * systemd-logind will now reload configuration on SIGHUP.
1424 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
1425 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
1426 enable. Also, support for the "io" cgroup controller in the unified
1427 hierarchy has been added, so that the "memory", "pids" and "io" are
1428 now the controllers that are supported on the unified hierarchy.
1430 WARNING: it is not possible to use previous systemd versions with
1431 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
1432 is necessary to also update systemd in the initramfs if using the
1433 unified hierarchy. An updated SELinux policy is also required.
1435 * LLDP support has been extended, and both passive (receive-only) and
1436 active (sender) modes are supported. Passive mode ("routers-only") is
1437 enabled by default in systemd-networkd. Active LLDP mode is enabled
1438 by default for containers on the internal network. The "networkctl
1439 lldp" command may be used to list information gathered. "networkctl
1440 status" will also show basic LLDP information on connected peers now.
1442 * The IAID and DUID unique identifier sent in DHCP requests may now be
1443 configured for the system and each .network file managed by
1444 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
1446 * systemd-networkd gained support for configuring proxy ARP support for
1447 each interface, via the ProxyArp= setting in .network files. It also
1448 gained support for configuring the multicast querier feature of
1449 bridge devices, via the new MulticastQuerier= setting in .netdev
1450 files. Similarly, snooping on the IGMP traffic can be controlled
1451 via the new setting MulticastSnooping=.
1453 A new setting PreferredLifetime= has been added for addresses
1454 configured in .network file to configure the lifetime intended for an
1457 The systemd-networkd DHCP server gained the option EmitRouter=, which
1458 defaults to yes, to configure whether the DHCP Option 3 (Router)
1461 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
1462 systemd-socket-activate and installed into /usr/bin. It is now fully
1465 * systemd-journald now uses separate threads to flush changes to disk
1466 when closing journal files, thus reducing impact of slow disk I/O on
1467 logging performance.
1469 * The sd-journal API gained two new calls
1470 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
1471 can be used to open journal files using file descriptors instead of
1472 file or directory paths. sd_journal_open_container() has been
1473 deprecated, sd_journal_open_directory_fd() should be used instead
1474 with the flag SD_JOURNAL_OS_ROOT.
1476 * journalctl learned a new output mode "-o short-unix" that outputs log
1477 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
1478 UTC). It also gained support for a new --no-hostname setting to
1479 suppress the hostname column in the family of "short" output modes.
1481 * systemd-ask-password now optionally skips printing of the password to
1482 stdout with --no-output which can be useful in scripts.
1484 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
1485 (devices tagged with ID_MAKER_TOOL) are now tagged with
1486 "uaccess" and are available to logged in users.
1488 * The DeviceAllow= unit setting now supports specifiers (with "%").
1490 * "systemctl show" gained a new --value switch, which allows print a
1491 only the contents of a specific unit property, without also printing
1492 the property's name. Similar support was added to "show*" verbs
1493 of loginctl and machinectl that output "key=value" lists.
1495 * A new unit type "generated" was added for files dynamically generated
1496 by generator tools. Similarly, a new unit type "transient" is used
1497 for unit files created using the runtime API. "systemctl enable" will
1498 refuse to operate on such files.
1500 * A new command "systemctl revert" has been added that may be used to
1501 revert to the vendor version of a unit file, in case local changes
1502 have been made by adding drop-ins or overriding the unit file.
1504 * "machinectl clean" gained a new verb to automatically remove all or
1505 just hidden container images.
1507 * systemd-tmpfiles gained support for a new line type "e" for emptying
1508 directories, if they exist, without creating them if they don't.
1510 * systemd-nspawn gained support for automatically patching the UID/GIDs
1511 of the owners and the ACLs of all files and directories in a
1512 container tree to match the UID/GID user namespacing range selected
1513 for the container invocation. This mode is enabled via the new
1514 --private-users-chown switch. It also gained support for
1515 automatically choosing a free, previously unused UID/GID range when
1516 starting a container, via the new --private-users=pick setting (which
1517 implies --private-users-chown). Together, these options for the first
1518 time make user namespacing for nspawn containers fully automatic and
1519 thus deployable. The systemd-nspawn@.service template unit file has
1520 been changed to use this functionality by default.
1522 * systemd-nspawn gained a new --network-zone= switch, that allows
1523 creating ad-hoc virtual Ethernet links between multiple containers,
1524 that only exist as long as at least one container referencing them is
1525 running. This allows easy connecting of multiple containers with a
1526 common link that implements an Ethernet broadcast domain. Each of
1527 these network "zones" may be named relatively freely by the user, and
1528 may be referenced by any number of containers, but each container may
1529 only reference one of these "zones". On the lower level, this is
1530 implemented by an automatically managed bridge network interface for
1531 each zone, that is created when the first container referencing its
1532 zone is created and removed when the last one referencing its zone
1535 * The default start timeout may now be configured on the kernel command
1536 line via systemd.default_timeout_start_sec=. It was already
1537 configurable via the DefaultTimeoutStartSec= option in
1538 /etc/systemd/system.conf.
1540 * Socket units gained a new TriggerLimitIntervalSec= and
1541 TriggerLimitBurst= setting to configure a limit on the activation
1542 rate of the socket unit.
1544 * The LimitNICE= setting now optionally takes normal UNIX nice values
1545 in addition to the raw integer limit value. If the specified
1546 parameter is prefixed with "+" or "-" and is in the range -20..19 the
1547 value is understood as UNIX nice value. If not prefixed like this it
1548 is understood as raw RLIMIT_NICE limit.
1550 * Note that the effect of the PrivateDevices= unit file setting changed
1551 slightly with this release: the per-device /dev file system will be
1552 mounted read-only from this version on, and will have "noexec"
1553 set. This (minor) change of behavior might cause some (exceptional)
1554 legacy software to break, when PrivateDevices=yes is set for its
1555 service. Please leave PrivateDevices= off if you run into problems
1558 * systemd-bootchart has been split out to a separate repository:
1559 https://github.com/systemd/systemd-bootchart
1561 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
1562 merged into the kernel in its current form.
1564 * The compatibility libraries libsystemd-daemon.so,
1565 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
1566 which have been deprecated since systemd-209 have been removed along
1567 with the corresponding pkg-config files. All symbols provided by
1568 those libraries are provided by libsystemd.so.
1570 * The Capabilities= unit file setting has been removed (it is ignored
1571 for backwards compatibility). AmbientCapabilities= and
1572 CapabilityBoundingSet= should be used instead.
1574 * A new special target has been added, initrd-root-device.target,
1575 which creates a synchronization point for dependencies of the root
1576 device in early userspace. Initramfs builders must ensure that this
1577 target is now included in early userspace.
1579 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
1580 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
1581 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
1582 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
1583 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
1584 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
1585 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
1586 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
1587 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
1588 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
1589 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
1590 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
1591 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
1592 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
1593 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
1594 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
1595 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
1596 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
1597 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
1598 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
1599 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
1600 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
1601 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
1602 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
1603 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
1606 — Fairfax, 2016-05-21
1610 * The systemd-resolved DNS resolver service has gained a substantial
1611 set of new features, most prominently it may now act as a DNSSEC
1612 validating stub resolver. DNSSEC mode is currently turned off by
1613 default, but is expected to be turned on by default in one of the
1614 next releases. For now, we invite everybody to test the DNSSEC logic
1615 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
1616 service also gained a full set of D-Bus interfaces, including calls
1617 to configure DNS and DNSSEC settings per link (for use by external
1618 network management software). systemd-resolved and systemd-networkd
1619 now distinguish between "search" and "routing" domains. The former
1620 are used to qualify single-label names, the latter are used purely
1621 for routing lookups within certain domains to specific links.
1622 resolved now also synthesizes RRs for all entries from /etc/hosts.
1624 * The systemd-resolve tool (which is a client utility for
1625 systemd-resolved) has been improved considerably and is now fully
1626 supported and documented. Hence it has moved from /usr/lib/systemd to
1629 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
1632 * The coredump collection logic has been reworked: when a coredump is
1633 collected it is now written to disk, compressed and processed
1634 (including stacktrace extraction) from a new instantiated service
1635 systemd-coredump@.service, instead of directly from the
1636 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
1637 processing large coredumps can take up a substantial amount of
1638 resources and time, and this previously happened entirely outside of
1639 systemd's service supervision. With the new logic the core_pattern
1640 hook only does minimal metadata collection before passing off control
1641 to the new instantiated service, which is configured with a time
1642 limit, a nice level and other settings to minimize negative impact on
1643 the rest of the system. Also note that the new logic will honour the
1644 RLIMIT_CORE setting of the crashed process, which now allows users
1645 and processes to turn off coredumping for their processes by setting
1648 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
1649 and all forked processes by default. Previously, PID 1 would leave
1650 the setting at "0" for all processes, as set by the kernel. Note that
1651 the resource limit traditionally has no effect on the generated
1652 coredumps on the system if the /proc/sys/kernel/core_pattern hook
1653 logic is used. Since the limit is now honoured (see above) its
1654 default has been changed so that the coredumping logic is enabled by
1655 default for all processes, while allowing specific opt-out.
1657 * When the stacktrace is extracted from processes of system users, this
1658 is now done as "systemd-coredump" user, in order to sandbox this
1659 potentially security sensitive parsing operation. (Note that when
1660 processing coredumps of normal users this is done under the user ID
1661 of process that crashed, as before.) Packagers should take notice
1662 that it is now necessary to create the "systemd-coredump" system user
1663 and group at package installation time.
1665 * The systemd-activate socket activation testing tool gained support
1666 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
1667 and --seqpacket switches. It also has been extended to support both
1668 new-style and inetd-style file descriptor passing. Use the new
1669 --inetd switch to request inetd-style file descriptor passing.
1671 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
1672 variable, which takes a boolean value. If set to false, ANSI color
1673 output is disabled in the tools even when run on a terminal that
1676 * The VXLAN support in networkd now supports two new settings
1677 DestinationPort= and PortRange=.
1679 * A new systemd.machine_id= kernel command line switch has been added,
1680 that may be used to set the machine ID in /etc/machine-id if it is
1681 not initialized yet. This command line option has no effect if the
1682 file is already initialized.
1684 * systemd-nspawn gained a new --as-pid2 switch that invokes any
1685 specified command line as PID 2 rather than PID 1 in the
1686 container. In this mode PID 1 is a minimal stub init process that
1687 implements the special POSIX and Linux semantics of PID 1 regarding
1688 signal and child process management. Note that this stub init process
1689 is implemented in nspawn itself and requires no support from the
1690 container image. This new logic is useful to support running
1691 arbitrary commands in the container, as normal processes are
1692 generally not prepared to run as PID 1.
1694 * systemd-nspawn gained a new --chdir= switch for setting the current
1695 working directory for the process started in the container.
1697 * "journalctl /dev/sda" will now output all kernel log messages for
1698 specified device from the current boot, in addition to all devices
1699 that are parents of it. This should make log output about devices
1700 pretty useful, as long as kernel drivers attach enough metadata to
1701 the log messages. (The usual SATA drivers do.)
1703 * The sd-journal API gained two new calls
1704 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
1705 that report whether log data from /run or /var has been found.
1707 * journalctl gained a new switch "--fields" that prints all journal
1708 record field names currently in use in the journal. This is backed
1709 by two new sd-journal API calls sd_journal_enumerate_fields() and
1710 sd_journal_restart_fields().
1712 * Most configurable timeouts in systemd now expect an argument of
1713 "infinity" to turn them off, instead of "0" as before. The semantics
1714 from now on is that a timeout of "0" means "now", and "infinity"
1715 means "never". To maintain backwards compatibility, "0" continues to
1716 turn off previously existing timeout settings.
1718 * "systemctl reload-or-try-restart" has been renamed to "systemctl
1719 try-reload-or-restart" to clarify what it actually does: the "try"
1720 logic applies to both reloading and restarting, not just restarting.
1721 The old name continues to be accepted for compatibility.
1723 * On boot-up, when PID 1 detects that the system clock is behind the
1724 release date of the systemd version in use, the clock is now set
1725 to the latter. Previously, this was already done in timesyncd, in order
1726 to avoid running with clocks set to the various clock epochs such as
1727 1902, 1938 or 1970. With this change the logic is now done in PID 1
1728 in addition to timesyncd during early boot-up, so that it is enforced
1729 before the first process is spawned by systemd. Note that the logic
1730 in timesyncd remains, as it is more comprehensive and ensures
1731 clock monotonicity by maintaining a persistent timestamp file in
1732 /var. Since /var is generally not available in earliest boot or the
1733 initrd, this part of the logic remains in timesyncd, and is not done
1736 * Support for tweaking details in net_cls.class_id through the
1737 NetClass= configuration directive has been removed, as the kernel
1738 people have decided to deprecate that controller in cgroup v2.
1739 Userspace tools such as nftables are moving over to setting rules
1740 that are specific to the full cgroup path of a task, which obsoletes
1741 these controllers anyway. The NetClass= directive is kept around for
1742 legacy compatibility reasons. For a more in-depth description of the
1743 kernel change, please refer to the respective upstream commit:
1745 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
1747 * A new service setting RuntimeMaxSec= has been added that may be used
1748 to specify a maximum runtime for a service. If the timeout is hit, the
1749 service is terminated and put into a failure state.
1751 * A new service setting AmbientCapabilities= has been added. It allows
1752 configuration of additional Linux process capabilities that are
1753 passed to the activated processes. This is only available on very
1756 * The process resource limit settings in service units may now be used
1757 to configure hard and soft limits individually.
1759 * The various libsystemd APIs such as sd-bus or sd-event now publicly
1760 expose support for gcc's __attribute__((cleanup())) C extension.
1761 Specifically, for many object destructor functions alternative
1762 versions have been added that have names suffixed with "p" and take a
1763 pointer to a pointer to the object to destroy, instead of just a
1764 pointer to the object itself. This is useful because these destructor
1765 functions may be used directly as parameters to the cleanup
1766 construct. Internally, systemd has been a heavy user of this GCC
1767 extension for a long time, and with this change similar support is
1768 now available to consumers of the library outside of systemd. Note
1769 that by using this extension in your sources compatibility with old
1770 and strictly ANSI compatible C compilers is lost. However, all gcc or
1771 LLVM versions of recent years support this extension.
1773 * Timer units gained support for a new setting RandomizedDelaySec= that
1774 allows configuring some additional randomized delay to the configured
1775 time. This is useful to spread out timer events to avoid load peaks in
1776 clusters or larger setups.
1778 * Calendar time specifications now support sub-second accuracy.
1780 * Socket units now support listening on SCTP and UDP-lite protocol
1783 * The sd-event API now comes with a full set of man pages.
1785 * Older versions of systemd contained experimental support for
1786 compressing journal files and coredumps with the LZ4 compressor that
1787 was not compatible with the lz4 binary (due to API limitations of the
1788 lz4 library). This support has been removed; only support for files
1789 compatible with the lz4 binary remains. This LZ4 logic is now
1790 officially supported and no longer considered experimental.
1792 * The dkr image import logic has been removed again from importd. dkr's
1793 micro-services focus doesn't fit into the machine image focus of
1794 importd, and quickly got out of date with the upstream dkr API.
1796 * Creation of the /run/lock/lockdev/ directory was dropped from
1797 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
1798 been available for many years. If you still need this, you need to
1799 create your own tmpfiles.d config file with:
1801 d /run/lock/lockdev 0775 root lock -
1803 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
1804 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
1805 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
1806 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
1807 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
1808 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
1809 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
1810 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
1811 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
1812 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
1813 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
1814 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
1815 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
1816 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
1817 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
1818 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
1819 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
1820 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
1821 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
1823 — Berlin, 2016-02-11
1827 * A number of properties previously only settable in unit
1828 files are now also available as properties to set when
1829 creating transient units programmatically via the bus, as it
1830 is exposed with systemd-run's --property=
1831 setting. Specifically, these are: SyslogIdentifier=,
1832 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
1833 EnvironmentFile=, ReadWriteDirectories=,
1834 ReadOnlyDirectories=, InaccessibleDirectories=,
1835 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
1837 * When creating transient services via the bus API it is now
1838 possible to pass in a set of file descriptors to use as
1839 STDIN/STDOUT/STDERR for the invoked process.
1841 * Slice units may now be created transiently via the bus APIs,
1842 similar to the way service and scope units may already be
1843 created transiently.
1845 * Wherever systemd expects a calendar timestamp specification
1846 (like in journalctl's --since= and --until= switches) UTC
1847 timestamps are now supported. Timestamps suffixed with "UTC"
1848 are now considered to be in Universal Time Coordinated
1849 instead of the local timezone. Also, timestamps may now
1850 optionally be specified with sub-second accuracy. Both of
1851 these additions also apply to recurring calendar event
1852 specification, such as OnCalendar= in timer units.
1854 * journalctl gained a new "--sync" switch that asks the
1855 journal daemon to write all so far unwritten log messages to
1856 disk and sync the files, before returning.
1858 * systemd-tmpfiles learned two new line types "q" and "Q" that
1859 operate like "v", but also set up a basic btrfs quota
1860 hierarchy when used on a btrfs file system with quota
1863 * tmpfiles' "v", "q" and "Q" will now create a plain directory
1864 instead of a subvolume (even on a btrfs file system) if the
1865 root directory is a plain directory, and not a
1866 subvolume. This should simplify things with certain chroot()
1867 environments which are not aware of the concept of btrfs
1870 * systemd-detect-virt gained a new --chroot switch to detect
1871 whether execution takes place in a chroot() environment.
1873 * CPUAffinity= now takes CPU index ranges in addition to
1876 * The various memory-related resource limit settings (such as
1877 LimitAS=) now understand the usual K, M, G, ... suffixes to
1878 the base of 1024 (IEC). Similar, the time-related resource
1879 limit settings understand the usual min, h, day, ...
1882 * There's a new system.conf setting DefaultTasksMax= to
1883 control the default TasksMax= setting for services and
1884 scopes running on the system. (TasksMax= is the primary
1885 setting that exposes the "pids" cgroup controller on systemd
1886 and was introduced in the previous systemd release.) The
1887 setting now defaults to 512, which means services that are
1888 not explicitly configured otherwise will only be able to
1889 create 512 processes or threads at maximum, from this
1890 version on. Note that this means that thread- or
1891 process-heavy services might need to be reconfigured to set
1892 TasksMax= to a higher value. It is sufficient to set
1893 TasksMax= in these specific unit files to a higher value, or
1894 even "infinity". Similar, there's now a logind.conf setting
1895 UserTasksMax= that defaults to 4096 and limits the total
1896 number of processes or tasks each user may own
1897 concurrently. nspawn containers also have the TasksMax=
1898 value set by default now, to 8192. Note that all of this
1899 only has an effect if the "pids" cgroup controller is
1900 enabled in the kernel. The general benefit of these changes
1901 should be a more robust and safer system, that provides a
1902 certain amount of per-service fork() bomb protection.
1904 * systemd-nspawn gained the new --network-veth-extra= switch
1905 to define additional and arbitrarily-named virtual Ethernet
1906 links between the host and the container.
1908 * A new service execution setting PassEnvironment= has been
1909 added that allows importing select environment variables
1910 from PID1's environment block into the environment block of
1913 * Timer units gained support for a new RemainAfterElapse=
1914 setting which takes a boolean argument. It defaults to on,
1915 exposing behaviour unchanged to previous releases. If set to
1916 off, timer units are unloaded after they elapsed if they
1917 cannot elapse again. This is particularly useful for
1918 transient timer units, which shall not stay around longer
1919 than until they first elapse.
1921 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
1922 default now (the kernel default is 16). This is beneficial
1923 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
1924 allows substantially larger numbers of queued
1925 datagrams. This should increase the capability of systemd to
1926 parallelize boot-up, as logging and sd_notify() are unlikely
1927 to stall execution anymore. If you need to change the value
1928 from the new defaults, use the usual sysctl.d/ snippets.
1930 * The compression framing format used by the journal or
1931 coredump processing has changed to be in line with what the
1932 official LZ4 tools generate. LZ4 compression support in
1933 systemd was considered unsupported previously, as the format
1934 was not compatible with the normal tools. With this release
1935 this has changed now, and it is hence safe for downstream
1936 distributions to turn it on. While not compressing as well
1937 as the XZ, LZ4 is substantially faster, which makes
1938 it a good default choice for the compression logic in the
1939 journal and in coredump handling.
1941 * Any reference to /etc/mtab has been dropped from
1942 systemd. The file has been obsolete since a while, but
1943 systemd refused to work on systems where it was incorrectly
1944 set up (it should be a symlink or non-existent). Please make
1945 sure to update to util-linux 2.27.1 or newer in conjunction
1946 with this systemd release, which also drops any reference to
1947 /etc/mtab. If you maintain a distribution make sure that no
1948 software you package still references it, as this is a
1949 likely source of bugs. There's also a glibc bug pending,
1950 asking for removal of any reference to this obsolete file:
1952 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
1954 Note that only util-linux versions built with
1955 --enable-libmount-force-mountinfo are supported.
1957 * Support for the ".snapshot" unit type has been removed. This
1958 feature turned out to be little useful and little used, and
1959 has now been removed from the core and from systemctl.
1961 * The dependency types RequiresOverridable= and
1962 RequisiteOverridable= have been removed from systemd. They
1963 have been used only very sparingly to our knowledge and
1964 other options that provide a similar effect (such as
1965 systemctl --mode=ignore-dependencies) are much more useful
1966 and commonly used. Moreover, they were only half-way
1967 implemented as the option to control behaviour regarding
1968 these dependencies was never added to systemctl. By removing
1969 these dependency types the execution engine becomes a bit
1970 simpler. Unit files that use these dependencies should be
1971 changed to use the non-Overridable dependency types
1972 instead. In fact, when parsing unit files with these
1973 options, that's what systemd will automatically convert them
1974 too, but it will also warn, asking users to fix the unit
1975 files accordingly. Removal of these dependency types should
1976 only affect a negligible number of unit files in the wild.
1978 * Behaviour of networkd's IPForward= option changed
1979 (again). It will no longer maintain a per-interface setting,
1980 but propagate one way from interfaces where this is enabled
1981 to the global kernel setting. The global setting will be
1982 enabled when requested by a network that is set up, but
1983 never be disabled again. This change was made to make sure
1984 IPv4 and IPv6 behaviour regarding packet forwarding is
1985 similar (as the Linux IPv6 stack does not support
1986 per-interface control of this setting) and to minimize
1989 * In unit files the behaviour of %u, %U, %h, %s has
1990 changed. These specifiers will now unconditionally resolve
1991 to the various user database fields of the user that the
1992 systemd instance is running as, instead of the user
1993 configured in the specific unit via User=. Note that this
1994 effectively doesn't change much, as resolving of these
1995 specifiers was already turned off in the --system instance
1996 of systemd, as we cannot do NSS lookups from PID 1. In the
1997 --user instance of systemd these specifiers where correctly
1998 resolved, but hardly made any sense, since the user instance
1999 lacks privileges to do user switches anyway, and User= is
2000 hence useless. Moreover, even in the --user instance of
2001 systemd behaviour was awkward as it would only take settings
2002 from User= assignment placed before the specifier into
2003 account. In order to unify and simplify the logic around
2004 this the specifiers will now always resolve to the
2005 credentials of the user invoking the manager (which in case
2006 of PID 1 is the root user).
2008 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
2009 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
2010 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
2011 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
2012 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
2013 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
2014 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
2015 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
2016 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
2017 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
2018 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
2019 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
2020 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
2021 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
2024 — Berlin, 2015-11-18
2028 * systemd now depends on util-linux v2.27. More specifically,
2029 the newly added mount monitor feature in libmount now
2030 replaces systemd's former own implementation.
2032 * libmount mandates /etc/mtab not to be regular file, and
2033 systemd now enforces this condition at early boot.
2034 /etc/mtab has been deprecated and warned about for a very
2035 long time, so systems running systemd should already have
2036 stopped having this file around as anything else than a
2037 symlink to /proc/self/mounts.
2039 * Support for the "pids" cgroup controller has been added. It
2040 allows accounting the number of tasks in a cgroup and
2041 enforcing limits on it. This adds two new setting
2042 TasksAccounting= and TasksMax= to each unit, as well as a
2043 global option DefaultTasksAccounting=.
2045 * Support for the "net_cls" cgroup controller has been added.
2046 It allows assigning a net class ID to each task in the
2047 cgroup, which can then be used in firewall rules and traffic
2048 shaping configurations. Note that the kernel netfilter net
2049 class code does not currently work reliably for ingress
2050 packets on unestablished sockets.
2052 This adds a new config directive called NetClass= to CGroup
2053 enabled units. Allowed values are positive numbers for fixed
2054 assignments and "auto" for picking a free value
2057 * 'systemctl is-system-running' now returns 'offline' if the
2058 system is not booted with systemd. This command can now be
2059 used as a substitute for 'systemd-notify --booted'.
2061 * Watchdog timeouts have been increased to 3 minutes for all
2062 in-tree service files. Apparently, disk IO issues are more
2063 frequent than we hoped, and user reported >1 minute waiting
2066 * 'machine-id-commit' functionality has been merged into
2067 'machine-id-setup --commit'. The separate binary has been
2070 * The WorkingDirectory= directive in unit files may now be set
2071 to the special value '~'. In this case, the working
2072 directory is set to the home directory of the user
2073 configured in User=.
2075 * "machinectl shell" will now open the shell in the home
2076 directory of the selected user by default.
2078 * The CrashChVT= configuration file setting is renamed to
2079 CrashChangeVT=, following our usual logic of not
2080 abbreviating unnecessarily. The old directive is still
2081 supported for compat reasons. Also, this directive now takes
2082 an integer value between 1 and 63, or a boolean value. The
2083 formerly supported '-1' value for disabling stays around for
2086 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
2087 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
2088 RootDirectory= properties can now be set for transient
2091 * The systemd-analyze tool gained a new "set-log-target" verb
2092 to change the logging target the system manager logs to
2093 dynamically during runtime. This is similar to how
2094 "systemd-analyze set-log-level" already changes the log
2097 * In nspawn /sys is now mounted as tmpfs, with only a selected
2098 set of subdirectories mounted in from the real sysfs. This
2099 enhances security slightly, and is useful for ensuring user
2100 namespaces work correctly.
2102 * Support for USB FunctionFS activation has been added. This
2103 allows implementation of USB gadget services that are
2104 activated as soon as they are requested, so that they don't
2105 have to run continuously, similar to classic socket
2108 * The "systemctl exit" command now optionally takes an
2109 additional parameter that sets the exit code to return from
2110 the systemd manager when exiting. This is only relevant when
2111 running the systemd user instance, or when running the
2112 system instance in a container.
2114 * sd-bus gained the new API calls sd_bus_path_encode_many()
2115 and sd_bus_path_decode_many() that allow easy encoding and
2116 decoding of multiple identifier strings inside a D-Bus
2117 object path. Another new call sd_bus_default_flush_close()
2118 has been added to flush and close per-thread default
2121 * systemd-cgtop gained support for a -M/--machine= switch to
2122 show the control groups within a certain container only.
2124 * "systemctl kill" gained support for an optional --fail
2125 switch. If specified the requested operation will fail of no
2126 processes have been killed, because the unit had no
2127 processes attached, or similar.
2129 * A new systemd.crash_reboot=1 kernel command line option has
2130 been added that triggers a reboot after crashing. This can
2131 also be set through CrashReboot= in systemd.conf.
2133 * The RuntimeDirectory= setting now understands unit
2134 specifiers like %i or %f.
2136 * A new (still internal) library API sd-ipv4acd has been added,
2137 that implements address conflict detection for IPv4. It's
2138 based on code from sd-ipv4ll, and will be useful for
2139 detecting DHCP address conflicts.
2141 * File descriptors passed during socket activation may now be
2142 named. A new API sd_listen_fds_with_names() is added to
2143 access the names. The default names may be overridden,
2144 either in the .socket file using the FileDescriptorName=
2145 parameter, or by passing FDNAME= when storing the file
2146 descriptors using sd_notify().
2148 * systemd-networkd gained support for:
2150 - Setting the IPv6 Router Advertisement settings via
2151 IPv6AcceptRouterAdvertisements= in .network files.
2153 - Configuring the HelloTimeSec=, MaxAgeSec= and
2154 ForwardDelaySec= bridge parameters in .netdev files.
2156 - Configuring PreferredSource= for static routes in
2159 * The "ask-password" framework used to query for LUKS harddisk
2160 passwords or SSL passwords during boot gained support for
2161 caching passwords in the kernel keyring, if it is
2162 available. This makes sure that the user only has to type in
2163 a passphrase once if there are multiple objects to unlock
2164 with the same one. Previously, such password caching was
2165 available only when Plymouth was used; this moves the
2166 caching logic into the systemd codebase itself. The
2167 "systemd-ask-password" utility gained a new --keyname=
2168 switch to control which kernel keyring key to use for
2169 caching a password in. This functionality is also useful for
2170 enabling display managers such as gdm to automatically
2171 unlock the user's GNOME keyring if its passphrase, the
2172 user's password and the harddisk password are the same, if
2173 gdm-autologin is used.
2175 * When downloading tar or raw images using "machinectl
2176 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
2177 file is now also downloaded, if it is available and stored
2178 next to the image file.
2180 * Units of type ".socket" gained a new boolean setting
2181 Writable= which is only useful in conjunction with
2182 ListenSpecial=. If true, enables opening the specified
2183 special file in O_RDWR mode rather than O_RDONLY mode.
2185 * systemd-rfkill has been reworked to become a singleton
2186 service that is activated through /dev/rfkill on each rfkill
2187 state change and saves the settings to disk. This way,
2188 systemd-rfkill is now compatible with devices that exist
2189 only intermittendly, and even restores state if the previous
2190 system shutdown was abrupt rather than clean.
2192 * The journal daemon gained support for vacuuming old journal
2193 files controlled by the number of files that shall remain,
2194 in addition to the already existing control by size and by
2195 date. This is useful as journal interleaving performance
2196 degrades with too many separate journal files, and allows
2197 putting an effective limit on them. The new setting defaults
2198 to 100, but this may be changed by setting SystemMaxFiles=
2199 and RuntimeMaxFiles= in journald.conf. Also, the
2200 "journalctl" tool gained the new --vacuum-files= switch to
2201 manually vacuum journal files to leave only the specified
2202 number of files in place.
2204 * udev will now create /dev/disk/by-path links for ATA devices
2205 on kernels where that is supported.
2207 * Galician, Serbian, Turkish and Korean translations were added.
2209 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
2210 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
2211 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
2212 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
2213 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
2214 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
2215 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
2216 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
2217 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
2218 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
2219 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
2220 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
2221 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
2222 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
2223 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
2224 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
2225 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
2226 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
2228 — Berlin, 2015-10-07
2232 * The DHCP implementation of systemd-networkd gained a set of
2235 - The DHCP server now supports emitting DNS and NTP
2236 information. It may be enabled and configured via
2237 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
2238 and NTP information is enabled, but no servers are
2239 configured, the corresponding uplink information (if there
2240 is any) is propagated.
2242 - Server and client now support transmission and reception
2243 of timezone information. It can be configured via the
2244 newly introduced network options UseTimezone=,
2245 EmitTimezone=, and Timezone=. Transmission of timezone
2246 information is enabled between host and containers by
2247 default now: the container will change its local timezone
2248 to what the host has set.
2250 - Lease timeouts can now be configured via
2251 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
2253 - The DHCP server improved on the stability of
2254 leases. Clients are more likely to get the same lease
2255 information back, even if the server loses state.
2257 - The DHCP server supports two new configuration options to
2258 control the lease address pool metrics, PoolOffset= and
2261 * The encapsulation limit of tunnels in systemd-networkd may
2262 now be configured via 'EncapsulationLimit='. It allows
2263 modifying the maximum additional levels of encapsulation
2264 that are permitted to be prepended to a packet.
2266 * systemd now supports the concept of user buses replacing
2267 session buses, if used with dbus-1.10 (and enabled via dbus
2268 --enable-user-session). It previously only supported this on
2269 kdbus-enabled systems, and this release expands this to
2270 'dbus-daemon' systems.
2272 * systemd-networkd now supports predictable interface names
2275 * systemd now optionally supports the new Linux kernel
2276 "unified" control group hierarchy. If enabled via the kernel
2277 command-line option 'systemd.unified_cgroup_hierarchy=1',
2278 systemd will try to mount the unified cgroup hierarchy
2279 directly on /sys/fs/cgroup. If not enabled, or not
2280 available, systemd will fall back to the legacy cgroup
2281 hierarchy setup, as before. Host system and containers can
2282 mix and match legacy and unified hierarchies as they
2283 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
2284 environment variable to individually select the hierarchy to
2285 use for executed containers. By default, nspawn will use the
2286 unified hierarchy for the containers if the host uses the
2287 unified hierarchy, and the legacy hierarchy otherwise.
2288 Please note that at this point the unified hierarchy is an
2289 experimental kernel feature and is likely to change in one
2290 of the next kernel releases. Therefore, it should not be
2291 enabled by default in downstream distributions yet. The
2292 minimum required kernel version for the unified hierarchy to
2293 work is 4.2. Note that when the unified hierarchy is used
2294 for the first time delegated access to controllers is
2295 safe. Because of this systemd-nspawn containers will get
2296 access to controllers now, as will systemd user
2297 sessions. This means containers and user sessions may now
2298 manage their own resources, partitioning up what the system
2301 * A new special scope unit "init.scope" has been introduced
2302 that encapsulates PID 1 of the system. It may be used to
2303 determine resource usage and enforce resource limits on PID
2304 1 itself. PID 1 hence moved out of the root of the control
2307 * The cgtop tool gained support for filtering out kernel
2308 threads when counting tasks in a control group. Also, the
2309 count of processes is now recursively summed up by
2310 default. Two options -k and --recursive= have been added to
2311 revert to old behaviour. The tool has also been updated to
2312 work correctly in containers now.
2314 * systemd-nspawn's --bind= and --bind-ro= options have been
2315 extended to allow creation of non-recursive bind mounts.
2317 * libsystemd gained two new calls sd_pid_get_cgroup() and
2318 sd_peer_get_cgroup() which return the control group path of
2319 a process or peer of a connected AF_UNIX socket. This
2320 function call is particularly useful when implementing
2321 delegated subtrees support in the control group hierarchy.
2323 * The "sd-event" event loop API of libsystemd now supports
2324 correct dequeuing of real-time signals, without losing
2327 * When systemd requests a PolicyKit decision when managing
2328 units it will now add additional fields to the request,
2329 including unit name and desired operation. This enables more
2330 powerful PolicyKit policies, that make decisions depending
2331 on these parameters.
2333 * nspawn learnt support for .nspawn settings files, that may
2334 accompany the image files or directories of containers, and
2335 may contain additional settings for the container. This is
2336 an alternative to configuring container parameters via the
2337 nspawn command line.
2339 Contributions from: Cristian Rodríguez, Daniel Mack, David
2340 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
2341 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
2342 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
2343 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
2344 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
2345 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
2346 Andersen, Tom Gundersen, Torstein Husebø
2348 — Berlin, 2015-09-08
2352 * machinectl gained a new verb 'shell' which opens a fresh
2353 shell on the target container or the host. It is similar to
2354 the existing 'login' command of machinectl, but spawns the
2355 shell directly without prompting for username or
2356 password. The pseudo machine '.host' now refers to the local
2357 host and is used by default. Hence, 'machinectl shell' can
2358 be used as replacement for 'su -' which spawns a session as
2359 a fresh systemd unit in a way that is fully isolated from
2360 the originating session.
2362 * systemd-networkd learned to cope with private-zone DHCP
2363 options and allows other programs to query the values.
2365 * SELinux access control when enabling/disabling units is no
2366 longer enforced with this release. The previous
2367 implementation was incorrect, and a new corrected
2368 implementation is not yet available. As unit file operations
2369 are still protected via PolicyKit and D-Bus policy this is
2370 not a security problem. Yet, distributions which care about
2371 optimal SELinux support should probably not stabilize on
2374 * sd-bus gained support for matches of type "arg0has=", that
2375 test for membership of strings in string arrays sent in bus
2378 * systemd-resolved now dumps the contents of its DNS and LLMNR
2379 caches to the logs on reception of the SIGUSR1 signal. This
2380 is useful to debug DNS behaviour.
2382 * The coredumpctl tool gained a new --directory= option to
2383 operate on journal files in a specific directory.
2385 * "systemctl reboot" and related commands gained a new
2386 "--message=" option which may be used to set a free-text
2387 wall message when shutting down or rebooting the
2388 system. This message is also logged, which is useful for
2389 figuring out the reason for a reboot or shutdown a
2392 * The "systemd-resolve-host" tool's -i switch now takes
2393 network interface numbers as alternative to interface names.
2395 * A new unit file setting for services has been introduced:
2396 UtmpMode= allows configuration of how precisely systemd
2397 handles utmp and wtmp entries for the service if this is
2398 enabled. This allows writing services that appear similar to
2399 user sessions in the output of the "w", "who", "last" and
2402 * systemd-resolved will now locally synthesize DNS resource
2403 records for the "localhost" and "gateway" domains as well as
2404 the local hostname. This should ensure that clients querying
2405 RRs via resolved will get similar results as those going via
2406 NSS, if nss-myhostname is enabled.
2408 Contributions from: Alastair Hughes, Alex Crawford, Daniel
2409 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
2410 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
2411 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
2412 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
2413 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
2414 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
2415 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
2416 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
2417 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
2418 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
2419 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
2421 — Berlin, 2015-08-27
2425 * The systemd-efi-boot-generator functionality was merged into
2426 systemd-gpt-auto-generator.
2428 * systemd-networkd now supports Group Policy for vxlan
2429 devices. It can be enabled via the new boolean configuration
2430 option called 'GroupPolicyExtension='.
2432 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
2433 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
2434 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
2436 — Berlin, 2015-07-31
2440 * The python-systemd code has been removed from the systemd repository.
2441 A new repository has been created which accommodates the code from
2442 now on, and we kindly ask distributions to create a separate package
2443 for this: https://github.com/systemd/python-systemd
2445 * The systemd daemon will now reload its main configuration
2446 (/etc/systemd/system.conf) on daemon-reload.
2448 * sd-dhcp now exposes vendor specific extensions via
2449 sd_dhcp_lease_get_vendor_specific().
2451 * systemd-networkd gained a number of new configuration options.
2453 - A new boolean configuration option for TAP devices called
2454 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
2455 device, thus allowing to send and receive GSO packets.
2457 - A new tunnel configuration option called 'CopyDSCP='.
2458 If enabled, the DSCP field of ip6 tunnels is copied into the
2459 decapsulated packet.
2461 - A set of boolean bridge configuration options were added.
2462 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
2463 and 'UnicastFlood=' are now parsed by networkd and applied to the
2464 respective bridge link device via the respective IFLA_BRPORT_*
2467 - A new string configuration option to override the hostname sent
2468 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
2469 is true, networkd will use the configured hostname instead of the
2470 system hostname when sending DHCP requests.
2472 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
2473 networkd will configure the IPv6 flow-label of the tunnel device
2474 according to RFC2460.
2476 - The 'macvtap' virtual network devices are now supported, similar to
2477 the already supported 'macvlan' devices.
2479 * systemd-resolved now implements RFC5452 to improve resilience against
2480 cache poisoning. Additionally, source port randomization is enabled
2481 by default to further protect against DNS spoofing attacks.
2483 * nss-mymachines now supports translating UIDs and GIDs of running
2484 containers with user-namespaces enabled. If a container 'foo'
2485 translates a host uid 'UID' to the container uid 'TUID', then
2486 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
2487 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
2488 mapped as 'vg-foo-TGID'.
2490 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
2491 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
2492 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
2493 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
2494 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
2495 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
2496 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
2497 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
2498 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
2499 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
2501 — Berlin, 2015-07-29
2505 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
2506 There are no known issues with current sysfs, and udev does not need
2507 or should be used to work around such bugs.
2509 * udev does no longer enable USB HID power management. Several reports
2510 indicate, that some devices cannot handle that setting.
2512 * The udev accelerometer helper was removed. The functionality
2513 is now fully included in iio-sensor-proxy. But this means,
2514 older iio-sensor-proxy versions will no longer provide
2515 accelerometer/orientation data with this systemd version.
2516 Please upgrade iio-sensor-proxy to version 1.0.
2518 * networkd gained a new configuration option IPv6PrivacyExtensions=
2519 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
2520 for Stateless Address") on selected networks.
2522 * For the sake of fewer build-time dependencies and less code in the
2523 main repository, the python bindings are about to be removed in the
2524 next release. A new repository has been created which accommodates
2525 the code from now on, and we kindly ask distributions to create a
2526 separate package for this. The removal will take place in v223.
2528 https://github.com/systemd/python-systemd
2530 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
2531 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
2532 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
2533 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
2534 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
2535 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
2536 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
2537 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
2538 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
2539 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
2541 — Berlin, 2015-07-07
2545 * The sd-bus.h and sd-event.h APIs have now been declared
2546 stable and have been added to the official interface of
2547 libsystemd.so. sd-bus implements an alternative D-Bus client
2548 library, that is relatively easy to use, very efficient and
2549 supports both classic D-Bus as well as kdbus as transport
2550 backend. sd-event is a generic event loop abstraction that
2551 is built around Linux epoll, but adds features such as event
2552 prioritization or efficient timer handling. Both APIs are good
2553 choices for C programs looking for a bus and/or event loop
2554 implementation that is minimal and does not have to be
2555 portable to other kernels.
2557 * kdbus support is no longer compile-time optional. It is now
2558 always built-in. However, it can still be disabled at
2559 runtime using the kdbus=0 kernel command line setting, and
2560 that setting may be changed to default to off, by specifying
2561 --disable-kdbus at build-time. Note though that the kernel
2562 command line setting has no effect if the kdbus.ko kernel
2563 module is not installed, in which case kdbus is (obviously)
2564 also disabled. We encourage all downstream distributions to
2565 begin testing kdbus by adding it to the kernel images in the
2566 development distributions, and leaving kdbus support in
2569 * The minimal required util-linux version has been bumped to
2572 * Support for chkconfig (--enable-chkconfig) was removed in
2573 favor of calling an abstraction tool
2574 /lib/systemd/systemd-sysv-install. This needs to be
2575 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
2576 in README for details.
2578 * If there's a systemd unit and a SysV init script for the
2579 same service name, and the user executes "systemctl enable"
2580 for it (or a related call), then this will now enable both
2581 (or execute the related operation on both), not just the
2584 * The libudev API documentation has been converted from gtkdoc
2587 * gudev has been removed from the systemd tree, it is now an
2590 * The systemd-cgtop tool learnt a new --raw switch to generate
2591 "raw" (machine parsable) output.
2593 * networkd's IPForwarding= .network file setting learnt the
2594 new setting "kernel", which ensures that networkd does not
2595 change the IP forwarding sysctl from the default kernel
2598 * The systemd-logind bus API now exposes a new boolean
2599 property "Docked" that reports whether logind considers the
2600 system "docked", i.e. connected to a docking station or not.
2602 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
2603 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
2604 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
2605 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
2606 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
2607 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
2608 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
2609 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
2610 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
2611 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
2612 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
2613 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
2614 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
2615 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
2616 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
2617 Fink, Zbigniew Jędrzejewski-Szmek
2619 — Berlin, 2015-06-19
2623 * The gudev library has been extracted into a separate repository
2624 available at: https://git.gnome.org/browse/libgudev/
2625 It is now managed as part of the Gnome project. Distributions
2626 are recommended to pass --disable-gudev to systemd and use
2627 gudev from the Gnome project instead. gudev is still included
2628 in systemd, for now. It will be removed soon, though. Please
2629 also see the announcement-thread on systemd-devel:
2630 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
2632 * systemd now exposes a CPUUsageNSec= property for each
2633 service unit on the bus, that contains the overall consumed
2634 CPU time of a service (the sum of what each process of the
2635 service consumed). This value is only available if
2636 CPUAccounting= is turned on for a service, and is then shown
2637 in the "systemctl status" output.
2639 * Support for configuring alternative mappings of the old SysV
2640 runlevels to systemd targets has been removed. They are now
2641 hardcoded in a way that runlevels 2, 3, 4 all map to
2642 multi-user.target and 5 to graphical.target (which
2643 previously was already the default behaviour).
2645 * The auto-mounter logic gained support for mount point
2646 expiry, using a new TimeoutIdleSec= setting in .automount
2647 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
2649 * The EFI System Partition (ESP) as mounted to /boot by
2650 systemd-efi-boot-generator will now be unmounted
2651 automatically after 2 minutes of not being used. This should
2652 minimize the risk of ESP corruptions.
2654 * New /etc/fstab options x-systemd.requires= and
2655 x-systemd.requires-mounts-for= are now supported to express
2656 additional dependencies for mounts. This is useful for
2657 journalling file systems that support external journal
2658 devices or overlay file systems that require underlying file
2659 systems to be mounted.
2661 * systemd does not support direct live-upgrades (via systemctl
2662 daemon-reexec) from versions older than v44 anymore. As no
2663 distribution we are aware of shipped such old versions in a
2664 stable release this should not be problematic.
2666 * When systemd forks off a new per-connection service instance
2667 it will now set the $REMOTE_ADDR environment variable to the
2668 remote IP address, and $REMOTE_PORT environment variable to
2669 the remote IP port. This behaviour is similar to the
2670 corresponding environment variables defined by CGI.
2672 * systemd-networkd gained support for uplink failure
2673 detection. The BindCarrier= option allows binding interface
2674 configuration dynamically to the link sense of other
2675 interfaces. This is useful to achieve behaviour like in
2678 * systemd-networkd gained support for configuring the DHCP
2679 client identifier to use when requesting leases.
2681 * systemd-networkd now has a per-network UseNTP= option to
2682 configure whether NTP server information acquired via DHCP
2683 is passed on to services like systemd-timesyncd.
2685 * systemd-networkd gained support for vti6 tunnels.
2687 * Note that systemd-networkd manages the sysctl variable
2688 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
2689 it is configured for since v219. The variable controls IP
2690 forwarding, and is a per-interface alternative to the global
2691 /proc/sys/net/ipv[46]/ip_forward. This setting is
2692 configurable in the IPForward= option, which defaults to
2693 "no". This means if networkd is used for an interface it is
2694 no longer sufficient to set the global sysctl option to turn
2695 on IP forwarding! Instead, the .network file option
2696 IPForward= needs to be turned on! Note that the
2697 implementation of this behaviour was broken in v219 and has
2700 * Many bonding and vxlan options are now configurable in
2703 * systemd-nspawn gained a new --property= setting to set unit
2704 properties for the container scope. This is useful for
2705 setting resource parameters (e.g. "CPUShares=500") on
2706 containers started from the command line.
2708 * systemd-nspawn gained a new --private-users= switch to make
2709 use of user namespacing available on recent Linux kernels.
2711 * systemd-nspawn may now be called as part of a shell pipeline
2712 in which case the pipes used for stdin and stdout are passed
2713 directly to the process invoked in the container, without
2714 indirection via a pseudo tty.
2716 * systemd-nspawn gained a new switch to control the UNIX
2717 signal to use when killing the init process of the container
2720 * systemd-nspawn gained a new --overlay= switch for mounting
2721 overlay file systems into the container using the new kernel
2724 * When a container image is imported via systemd-importd and
2725 the host file system is not btrfs, a loopback block device
2726 file is created in /var/lib/machines.raw with a btrfs file
2727 system inside. It is then mounted to /var/lib/machines to
2728 enable btrfs features for container management. The loopback
2729 file and btrfs file system is grown as needed when container
2730 images are imported via systemd-importd.
2732 * systemd-machined/systemd-importd gained support for btrfs
2733 quota, to enforce container disk space limits on disk. This
2734 is exposed in "machinectl set-limit".
2736 * systemd-importd now can import containers from local .tar,
2737 .raw and .qcow2 images, and export them to .tar and .raw. It
2738 can also import dkr v2 images now from the network (on top
2741 * systemd-importd gained support for verifying downloaded
2742 images with gpg2 (previously only gpg1 was supported).
2744 * systemd-machined, systemd-logind, systemd: most bus calls
2745 are now accessible to unprivileged processes via
2746 PolicyKit. Also, systemd-logind will now allow users to kill
2747 their own sessions without further privileges or
2750 * systemd-shutdownd has been removed. This service was
2751 previously responsible for implementing scheduled shutdowns
2752 as exposed in /usr/bin/shutdown's time parameter. This
2753 functionality has now been moved into systemd-logind and is
2754 accessible via a bus interface.
2756 * "systemctl reboot" gained a new switch --firmware-setup that
2757 can be used to reboot into the EFI firmware setup, if that
2758 is available. systemd-logind now exposes an API on the bus
2759 to trigger such reboots, in case graphical desktop UIs want
2760 to cover this functionality.
2762 * "systemctl enable", "systemctl disable" and "systemctl mask"
2763 now support a new "--now" switch. If specified the units
2764 that are enabled will also be started, and the ones
2765 disabled/masked also stopped.
2767 * The Gummiboot EFI boot loader tool has been merged into
2768 systemd, and renamed to "systemd-boot". The bootctl tool has been
2769 updated to support systemd-boot.
2771 * An EFI kernel stub has been added that may be used to create
2772 kernel EFI binaries that contain not only the actual kernel,
2773 but also an initrd, boot splash, command line and OS release
2774 information. This combined binary can then be signed as a
2775 single image, so that the firmware can verify it all in one
2776 step. systemd-boot has special support for EFI binaries created
2777 like this and can extract OS release information from them
2778 and show them in the boot menu. This functionality is useful
2779 to implement cryptographically verified boot schemes.
2781 * Optional support has been added to systemd-fsck to pass
2782 fsck's progress report to an AF_UNIX socket in the file
2785 * udev will no longer create device symlinks for all block
2786 devices by default. A blacklist for excluding special block
2787 devices from this logic has been turned into a whitelist
2788 that requires picking block devices explicitly that require
2791 * A new (currently still internal) API sd-device.h has been
2792 added to libsystemd. This modernized API is supposed to
2793 replace libudev eventually. In fact, already much of libudev
2794 is now just a wrapper around sd-device.h.
2796 * A new hwdb database for storing metadata about pointing
2797 stick devices has been added.
2799 * systemd-tmpfiles gained support for setting file attributes
2800 similar to the "chattr" tool with new 'h' and 'H' lines.
2802 * systemd-journald will no longer unconditionally set the
2803 btrfs NOCOW flag on new journal files. This is instead done
2804 with tmpfiles snippet using the new 'h' line type. This
2805 allows easy disabling of this logic, by masking the
2806 journal-nocow.conf tmpfiles file.
2808 * systemd-journald will now translate audit message types to
2809 human readable identifiers when writing them to the
2810 journal. This should improve readability of audit messages.
2812 * The LUKS logic gained support for the offset= and skip=
2813 options in /etc/crypttab, as previously implemented by
2816 * /usr/lib/os-release gained a new optional field VARIANT= for
2817 distributions that support multiple variants (such as a
2818 desktop edition, a server edition, ...)
2820 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
2821 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
2822 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
2823 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
2824 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
2825 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
2826 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
2827 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
2828 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
2829 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
2830 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
2831 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
2832 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
2833 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
2834 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
2835 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
2836 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
2837 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
2838 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
2839 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
2840 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
2841 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
2842 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
2843 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
2844 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
2845 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
2846 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
2848 — Berlin, 2015-05-22
2852 * Introduce a new API "sd-hwdb.h" for querying the hardware
2853 metadata database. With this minimal interface one can query
2854 and enumerate the udev hwdb, decoupled from the old libudev
2855 library. libudev's interface for this is now only a wrapper
2856 around sd-hwdb. A new tool systemd-hwdb has been added to
2857 interface with and update the database.
2859 * When any of systemd's tools copies files (for example due to
2860 tmpfiles' C lines) a btrfs reflink will attempted first,
2861 before bytewise copying is done.
2863 * systemd-nspawn gained a new --ephemeral switch. When
2864 specified a btrfs snapshot is taken of the container's root
2865 directory, and immediately removed when the container
2866 terminates again. Thus, a container can be started whose
2867 changes never alter the container's root directory, and are
2868 lost on container termination. This switch can also be used
2869 for starting a container off the root file system of the
2870 host without affecting the host OS. This switch is only
2871 available on btrfs file systems.
2873 * systemd-nspawn gained a new --template= switch. It takes the
2874 path to a container tree to use as template for the tree
2875 specified via --directory=, should that directory be
2876 missing. This allows instantiating containers dynamically,
2877 on first run. This switch is only available on btrfs file
2880 * When a .mount unit refers to a mount point on which multiple
2881 mounts are stacked, and the .mount unit is stopped all of
2882 the stacked mount points will now be unmounted until no
2883 mount point remains.
2885 * systemd now has an explicit notion of supported and
2886 unsupported unit types. Jobs enqueued for unsupported unit
2887 types will now fail with an "unsupported" error code. More
2888 specifically .swap, .automount and .device units are not
2889 supported in containers, .busname units are not supported on
2890 non-kdbus systems. .swap and .automount are also not
2891 supported if their respective kernel compile time options
2894 * machinectl gained support for two new "copy-from" and
2895 "copy-to" commands for copying files from a running
2896 container to the host or vice versa.
2898 * machinectl gained support for a new "bind" command to bind
2899 mount host directories into local containers. This is
2900 currently only supported for nspawn containers.
2902 * networkd gained support for configuring bridge forwarding
2903 database entries (fdb) from .network files.
2905 * A new tiny daemon "systemd-importd" has been added that can
2906 download container images in tar, raw, qcow2 or dkr formats,
2907 and make them available locally in /var/lib/machines, so
2908 that they can run as nspawn containers. The daemon can GPG
2909 verify the downloads (not supported for dkr, since it has no
2910 provisions for verifying downloads). It will transparently
2911 decompress bz2, xz, gzip compressed downloads if necessary,
2912 and restore sparse files on disk. The daemon uses privilege
2913 separation to ensure the actual download logic runs with
2914 fewer privileges than the daemon itself. machinectl has
2915 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
2916 make the functionality of importd available to the
2917 user. With this in place the Fedora and Ubuntu "Cloud"
2918 images can be downloaded and booted as containers unmodified
2919 (the Fedora images lack the appropriate GPG signature files
2920 currently, so they cannot be verified, but this will change
2921 soon, hopefully). Note that downloading images is currently
2922 only fully supported on btrfs.
2924 * machinectl is now able to list container images found in
2925 /var/lib/machines, along with some metadata about sizes of
2926 disk and similar. If the directory is located on btrfs and
2927 quota is enabled, this includes quota display. A new command
2928 "image-status" has been added that shows additional
2929 information about images.
2931 * machinectl is now able to clone container images
2932 efficiently, if the underlying file system (btrfs) supports
2933 it, with the new "machinectl clone" command. It also
2934 gained commands for renaming and removing images, as well as
2935 marking them read-only or read-write (supported also on
2936 legacy file systems).
2938 * networkd gained support for collecting LLDP network
2939 announcements, from hardware that supports this. This is
2940 shown in networkctl output.
2942 * systemd-run gained support for a new -t (--pty) switch for
2943 invoking a binary on a pty whose input and output is
2944 connected to the invoking terminal. This allows executing
2945 processes as system services while interactively
2946 communicating with them via the terminal. Most interestingly
2947 this is supported across container boundaries. Invoking
2948 "systemd-run -t /bin/bash" is an alternative to running a
2949 full login session, the difference being that the former
2950 will not register a session, nor go through the PAM session
2953 * tmpfiles gained support for a new "v" line type for creating
2954 btrfs subvolumes. If the underlying file system is a legacy
2955 file system, this automatically degrades to creating a
2956 normal directory. Among others /var/lib/machines is now
2957 created like this at boot, should it be missing.
2959 * The directory /var/lib/containers/ has been deprecated and
2960 been replaced by /var/lib/machines. The term "machines" has
2961 been used in the systemd context as generic term for both
2962 VMs and containers, and hence appears more appropriate for
2963 this, as the directory can also contain raw images bootable
2966 * systemd-nspawn when invoked with -M but without --directory=
2967 or --image= is now capable of searching for the container
2968 root directory, subvolume or disk image automatically, in
2969 /var/lib/machines. systemd-nspawn@.service has been updated
2970 to make use of this, thus allowing it to be used for raw
2973 * A new machines.target unit has been introduced that is
2974 supposed to group all containers/VMs invoked as services on
2975 the system. systemd-nspawn@.service has been updated to
2976 integrate with that.
2978 * machinectl gained a new "start" command, for invoking a
2979 container as a service. "machinectl start foo" is mostly
2980 equivalent to "systemctl start systemd-nspawn@foo.service",
2981 but handles escaping in a nicer way.
2983 * systemd-nspawn will now mount most of the cgroupfs tree
2984 read-only into each container, with the exception of the
2985 container's own subtree in the name=systemd hierarchy.
2987 * journald now sets the special FS_NOCOW file flag for its
2988 journal files. This should improve performance on btrfs, by
2989 avoiding heavy fragmentation when journald's write-pattern
2990 is used on COW file systems. It degrades btrfs' data
2991 integrity guarantees for the files to the same levels as for
2992 ext3/ext4 however. This should be OK though as journald does
2993 its own data integrity checks and all its objects are
2994 checksummed on disk. Also, journald should handle btrfs disk
2995 full events a lot more gracefully now, by processing SIGBUS
2996 errors, and not relying on fallocate() anymore.
2998 * When journald detects that journal files it is writing to
2999 have been deleted it will immediately start new journal
3002 * systemd now provides a way to store file descriptors
3003 per-service in PID 1. This is useful for daemons to ensure
3004 that fds they require are not lost during a daemon
3005 restart. The fds are passed to the daemon on the next
3006 invocation in the same way socket activation fds are
3007 passed. This is now used by journald to ensure that the
3008 various sockets connected to all the system's stdout/stderr
3009 are not lost when journald is restarted. File descriptors
3010 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
3011 an extension to sd_notify(). Note that a limit is enforced
3012 on the number of fds a service can store in PID 1, and it
3013 defaults to 0, so that no fds may be stored, unless this is
3014 explicitly turned on.
3016 * The default TERM variable to use for units connected to a
3017 terminal, when no other value is explicitly is set is now
3018 vt220 rather than vt102. This should be fairly safe still,
3019 but allows PgUp/PgDn work.
3021 * The /etc/crypttab option header= as known from Debian is now
3024 * "loginctl user-status" and "loginctl session-status" will
3025 now show the last 10 lines of log messages of the
3026 user/session following the status output. Similar,
3027 "machinectl status" will show the last 10 log lines
3028 associated with a virtual machine or container
3029 service. (Note that this is usually not the log messages
3030 done in the VM/container itself, but simply what the
3031 container manager logs. For nspawn this includes all console
3034 * "loginctl session-status" without further argument will now
3035 show the status of the session of the caller. Similar,
3036 "lock-session", "unlock-session", "activate",
3037 "enable-linger", "disable-linger" may now be called without
3038 session/user parameter in which case they apply to the
3039 caller's session/user.
3041 * An X11 session scriptlet is now shipped that uploads
3042 $DISPLAY and $XAUTHORITY into the environment of the systemd
3043 --user daemon if a session begins. This should improve
3044 compatibility with X11 enabled applications run as systemd
3047 * Generators are now subject to masking via /etc and /run, the
3048 same way as unit files.
3050 * networkd .network files gained support for configuring
3051 per-link IPv4/IPv6 packet forwarding as well as IPv4
3052 masquerading. This is by default turned on for veth links to
3053 containers, as registered by systemd-nspawn. This means that
3054 nspawn containers run with --network-veth will now get
3055 automatic routed access to the host's networks without any
3056 further configuration or setup, as long as networkd runs on
3059 * systemd-nspawn gained the --port= (-p) switch to expose TCP
3060 or UDP posts of a container on the host. With this in place
3061 it is possible to run containers with private veth links
3062 (--network-veth), and have their functionality exposed on
3063 the host as if their services were running directly on the
3066 * systemd-nspawn's --network-veth switch now gained a short
3067 version "-n", since with the changes above it is now truly
3068 useful out-of-the-box. The systemd-nspawn@.service has been
3069 updated to make use of it too by default.
3071 * systemd-nspawn will now maintain a per-image R/W lock, to
3072 ensure that the same image is not started more than once
3073 writable. (It's OK to run an image multiple times
3074 simultaneously in read-only mode.)
3076 * systemd-nspawn's --image= option is now capable of
3077 dissecting and booting MBR and GPT disk images that contain
3078 only a single active Linux partition. Previously it
3079 supported only GPT disk images with proper GPT type
3080 IDs. This allows running cloud images from major
3081 distributions directly with systemd-nspawn, without
3084 * In addition to collecting mouse dpi data in the udev
3085 hardware database, there's now support for collecting angle
3086 information for mouse scroll wheels. The database is
3087 supposed to guarantee similar scrolling behavior on mice
3088 that it knows about. There's also support for collecting
3089 information about Touchpad types.
3091 * udev's input_id built-in will now also collect touch screen
3092 dimension data and attach it to probed devices.
3094 * /etc/os-release gained support for a Distribution Privacy
3097 * networkd gained support for creating "ipvlan", "gretap",
3098 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
3100 * systemd-tmpfiles gained support for "a" lines for setting
3103 * systemd-nspawn will now mount /tmp in the container to
3104 tmpfs, automatically.
3106 * systemd now exposes the memory.usage_in_bytes cgroup
3107 attribute and shows it for each service in the "systemctl
3108 status" output, if available.
3110 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
3111 immediate reboot is triggered. This useful if shutdown is
3112 hung and is unable to complete, to expedite the
3113 operation. Note that this kind of reboot will still unmount
3114 all file systems, and hence should not result in fsck being
3117 * A .device unit for an optical block device will now be
3118 considered active only when a medium is in the drive. Also,
3119 mount units are now bound to their backing devices thus
3120 triggering automatic unmounting when devices become
3121 unavailable. With this in place systemd will now
3122 automatically unmount left-over mounts when a CD-ROM is
3123 ejected or an USB stick is yanked from the system.
3125 * networkd-wait-online now has support for waiting for
3126 specific interfaces only (with globbing), and for giving up
3127 after a configurable timeout.
3129 * networkd now exits when idle. It will be automatically
3130 restarted as soon as interfaces show up, are removed or
3131 change state. networkd will stay around as long as there is
3132 at least one DHCP state machine or similar around, that keep
3135 * networkd may now configure IPv6 link-local addressing in
3136 addition to IPv4 link-local addressing.
3138 * The IPv6 "token" for use in SLAAC may now be configured for
3139 each .network interface in networkd.
3141 * Routes configured with networkd may now be assigned a scope
3144 * networkd's [Match] sections now support globbing and lists
3145 of multiple space-separated matches per item.
3147 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
3148 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
3149 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
3150 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
3151 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
3152 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
3153 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
3154 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
3155 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
3156 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
3157 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
3158 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
3159 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
3160 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
3161 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
3162 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
3163 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
3164 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
3165 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
3166 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
3167 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
3168 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
3169 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
3170 Hoffmann, Zbigniew Jędrzejewski-Szmek
3172 — Berlin, 2015-02-16
3176 * When querying unit file enablement status (for example via
3177 "systemctl is-enabled"), a new state "indirect" is now known
3178 which indicates that a unit might not be enabled itself, but
3179 another unit listed in its Also= setting might be.
3181 * Similar to the various existing ConditionXYZ= settings for
3182 units, there are now matching AssertXYZ= settings. While
3183 failing conditions cause a unit to be skipped, but its job
3184 to succeed, failing assertions declared like this will cause
3185 a unit start operation and its job to fail.
3187 * hostnamed now knows a new chassis type "embedded".
3189 * systemctl gained a new "edit" command. When used on a unit
3190 file, this allows extending unit files with .d/ drop-in
3191 configuration snippets or editing the full file (after
3192 copying it from /usr/lib to /etc). This will invoke the
3193 user's editor (as configured with $EDITOR), and reload the
3194 modified configuration after editing.
3196 * "systemctl status" now shows the suggested enablement state
3197 for a unit, as declared in the (usually vendor-supplied)
3198 system preset files.
3200 * nss-myhostname will now resolve the single-label host name
3201 "gateway" to the locally configured default IP routing
3202 gateways, ordered by their metrics. This assigns a stable
3203 name to the used gateways, regardless which ones are
3204 currently configured. Note that the name will only be
3205 resolved after all other name sources (if nss-myhostname is
3206 configured properly) and should hence not negatively impact
3207 systems that use the single-label host name "gateway" in
3210 * systemd-inhibit now allows filtering by mode when listing
3213 * Scope and service units gained a new "Delegate" boolean
3214 property, which, when set, allows processes running inside the
3215 unit to further partition resources. This is primarily
3216 useful for systemd user instances as well as container
3219 * journald will now pick up audit messages directly from
3220 the kernel, and log them like any other log message. The
3221 audit fields are split up and fully indexed. This means that
3222 journalctl in many ways is now a (nicer!) alternative to
3223 ausearch, the traditional audit client. Note that this
3224 implements only a minimal audit client. If you want the
3225 special audit modes like reboot-on-log-overflow, please use
3226 the traditional auditd instead, which can be used in
3227 parallel to journald.
3229 * The ConditionSecurity= unit file option now understands the
3230 special string "audit" to check whether auditing is
3233 * journalctl gained two new commands --vacuum-size= and
3234 --vacuum-time= to delete old journal files until the
3235 remaining ones take up no more than the specified size on disk,
3236 or are not older than the specified time.
3238 * A new, native PPPoE library has been added to sd-network,
3239 systemd's library of light-weight networking protocols. This
3240 library will be used in a future version of networkd to
3241 enable PPPoE communication without an external pppd daemon.
3243 * The busctl tool now understands a new "capture" verb that
3244 works similar to "monitor", but writes a packet capture
3245 trace to STDOUT that can be redirected to a file which is
3246 compatible with libcap's capture file format. This can then
3247 be loaded in Wireshark and similar tools to inspect bus
3250 * The busctl tool now understands a new "tree" verb that shows
3251 the object trees of a specific service on the bus, or of all
3254 * The busctl tool now understands a new "introspect" verb that
3255 shows all interfaces and members of objects on the bus,
3256 including their signature and values. This is particularly
3257 useful to get more information about bus objects shown by
3258 the new "busctl tree" command.
3260 * The busctl tool now understands new verbs "call",
3261 "set-property" and "get-property" for invoking bus method
3262 calls, setting and getting bus object properties in a
3265 * busctl gained a new --augment-creds= argument that controls
3266 whether the tool shall augment credential information it
3267 gets from the bus with data from /proc, in a possibly
3270 * nspawn's --link-journal= switch gained two new values
3271 "try-guest" and "try-host" that work like "guest" and
3272 "host", but do not fail if the host has no persistent
3273 journalling enabled. -j is now equivalent to
3274 --link-journal=try-guest.
3276 * macvlan network devices created by nspawn will now have
3277 stable MAC addresses.
3279 * A new SmackProcessLabel= unit setting has been added, which
3280 controls the SMACK security label processes forked off by
3281 the respective unit shall use.
3283 * If compiled with --enable-xkbcommon, systemd-localed will
3284 verify x11 keymap settings by compiling the given keymap. It
3285 will spew out warnings if the compilation fails. This
3286 requires libxkbcommon to be installed.
3288 * When a coredump is collected, a larger number of metadata
3289 fields is now collected and included in the journal records
3290 created for it. More specifically, control group membership,
3291 environment variables, memory maps, working directory,
3292 chroot directory, /proc/$PID/status, and a list of open file
3293 descriptors is now stored in the log entry.
3295 * The udev hwdb now contains DPI information for mice. For
3298 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
3300 * All systemd programs that read standalone configuration
3301 files in /etc now also support a corresponding series of
3302 .conf.d configuration directories in /etc/, /run/,
3303 /usr/local/lib/, /usr/lib/, and (if configured with
3304 --enable-split-usr) /lib/. In particular, the following
3305 configuration files now have corresponding configuration
3306 directories: system.conf user.conf, logind.conf,
3307 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
3308 resolved.conf, timesyncd.conf, journal-remote.conf, and
3309 journal-upload.conf. Note that distributions should use the
3310 configuration directories in /usr/lib/; the directories in
3311 /etc/ are reserved for the system administrator.
3313 * systemd-rfkill will no longer take the rfkill device name
3314 into account when storing rfkill state on disk, as the name
3315 might be dynamically assigned and not stable. Instead, the
3316 ID_PATH udev variable combined with the rfkill type (wlan,
3317 bluetooth, ...) is used.
3319 * A new service systemd-machine-id-commit.service has been
3320 added. When used on systems where /etc is read-only during
3321 boot, and /etc/machine-id is not initialized (but an empty
3322 file), this service will copy the temporary machine ID
3323 created as replacement into /etc after the system is fully
3324 booted up. This is useful for systems that are freshly
3325 installed with a non-initialized machine ID, but should get
3326 a fixed machine ID for subsequent boots.
3328 * networkd's .netdev files now provide a large set of
3329 configuration parameters for VXLAN devices. Similarly, the
3330 bridge port cost parameter is now configurable in .network
3331 files. There's also new support for configuring IP source
3332 routing. networkd .link files gained support for a new
3333 OriginalName= match that is useful to match against the
3334 original interface name the kernel assigned. .network files
3335 may include MTU= and MACAddress= fields for altering the MTU
3336 and MAC address while being connected to a specific network
3339 * The LUKS logic gained supported for configuring
3340 UUID-specific key files. There's also new support for naming
3341 LUKS device from the kernel command line, using the new
3342 luks.name= argument.
3344 * Timer units may now be transiently created via the bus API
3345 (this was previously already available for scope and service
3346 units). In addition it is now possible to create multiple
3347 transient units at the same time with a single bus call. The
3348 "systemd-run" tool has been updated to make use of this for
3349 running commands on a specified time, in at(1)-style.
3351 * tmpfiles gained support for "t" lines, for assigning
3352 extended attributes to files. Among other uses this may be
3353 used to assign SMACK labels to files.
3355 Contributions from: Alin Rauta, Alison Chaiken, Andrej
3356 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
3357 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
3358 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
3359 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
3360 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
3361 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
3362 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
3363 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
3364 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
3365 Michael Biebl, Michael Chapman, Michael Marineau, Michal
3366 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
3367 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
3368 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
3369 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
3370 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
3371 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
3372 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
3374 — Berlin, 2014-12-10
3378 * journalctl gained the new options -t/--identifier= to match
3379 on the syslog identifier (aka "tag"), as well as --utc to
3380 show log timestamps in the UTC timezone. journalctl now also
3381 accepts -n/--lines=all to disable line capping in a pager.
3383 * journalctl gained a new switch, --flush, that synchronously
3384 flushes logs from /run/log/journal to /var/log/journal if
3385 persistent storage is enabled. systemd-journal-flush.service
3386 now waits until the operation is complete.
3388 * Services can notify the manager before they start a reload
3389 (by sending RELOADING=1) or shutdown (by sending
3390 STOPPING=1). This allows the manager to track and show the
3391 internal state of daemons and closes a race condition when
3392 the process is still running but has closed its D-Bus
3395 * Services with Type=oneshot do not have to have any ExecStart
3398 * User units are now loaded also from
3399 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
3400 /run/systemd/user directory that was already previously
3401 supported, but is under the control of the user.
3403 * Job timeouts (i.e. time-outs on the time a job that is
3404 queued stays in the run queue) can now optionally result in
3405 immediate reboot or power-off actions (JobTimeoutAction= and
3406 JobTimeoutRebootArgument=). This is useful on ".target"
3407 units, to limit the maximum time a target remains
3408 undispatched in the run queue, and to trigger an emergency
3409 operation in such a case. This is now used by default to
3410 turn off the system if boot-up (as defined by everything in
3411 basic.target) hangs and does not complete for at least
3412 15min. Also, if power-off or reboot hang for at least 30min
3413 an immediate power-off/reboot operation is triggered. This
3414 functionality is particularly useful to increase reliability
3415 on embedded devices, but also on laptops which might
3416 accidentally get powered on when carried in a backpack and
3417 whose boot stays stuck in a hard disk encryption passphrase
3420 * systemd-logind can be configured to also handle lid switch
3421 events even when the machine is docked or multiple displays
3422 are attached (HandleLidSwitchDocked= option).
3424 * A helper binary and a service have been added which can be
3425 used to resume from hibernation in the initramfs. A
3426 generator will parse the resume= option on the kernel
3427 command line to trigger resume.
3429 * A user console daemon systemd-consoled has been
3430 added. Currently, it is a preview, and will so far open a
3431 single terminal on each session of the user marked as
3432 Desktop=systemd-console.
3434 * Route metrics can be specified for DHCP routes added by
3437 * The SELinux context of socket-activated services can be set
3438 from the information provided by the networking stack
3439 (SELinuxContextFromNet= option).
3441 * Userspace firmware loading support has been removed and
3442 the minimum supported kernel version is thus bumped to 3.7.
3444 * Timeout for udev workers has been increased from 1 to 3
3445 minutes, but a warning will be printed after 1 minute to
3446 help diagnose kernel modules that take a long time to load.
3448 * Udev rules can now remove tags on devices with TAG-="foobar".
3450 * systemd's readahead implementation has been removed. In many
3451 circumstances it didn't give expected benefits even for
3452 rotational disk drives and was becoming less relevant in the
3453 age of SSDs. As none of the developers has been using
3454 rotating media anymore, and nobody stepped up to actively
3455 maintain this component of systemd it has now been removed.
3457 * Swap units can use Options= to specify discard options.
3458 Discard options specified for swaps in /etc/fstab are now
3461 * Docker containers are now detected as a separate type of
3464 * The Password Agent protocol gained support for queries where
3465 the user input is shown, useful e.g. for user names.
3466 systemd-ask-password gained a new --echo option to turn that
3469 * The default sysctl.d/ snippets will now set:
3471 net.core.default_qdisc = fq_codel
3473 This selects Fair Queuing Controlled Delay as the default
3474 queuing discipline for network interfaces. fq_codel helps
3475 fight the network bufferbloat problem. It is believed to be
3476 a good default with no tuning required for most workloads.
3477 Downstream distributions may override this choice. On 10Gbit
3478 servers that do not do forwarding, "fq" may perform better.
3479 Systems without a good clocksource should use "pfifo_fast".
3481 * If kdbus is enabled during build a new option BusPolicy= is
3482 available for service units, that allows locking all service
3483 processes into a stricter bus policy, in order to limit
3484 access to various bus services, or even hide most of them
3485 from the service's view entirely.
3487 * networkctl will now show the .network and .link file
3488 networkd has applied to a specific interface.
3490 * sd-login gained a new API call sd_session_get_desktop() to
3491 query which desktop environment has been selected for a
3494 * UNIX utmp support is now compile-time optional to support
3495 legacy-free systems.
3497 * systemctl gained two new commands "add-wants" and
3498 "add-requires" for pulling in units from specific targets
3501 * If the word "rescue" is specified on the kernel command line
3502 the system will now boot into rescue mode (aka
3503 rescue.target), which was previously available only by
3504 specifying "1" or "systemd.unit=rescue.target" on the kernel
3505 command line. This new kernel command line option nicely
3506 mirrors the already existing "emergency" kernel command line
3509 * New kernel command line options mount.usr=, mount.usrflags=,
3510 mount.usrfstype= have been added that match root=, rootflags=,
3511 rootfstype= but allow mounting a specific file system to
3514 * The $NOTIFY_SOCKET is now also passed to control processes of
3515 services, not only the main process.
3517 * This version reenables support for fsck's -l switch. This
3518 means at least version v2.25 of util-linux is required for
3519 operation, otherwise dead-locks on device nodes may
3520 occur. Again: you need to update util-linux to at least
3521 v2.25 when updating systemd to v217.
3523 * The "multi-seat-x" tool has been removed from systemd, as
3524 its functionality has been integrated into X servers 1.16,
3525 and the tool is hence redundant. It is recommended to update
3526 display managers invoking this tool to simply invoke X
3527 directly from now on, again.
3529 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
3530 message flag has been added for all of systemd's PolicyKit
3531 authenticated method calls has been added. In particular
3532 this now allows optional interactive authorization via
3533 PolicyKit for many of PID1's privileged operations such as
3534 unit file enabling and disabling.
3536 * "udevadm hwdb --update" learnt a new switch "--usr" for
3537 placing the rebuilt hardware database in /usr instead of
3538 /etc. When used only hardware database entries stored in
3539 /usr will be used, and any user database entries in /etc are
3540 ignored. This functionality is useful for vendors to ship a
3541 pre-built database on systems where local configuration is
3542 unnecessary or unlikely.
3544 * Calendar time specifications in .timer units now also
3545 understand the strings "semi-annually", "quarterly" and
3546 "minutely" as shortcuts (in addition to the preexisting
3547 "anually", "hourly", ...).
3549 * systemd-tmpfiles will now correctly create files in /dev
3550 at boot which are marked for creation only at boot. It is
3551 recommended to always create static device nodes with 'c!'
3552 and 'b!', so that they are created only at boot and not
3553 overwritten at runtime.
3555 * When the watchdog logic is used for a service (WatchdogSec=)
3556 and the watchdog timeout is hit the service will now be
3557 terminated with SIGABRT (instead of just SIGTERM), in order
3558 to make sure a proper coredump and backtrace is
3559 generated. This ensures that hanging services will result in
3560 similar coredump/backtrace behaviour as services that hit a
3563 Contributions from: Andreas Henriksson, Andrei Borzenkov,
3564 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
3565 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
3566 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
3567 Herrmann, David Sommerseth, David Strauss, Emil Renner
3568 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
3569 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
3570 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
3571 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
3572 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
3573 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
3574 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
3575 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
3576 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
3577 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
3578 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
3579 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
3580 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
3581 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
3582 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
3583 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
3586 — Berlin, 2014-10-28
3590 * timedated no longer reads NTP implementation unit names from
3591 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
3592 implementations should add a
3594 Conflicts=systemd-timesyncd.service
3596 to their unit files to take over and replace systemd's NTP
3597 default functionality.
3599 * systemd-sysusers gained a new line type "r" for configuring
3600 which UID/GID ranges to allocate system users/groups
3601 from. Lines of type "u" may now add an additional column
3602 that specifies the home directory for the system user to be
3603 created. Also, systemd-sysusers may now optionally read user
3604 information from STDIN instead of a file. This is useful for
3605 invoking it from RPM preinst scriptlets that need to create
3606 users before the first RPM file is installed since these
3607 files might need to be owned by them. A new
3608 %sysusers_create_inline RPM macro has been introduced to do
3609 just that. systemd-sysusers now updates the shadow files as
3610 well as the user/group databases, which should enhance
3611 compatibility with certain tools like grpck.
3613 * A number of bus APIs of PID 1 now optionally consult
3614 PolicyKit to permit access for otherwise unprivileged
3615 clients under certain conditions. Note that this currently
3616 doesn't support interactive authentication yet, but this is
3617 expected to be added eventually, too.
3619 * /etc/machine-info now has new fields for configuring the
3620 deployment environment of the machine, as well as the
3621 location of the machine. hostnamectl has been updated with
3622 new command to update these fields.
3624 * systemd-timesyncd has been updated to automatically acquire
3625 NTP server information from systemd-networkd, which might
3626 have been discovered via DHCP.
3628 * systemd-resolved now includes a caching DNS stub resolver
3629 and a complete LLMNR name resolution implementation. A new
3630 NSS module "nss-resolve" has been added which can be used
3631 instead of glibc's own "nss-dns" to resolve hostnames via
3632 systemd-resolved. Hostnames, addresses and arbitrary RRs may
3633 be resolved via systemd-resolved D-Bus APIs. In contrast to
3634 the glibc internal resolver systemd-resolved is aware of
3635 multi-homed system, and keeps DNS server and caches separate
3636 and per-interface. Queries are sent simultaneously on all
3637 interfaces that have DNS servers configured, in order to
3638 properly handle VPNs and local LANs which might resolve
3639 separate sets of domain names. systemd-resolved may acquire
3640 DNS server information from systemd-networkd automatically,
3641 which in turn might have discovered them via DHCP. A tool
3642 "systemd-resolve-host" has been added that may be used to
3643 query the DNS logic in resolved. systemd-resolved implements
3644 IDNA and automatically uses IDNA or UTF-8 encoding depending
3645 on whether classic DNS or LLMNR is used as transport. In the
3646 next releases we intend to add a DNSSEC and mDNS/DNS-SD
3647 implementation to systemd-resolved.
3649 * A new NSS module nss-mymachines has been added, that
3650 automatically resolves the names of all local registered
3651 containers to their respective IP addresses.
3653 * A new client tool "networkctl" for systemd-networkd has been
3654 added. It currently is entirely passive and will query
3655 networking configuration from udev, rtnetlink and networkd,
3656 and present it to the user in a very friendly
3657 way. Eventually, we hope to extend it to become a full
3658 control utility for networkd.
3660 * .socket units gained a new DeferAcceptSec= setting that
3661 controls the kernels' TCP_DEFER_ACCEPT sockopt for
3662 TCP. Similarly, support for controlling TCP keep-alive
3663 settings has been added (KeepAliveTimeSec=,
3664 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
3665 turning off Nagle's algorithm on TCP has been added
3668 * logind learned a new session type "web", for use in projects
3669 like Cockpit which register web clients as PAM sessions.
3671 * timer units with at least one OnCalendar= setting will now
3672 be started only after time-sync.target has been
3673 reached. This way they will not elapse before the system
3674 clock has been corrected by a local NTP client or
3675 similar. This is particular useful on RTC-less embedded
3676 machines, that come up with an invalid system clock.
3678 * systemd-nspawn's --network-veth= switch should now result in
3679 stable MAC addresses for both the outer and the inner side
3682 * systemd-nspawn gained a new --volatile= switch for running
3683 container instances with /etc or /var unpopulated.
3685 * The kdbus client code has been updated to use the new Linux
3686 3.17 memfd subsystem instead of the old kdbus-specific one.
3688 * systemd-networkd's DHCP client and server now support
3689 FORCERENEW. There are also new configuration options to
3690 configure the vendor client identifier and broadcast mode
3693 * systemd will no longer inform the kernel about the current
3694 timezone, as this is necessarily incorrect and racy as the
3695 kernel has no understanding of DST and similar
3696 concepts. This hence means FAT timestamps will be always
3697 considered UTC, similar to what Android is already
3698 doing. Also, when the RTC is configured to the local time
3699 (rather than UTC) systemd will never synchronize back to it,
3700 as this might confuse Windows at a later boot.
3702 * systemd-analyze gained a new command "verify" for offline
3703 validation of unit files.
3705 * systemd-networkd gained support for a couple of additional
3706 settings for bonding networking setups. Also, the metric for
3707 statically configured routes may now be configured. For
3708 network interfaces where this is appropriate the peer IP
3709 address may now be configured.
3711 * systemd-networkd's DHCP client will no longer request
3712 broadcasting by default, as this tripped up some networks.
3713 For hardware where broadcast is required the feature should
3714 be switched back on using RequestBroadcast=yes.
3716 * systemd-networkd will now set up IPv4LL addresses (when
3717 enabled) even if DHCP is configured successfully.
3719 * udev will now default to respect network device names given
3720 by the kernel when the kernel indicates that these are
3721 predictable. This behavior can be tweaked by changing
3722 NamePolicy= in the relevant .link file.
3724 * A new library systemd-terminal has been added that
3725 implements full TTY stream parsing and rendering. This
3726 library is supposed to be used later on for implementing a
3727 full userspace VT subsystem, replacing the current kernel
3730 * A new tool systemd-journal-upload has been added to push
3731 journal data to a remote system running
3732 systemd-journal-remote.
3734 * journald will no longer forward all local data to another
3735 running syslog daemon. This change has been made because
3736 rsyslog (which appears to be the most commonly used syslog
3737 implementation these days) no longer makes use of this, and
3738 instead pulls the data out of the journal on its own. Since
3739 forwarding the messages to a non-existent syslog server is
3740 more expensive than we assumed we have now turned this
3741 off. If you run a syslog server that is not a recent rsyslog
3742 version, you have to turn this option on again
3743 (ForwardToSyslog= in journald.conf).
3745 * journald now optionally supports the LZ4 compressor for
3746 larger journal fields. This compressor should perform much
3747 better than XZ which was the previous default.
3749 * machinectl now shows the IP addresses of local containers,
3750 if it knows them, plus the interface name of the container.
3752 * A new tool "systemd-escape" has been added that makes it
3753 easy to escape strings to build unit names and similar.
3755 * sd_notify() messages may now include a new ERRNO= field
3756 which is parsed and collected by systemd and shown among the
3757 "systemctl status" output for a service.
3759 * A new component "systemd-firstboot" has been added that
3760 queries the most basic systemd information (timezone,
3761 hostname, root password) interactively on first
3762 boot. Alternatively it may also be used to provision these
3763 things offline on OS images installed into directories.
3765 * The default sysctl.d/ snippets will now set
3767 net.ipv4.conf.default.promote_secondaries=1
3769 This has the benefit of no flushing secondary IP addresses
3770 when primary addresses are removed.
3772 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
3773 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
3774 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
3775 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
3776 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
3777 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
3778 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
3779 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
3780 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
3781 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
3782 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
3783 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
3784 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
3785 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
3786 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
3788 — Berlin, 2014-08-19
3792 * A new tool systemd-sysusers has been added. This tool
3793 creates system users and groups in /etc/passwd and
3794 /etc/group, based on static declarative system user/group
3795 definitions in /usr/lib/sysusers.d/. This is useful to
3796 enable factory resets and volatile systems that boot up with
3797 an empty /etc directory, and thus need system users and
3798 groups created during early boot. systemd now also ships
3799 with two default sysusers.d/ files for the most basic
3800 users and groups systemd and the core operating system
3803 * A new tmpfiles snippet has been added that rebuilds the
3804 essential files in /etc on boot, should they be missing.
3806 * A directive for ensuring automatic clean-up of
3807 /var/cache/man/ has been removed from the default
3808 configuration. This line should now be shipped by the man
3809 implementation. The necessary change has been made to the
3810 man-db implementation. Note that you need to update your man
3811 implementation to one that ships this line, otherwise no
3812 automatic clean-up of /var/cache/man will take place.
3814 * A new condition ConditionNeedsUpdate= has been added that
3815 may conditionalize services to only run when /etc or /var
3816 are "older" than the vendor operating system resources in
3817 /usr. This is useful for reconstructing or updating /etc
3818 after an offline update of /usr or a factory reset, on the
3819 next reboot. Services that want to run once after such an
3820 update or reset should use this condition and order
3821 themselves before the new systemd-update-done.service, which
3822 will mark the two directories as fully updated. A number of
3823 service files have been added making use of this, to rebuild
3824 the udev hardware database, the journald message catalog and
3825 dynamic loader cache (ldconfig). The systemd-sysusers tool
3826 described above also makes use of this now. With this in
3827 place it is now possible to start up a minimal operating
3828 system with /etc empty cleanly. For more information on the
3829 concepts involved see this recent blog story:
3831 http://0pointer.de/blog/projects/stateless.html
3833 * A new system group "input" has been introduced, and all
3834 input device nodes get this group assigned. This is useful
3835 for system-level software to get access to input devices. It
3836 complements what is already done for "audio" and "video".
3838 * systemd-networkd learnt minimal DHCPv4 server support in
3839 addition to the existing DHCPv4 client support. It also
3840 learnt DHCPv6 client and IPv6 Router Solicitation client
3841 support. The DHCPv4 client gained support for static routes
3842 passed in from the server. Note that the [DHCPv4] section
3843 known in older systemd-networkd versions has been renamed to
3844 [DHCP] and is now also used by the DHCPv6 client. Existing
3845 .network files using settings of this section should be
3846 updated, though compatibility is maintained. Optionally, the
3847 client hostname may now be sent to the DHCP server.
3849 * networkd gained support for vxlan virtual networks as well
3850 as tun/tap and dummy devices.
3852 * networkd gained support for automatic allocation of address
3853 ranges for interfaces from a system-wide pool of
3854 addresses. This is useful for dynamically managing a large
3855 number of interfaces with a single network configuration
3856 file. In particular this is useful to easily assign
3857 appropriate IP addresses to the veth links of a large number
3858 of nspawn instances.
3860 * RPM macros for processing sysusers, sysctl and binfmt
3861 drop-in snippets at package installation time have been
3864 * The /etc/os-release file should now be placed in
3865 /usr/lib/os-release. The old location is automatically
3866 created as symlink. /usr/lib is the more appropriate
3867 location of this file, since it shall actually describe the
3868 vendor operating system shipped in /usr, and not the
3869 configuration stored in /etc.
3871 * .mount units gained a new boolean SloppyOptions= setting
3872 that maps to mount(8)'s -s option which enables permissive
3873 parsing of unknown mount options.
3875 * tmpfiles learnt a new "L+" directive which creates a symlink
3876 but (unlike "L") deletes a pre-existing file first, should
3877 it already exist and not already be the correct
3878 symlink. Similarly, "b+", "c+" and "p+" directives have been
3879 added as well, which create block and character devices, as
3880 well as fifos in the filesystem, possibly removing any
3881 pre-existing files of different types.
3883 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
3884 'argument' field (which so far specified the source to
3885 symlink/copy the files from) is now optional. If omitted the
3886 same file os copied from /usr/share/factory/ suffixed by the
3887 full destination path. This is useful for populating /etc
3888 with essential files, by copying them from vendor defaults
3889 shipped in /usr/share/factory/etc.
3891 * A new command "systemctl preset-all" has been added that
3892 applies the service preset settings to all installed unit
3893 files. A new switch --preset-mode= has been added that
3894 controls whether only enable or only disable operations
3897 * A new command "systemctl is-system-running" has been added
3898 that allows checking the overall state of the system, for
3899 example whether it is fully up and running.
3901 * When the system boots up with an empty /etc, the equivalent
3902 to "systemctl preset-all" is executed during early boot, to
3903 make sure all default services are enabled after a factory
3906 * systemd now contains a minimal preset file that enables the
3907 most basic services systemd ships by default.
3909 * Unit files' [Install] section gained a new DefaultInstance=
3910 field for defining the default instance to create if a
3911 template unit is enabled with no instance specified.
3913 * A new passive target cryptsetup-pre.target has been added
3914 that may be used by services that need to make they run and
3915 finish before the first LUKS cryptographic device is set up.
3917 * The /dev/loop-control and /dev/btrfs-control device nodes
3918 are now owned by the "disk" group by default, opening up
3919 access to this group.
3921 * systemd-coredump will now automatically generate a
3922 stack trace of all core dumps taking place on the system,
3923 based on elfutils' libdw library. This stack trace is logged
3926 * systemd-coredump may now optionally store coredumps directly
3927 on disk (in /var/lib/systemd/coredump, possibly compressed),
3928 instead of storing them unconditionally in the journal. This
3929 mode is the new default. A new configuration file
3930 /etc/systemd/coredump.conf has been added to configure this
3931 and other parameters of systemd-coredump.
3933 * coredumpctl gained a new "info" verb to show details about a
3934 specific coredump. A new switch "-1" has also been added
3935 that makes sure to only show information about the most
3936 recent entry instead of all entries. Also, as the tool is
3937 generally useful now the "systemd-" prefix of the binary
3938 name has been removed. Distributions that want to maintain
3939 compatibility with the old name should add a symlink from
3940 the old name to the new name.
3942 * journald's SplitMode= now defaults to "uid". This makes sure
3943 that unprivileged users can access their own coredumps with
3944 coredumpctl without restrictions.
3946 * New kernel command line options "systemd.wants=" (for
3947 pulling an additional unit during boot), "systemd.mask="
3948 (for masking a specific unit for the boot), and
3949 "systemd.debug-shell" (for enabling the debug shell on tty9)
3950 have been added. This is implemented in the new generator
3951 "systemd-debug-generator".
3953 * systemd-nspawn will now by default filter a couple of
3954 syscalls for containers, among them those required for
3955 kernel module loading, direct x86 IO port access, swap
3956 management, and kexec. Most importantly though
3957 open_by_handle_at() is now prohibited for containers,
3958 closing a hole similar to a recently discussed vulnerability
3959 in docker regarding access to files on file hierarchies the
3960 container should normally not have access to. Note that, for
3961 nspawn, we generally make no security claims anyway (and
3962 this is explicitly documented in the man page), so this is
3963 just a fix for one of the most obvious problems.
3965 * A new man page file-hierarchy(7) has been added that
3966 contains a minimized, modernized version of the file system
3967 layout systemd expects, similar in style to the FHS
3968 specification or hier(5). A new tool systemd-path(1) has
3969 been added to query many of these paths for the local
3972 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
3973 longer done. Since the directory now has a per-user size
3974 limit, and is cleaned on logout this appears unnecessary,
3975 in particular since this now brings the lifecycle of this
3976 directory closer in line with how IPC objects are handled.
3978 * systemd.pc now exports a number of additional directories,
3979 including $libdir (which is useful to identify the library
3980 path for the primary architecture of the system), and a
3981 couple of drop-in directories.
3983 * udev's predictable network interface names now use the dev_port
3984 sysfs attribute, introduced in linux 3.15 instead of dev_id to
3985 distinguish between ports of the same PCI function. dev_id should
3986 only be used for ports using the same HW address, hence the need
3989 * machined has been updated to export the OS version of a
3990 container (read from /etc/os-release and
3991 /usr/lib/os-release) on the bus. This is now shown in
3992 "machinectl status" for a machine.
3994 * A new service setting RestartForceExitStatus= has been
3995 added. If configured to a set of exit signals or process
3996 return values, the service will be restarted when the main
3997 daemon process exits with any of them, regardless of the
4000 * systemctl's -H switch for connecting to remote systemd
4001 machines has been extended so that it may be used to
4002 directly connect to a specific container on the
4003 host. "systemctl -H root@foobar:waldi" will now connect as
4004 user "root" to host "foobar", and then proceed directly to
4005 the container named "waldi". Note that currently you have to
4006 authenticate as user "root" for this to work, as entering
4007 containers is a privileged operation.
4009 Contributions from: Andreas Henriksson, Benjamin Steinwender,
4010 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
4011 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
4012 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
4013 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
4014 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
4015 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
4016 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
4017 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
4018 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
4019 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
4020 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
4022 — Berlin, 2014-07-03
4026 * As an experimental feature, udev now tries to lock the
4027 disk device node (flock(LOCK_SH|LOCK_NB)) while it
4028 executes events for the disk or any of its partitions.
4029 Applications like partitioning programs can lock the
4030 disk device node (flock(LOCK_EX)) and claim temporary
4031 device ownership that way; udev will entirely skip all event
4032 handling for this disk and its partitions. If the disk
4033 was opened for writing, the close will trigger a partition
4034 table rescan in udev's "watch" facility, and if needed
4035 synthesize "change" events for the disk and all its partitions.
4036 This is now unconditionally enabled, and if it turns out to
4037 cause major problems, we might turn it on only for specific
4038 devices, or might need to disable it entirely. Device Mapper
4039 devices are excluded from this logic.
4041 * We temporarily dropped the "-l" switch for fsck invocations,
4042 since they collide with the flock() logic above. util-linux
4043 upstream has been changed already to avoid this conflict,
4044 and we will readd "-l" as soon as util-linux with this
4045 change has been released.
4047 * The dependency on libattr has been removed. Since a long
4048 time, the extended attribute calls have moved to glibc, and
4049 libattr is thus unnecessary.
4051 * Virtualization detection works without privileges now. This
4052 means the systemd-detect-virt binary no longer requires
4053 CAP_SYS_PTRACE file capabilities, and our daemons can run
4054 with fewer privileges.
4056 * systemd-networkd now runs under its own "systemd-network"
4057 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
4058 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
4059 loses the ability to write to files owned by root this way.
4061 * Similarly, systemd-resolved now runs under its own
4062 "systemd-resolve" user with no capabilities remaining.
4064 * Similarly, systemd-bus-proxyd now runs under its own
4065 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
4067 * systemd-networkd gained support for setting up "veth"
4068 virtual Ethernet devices for container connectivity, as well
4069 as GRE and VTI tunnels.
4071 * systemd-networkd will no longer automatically attempt to
4072 manually load kernel modules necessary for certain tunnel
4073 transports. Instead, it is assumed the kernel loads them
4074 automatically when required. This only works correctly on
4075 very new kernels. On older kernels, please consider adding
4076 the kernel modules to /etc/modules-load.d/ as a work-around.
4078 * The resolv.conf file systemd-resolved generates has been
4079 moved to /run/systemd/resolve/. If you have a symlink from
4080 /etc/resolv.conf, it might be necessary to correct it.
4082 * Two new service settings, ProtectHome= and ProtectSystem=,
4083 have been added. When enabled, they will make the user data
4084 (such as /home) inaccessible or read-only and the system
4085 (such as /usr) read-only, for specific services. This allows
4086 very light-weight per-service sandboxing to avoid
4087 modifications of user data or system files from
4088 services. These two new switches have been enabled for all
4089 of systemd's long-running services, where appropriate.
4091 * Socket units gained new SocketUser= and SocketGroup=
4092 settings to set the owner user and group of AF_UNIX sockets
4093 and FIFOs in the file system.
4095 * Socket units gained a new RemoveOnStop= setting. If enabled,
4096 all FIFOS and sockets in the file system will be removed
4097 when the specific socket unit is stopped.
4099 * Socket units gained a new Symlinks= setting. It takes a list
4100 of symlinks to create to file system sockets or FIFOs
4101 created by the specific Unix sockets. This is useful to
4102 manage symlinks to socket nodes with the same life-cycle as
4105 * The /dev/log socket and /dev/initctl FIFO have been moved to
4106 /run, and have been replaced by symlinks. This allows
4107 connecting to these facilities even if PrivateDevices=yes is
4108 used for a service (which makes /dev/log itself unavailable,
4109 but /run is left). This also has the benefit of ensuring
4110 that /dev only contains device nodes, directories and
4111 symlinks, and nothing else.
4113 * sd-daemon gained two new calls sd_pid_notify() and
4114 sd_pid_notifyf(). They are similar to sd_notify() and
4115 sd_notifyf(), but allow overriding of the source PID of
4116 notification messages if permissions permit this. This is
4117 useful to send notify messages on behalf of a different
4118 process (for example, the parent process). The
4119 systemd-notify tool has been updated to make use of this
4120 when sending messages (so that notification messages now
4121 originate from the shell script invoking systemd-notify and
4122 not the systemd-notify process itself. This should minimize
4123 a race where systemd fails to associate notification
4124 messages to services when the originating process already
4127 * A new "on-abnormal" setting for Restart= has been added. If
4128 set, it will result in automatic restarts on all "abnormal"
4129 reasons for a process to exit, which includes unclean
4130 signals, core dumps, timeouts and watchdog timeouts, but
4131 does not include clean and unclean exit codes or clean
4132 signals. Restart=on-abnormal is an alternative for
4133 Restart=on-failure for services that shall be able to
4134 terminate and avoid restarts on certain errors, by
4135 indicating so with an unclean exit code. Restart=on-failure
4136 or Restart=on-abnormal is now the recommended setting for
4137 all long-running services.
4139 * If the InaccessibleDirectories= service setting points to a
4140 mount point (or if there are any submounts contained within
4141 it), it is now attempted to completely unmount it, to make
4142 the file systems truly unavailable for the respective
4145 * The ReadOnlyDirectories= service setting and
4146 systemd-nspawn's --read-only parameter are now recursively
4147 applied to all submounts, too.
4149 * Mount units may now be created transiently via the bus APIs.
4151 * The support for SysV and LSB init scripts has been removed
4152 from the systemd daemon itself. Instead, it is now
4153 implemented as a generator that creates native systemd units
4154 from these scripts when needed. This enables us to remove a
4155 substantial amount of legacy code from PID 1, following the
4156 fact that many distributions only ship a very small number
4157 of LSB/SysV init scripts nowadays.
4159 * Privileged Xen (dom0) domains are not considered
4160 virtualization anymore by the virtualization detection
4161 logic. After all, they generally have unrestricted access to
4162 the hardware and usually are used to manage the unprivileged
4165 * systemd-tmpfiles gained a new "C" line type, for copying
4166 files or entire directories.
4168 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
4169 lines. So far, they have been non-globbing versions of the
4170 latter, and have thus been redundant. In future, it is
4171 recommended to only use "z". "m" has hence been removed
4172 from the documentation, even though it stays supported.
4174 * A tmpfiles snippet to recreate the most basic structure in
4175 /var has been added. This is enough to create the /var/run →
4176 /run symlink and create a couple of structural
4177 directories. This allows systems to boot up with an empty or
4178 volatile /var. Of course, while with this change, the core OS
4179 now is capable with dealing with a volatile /var, not all
4180 user services are ready for it. However, we hope that sooner
4181 or later, many service daemons will be changed upstream so
4182 that they are able to automatically create their necessary
4183 directories in /var at boot, should they be missing. This is
4184 the first step to allow state-less systems that only require
4185 the vendor image for /usr to boot.
4187 * systemd-nspawn has gained a new --tmpfs= switch to mount an
4188 empty tmpfs instance to a specific directory. This is
4189 particularly useful for making use of the automatic
4190 reconstruction of /var (see above), by passing --tmpfs=/var.
4192 * Access modes specified in tmpfiles snippets may now be
4193 prefixed with "~", which indicates that they shall be masked
4194 by whether the existing file or directory is currently
4195 writable, readable or executable at all. Also, if specified,
4196 the sgid/suid/sticky bits will be masked for all
4199 * A new passive target unit "network-pre.target" has been
4200 added which is useful for services that shall run before any
4201 network is configured, for example firewall scripts.
4203 * The "floppy" group that previously owned the /dev/fd*
4204 devices is no longer used. The "disk" group is now used
4205 instead. Distributions should probably deprecate usage of
4208 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
4209 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
4210 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
4211 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
4212 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
4213 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
4216 — Berlin, 2014-06-11
4220 * A new "systemd-timesyncd" daemon has been added for
4221 synchronizing the system clock across the network. It
4222 implements an SNTP client. In contrast to NTP
4223 implementations such as chrony or the NTP reference server,
4224 this only implements a client side, and does not bother with
4225 the full NTP complexity, focusing only on querying time from
4226 one remote server and synchronizing the local clock to
4227 it. Unless you intend to serve NTP to networked clients or
4228 want to connect to local hardware clocks, this simple NTP
4229 client should be more than appropriate for most
4230 installations. The daemon runs with minimal privileges, and
4231 has been hooked up with networkd to only operate when
4232 network connectivity is available. The daemon saves the
4233 current clock to disk every time a new NTP sync has been
4234 acquired, and uses this to possibly correct the system clock
4235 early at bootup, in order to accommodate for systems that
4236 lack an RTC such as the Raspberry Pi and embedded devices,
4237 and to make sure that time monotonically progresses on these
4238 systems, even if it is not always correct. To make use of
4239 this daemon, a new system user and group "systemd-timesync"
4240 needs to be created on installation of systemd.
4242 * The queue "seqnum" interface of libudev has been disabled, as
4243 it was generally incompatible with device namespacing as
4244 sequence numbers of devices go "missing" if the devices are
4245 part of a different namespace.
4247 * "systemctl list-timers" and "systemctl list-sockets" gained
4248 a --recursive switch for showing units of these types also
4249 for all local containers, similar in style to the already
4250 supported --recursive switch for "systemctl list-units".
4252 * A new RebootArgument= setting has been added for service
4253 units, which may be used to specify a kernel reboot argument
4254 to use when triggering reboots with StartLimitAction=.
4256 * A new FailureAction= setting has been added for service
4257 units which may be used to specify an operation to trigger
4258 when a service fails. This works similarly to
4259 StartLimitAction=, but unlike it, controls what is done
4260 immediately rather than only after several attempts to
4261 restart the service in question.
4263 * hostnamed got updated to also expose the kernel name,
4264 release, and version on the bus. This is useful for
4265 executing commands like hostnamectl with the -H switch.
4266 systemd-analyze makes use of this to properly display
4267 details when running non-locally.
4269 * The bootchart tool can now show cgroup information in the
4270 graphs it generates.
4272 * The CFS CPU quota cgroup attribute is now exposed for
4273 services. The new CPUQuota= switch has been added for this
4274 which takes a percentage value. Setting this will have the
4275 result that a service may never get more CPU time than the
4276 specified percentage, even if the machine is otherwise idle.
4278 * systemd-networkd learned IPIP and SIT tunnel support.
4280 * LSB init scripts exposing a dependency on $network will now
4281 get a dependency on network-online.target rather than simply
4282 network.target. This should bring LSB handling closer to
4283 what it was on SysV systems.
4285 * A new fsck.repair= kernel option has been added to control
4286 how fsck shall deal with unclean file systems at boot.
4288 * The (.ini) configuration file parser will now silently
4289 ignore sections whose name begins with "X-". This may be
4290 used to maintain application-specific extension sections in unit
4293 * machined gained a new API to query the IP addresses of
4294 registered containers. "machinectl status" has been updated
4295 to show these addresses in its output.
4297 * A new call sd_uid_get_display() has been added to the
4298 sd-login APIs for querying the "primary" session of a
4299 user. The "primary" session of the user is elected from the
4300 user's sessions and generally a graphical session is
4301 preferred over a text one.
4303 * A minimal systemd-resolved daemon has been added. It
4304 currently simply acts as a companion to systemd-networkd and
4305 manages resolv.conf based on per-interface DNS
4306 configuration, possibly supplied via DHCP. In the long run
4307 we hope to extend this into a local DNSSEC enabled DNS and
4310 * The systemd-networkd-wait-online tool is now enabled by
4311 default. It will delay network-online.target until a network
4312 connection has been configured. The tool primarily integrates
4313 with networkd, but will also make a best effort to make sense
4314 of network configuration performed in some other way.
4316 * Two new service options StartupCPUShares= and
4317 StartupBlockIOWeight= have been added that work similarly to
4318 CPUShares= and BlockIOWeight= however only apply during
4319 system startup. This is useful to prioritize certain services
4320 differently during bootup than during normal runtime.
4322 * hostnamed has been changed to prefer the statically
4323 configured hostname in /etc/hostname (unless set to
4324 'localhost' or empty) over any dynamic one supplied by
4325 dhcp. With this change, the rules for picking the hostname
4326 match more closely the rules of other configuration settings
4327 where the local administrator's configuration in /etc always
4328 overrides any other settings.
4330 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
4331 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
4332 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
4333 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
4334 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
4335 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
4336 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
4337 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
4338 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
4339 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
4340 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
4341 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
4342 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
4343 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
4344 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
4345 Lindskog, WaLyong Cho, Will Woods, Zbigniew
4348 — Beijing, 2014-05-28
4352 * When restoring the screen brightness at boot, stay away from
4353 the darkest setting or from the lowest 5% of the available
4354 range, depending on which is the larger value of both. This
4355 should effectively protect the user from rebooting into a
4356 black screen, should the brightness have been set to minimum
4359 * sd-login gained a new sd_machine_get_class() call to
4360 determine the class ("vm" or "container") of a machine
4361 registered with machined.
4363 * sd-login gained new calls
4364 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
4365 to query the identity of the peer of a local AF_UNIX
4366 connection. They operate similarly to their sd_pid_get_xyz()
4369 * PID 1 will now maintain a system-wide system state engine
4370 with the states "starting", "running", "degraded",
4371 "maintenance", "stopping". These states are bound to system
4372 startup, normal runtime, runtime with at least one failed
4373 service, rescue/emergency mode and system shutdown. This
4374 state is shown in the "systemctl status" output when no unit
4375 name is passed. It is useful to determine system state, in
4376 particularly when doing so for many systems or containers at
4379 * A new command "list-machines" has been added to "systemctl"
4380 that lists all local OS containers and shows their system
4381 state (see above), if systemd runs inside of them.
4383 * systemctl gained a new "-r" switch to recursively enumerate
4384 units on all local containers, when used with the
4385 "list-unit" command (which is the default one that is
4386 executed when no parameters are specified).
4388 * The GPT automatic partition discovery logic will now honour
4389 two GPT partition flags: one may be set on a partition to
4390 cause it to be mounted read-only, and the other may be set
4391 on a partition to ignore it during automatic discovery.
4393 * Two new GPT type UUIDs have been added for automatic root
4394 partition discovery, for 32-bit and 64-bit ARM. This is not
4395 particularly useful for discovering the root directory on
4396 these architectures during bare-metal boots (since UEFI is
4397 not common there), but still very useful to allow booting of
4398 ARM disk images in nspawn with the -i option.
4400 * MAC addresses of interfaces created with nspawn's
4401 --network-interface= switch will now be generated from the
4402 machine name, and thus be stable between multiple invocations
4405 * logind will now automatically remove all IPC objects owned
4406 by a user if she or he fully logs out. This makes sure that
4407 users who are logged out cannot continue to consume IPC
4408 resources. This covers SysV memory, semaphores and message
4409 queues as well as POSIX shared memory and message
4410 queues. Traditionally, SysV and POSIX IPC had no life-cycle
4411 limits. With this functionality, that is corrected. This may
4412 be turned off by using the RemoveIPC= switch of logind.conf.
4414 * The systemd-machine-id-setup and tmpfiles tools gained a
4415 --root= switch to operate on a specific root directory,
4418 * journald can now forward logged messages to the TTYs of all
4419 logged in users ("wall"). This is the default for all
4420 emergency messages now.
4422 * A new tool systemd-journal-remote has been added to stream
4423 journal log messages across the network.
4425 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
4426 controller trees are mounted into it. Note that the
4427 directories mounted beneath it are not read-only. This is a
4428 security measure and is particularly useful because glibc
4429 actually includes a search logic to pick any tmpfs it can
4430 find to implement shm_open() if /dev/shm is not available
4431 (which it might very well be in namespaced setups).
4433 * machinectl gained a new "poweroff" command to cleanly power
4434 down a local OS container.
4436 * The PrivateDevices= unit file setting will now also drop the
4437 CAP_MKNOD capability from the capability bound set, and
4438 imply DevicePolicy=closed.
4440 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
4441 comprehensively on all long-running systemd services where
4442 this is appropriate.
4444 * systemd-udevd will now run in a disassociated mount
4445 namespace. To mount directories from udev rules, make sure to
4446 pull in mount units via SYSTEMD_WANTS properties.
4448 * The kdbus support gained support for uploading policy into
4449 the kernel. sd-bus gained support for creating "monitoring"
4450 connections that can eavesdrop into all bus communication
4451 for debugging purposes.
4453 * Timestamps may now be specified in seconds since the UNIX
4454 epoch Jan 1st, 1970 by specifying "@" followed by the value
4457 * Native tcpwrap support in systemd has been removed. tcpwrap
4458 is old code, not really maintained anymore and has serious
4459 shortcomings, and better options such as firewalls
4460 exist. For setups that require tcpwrap usage, please
4461 consider invoking your socket-activated service via tcpd,
4462 like on traditional inetd.
4464 * A new system.conf configuration option
4465 DefaultTimerAccuracySec= has been added that controls the
4466 default AccuracySec= setting of .timer units.
4468 * Timer units gained a new WakeSystem= switch. If enabled,
4469 timers configured this way will cause the system to resume
4470 from system suspend (if the system supports that, which most
4473 * Timer units gained a new Persistent= switch. If enabled,
4474 timers configured this way will save to disk when they have
4475 been last triggered. This information is then used on next
4476 reboot to possible execute overdue timer events, that
4477 could not take place because the system was powered off.
4478 This enables simple anacron-like behaviour for timer units.
4480 * systemctl's "list-timers" will now also list the time a
4481 timer unit was last triggered in addition to the next time
4482 it will be triggered.
4484 * systemd-networkd will now assign predictable IPv4LL
4485 addresses to its local interfaces.
4487 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
4488 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
4489 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
4490 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
4491 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
4492 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
4493 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
4494 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
4497 — Berlin, 2014-03-25
4501 * A new unit file setting RestrictAddressFamilies= has been
4502 added to restrict which socket address families unit
4503 processes gain access to. This takes address family names
4504 like "AF_INET" or "AF_UNIX", and is useful to minimize the
4505 attack surface of services via exotic protocol stacks. This
4506 is built on seccomp system call filters.
4508 * Two new unit file settings RuntimeDirectory= and
4509 RuntimeDirectoryMode= have been added that may be used to
4510 manage a per-daemon runtime directories below /run. This is
4511 an alternative for setting up directory permissions with
4512 tmpfiles snippets, and has the advantage that the runtime
4513 directory's lifetime is bound to the daemon runtime and that
4514 the daemon starts up with an empty directory each time. This
4515 is particularly useful when writing services that drop
4516 privileges using the User= or Group= setting.
4518 * The DeviceAllow= unit setting now supports globbing for
4519 matching against device group names.
4521 * The systemd configuration file system.conf gained new
4522 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
4523 DefaultMemoryAccounting= to globally turn on/off accounting
4524 for specific resources (cgroups) for all units. These
4525 settings may still be overridden individually in each unit
4528 * systemd-gpt-auto-generator is now able to discover /srv and
4529 root partitions in addition to /home and swap partitions. It
4530 also supports LUKS-encrypted partitions now. With this in
4531 place, automatic discovery of partitions to mount following
4532 the Discoverable Partitions Specification
4533 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
4534 is now a lot more complete. This allows booting without
4535 /etc/fstab and without root= on the kernel command line on
4536 systems prepared appropriately.
4538 * systemd-nspawn gained a new --image= switch which allows
4539 booting up disk images and Linux installations on any block
4540 device that follow the Discoverable Partitions Specification
4541 (see above). This means that installations made with
4542 appropriately updated installers may now be started and
4543 deployed using container managers, completely
4544 unmodified. (We hope that libvirt-lxc will add support for
4545 this feature soon, too.)
4547 * systemd-nspawn gained a new --network-macvlan= setting to
4548 set up a private macvlan interface for the
4549 container. Similarly, systemd-networkd gained a new
4550 Kind=macvlan setting in .netdev files.
4552 * systemd-networkd now supports configuring local addresses
4555 * A new tool systemd-network-wait-online has been added to
4556 synchronously wait for network connectivity using
4559 * The sd-bus.h bus API gained a new sd_bus_track object for
4560 tracking the life-cycle of bus peers. Note that sd-bus.h is
4561 still not a public API though (unless you specify
4562 --enable-kdbus on the configure command line, which however
4563 voids your warranty and you get no API stability guarantee).
4565 * The $XDG_RUNTIME_DIR runtime directories for each user are
4566 now individual tmpfs instances, which has the benefit of
4567 introducing separate pools for each user, with individual
4568 size limits, and thus making sure that unprivileged clients
4569 can no longer negatively impact the system or other users by
4570 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
4571 RuntimeDirectorySize= has been introduced that allows
4572 controlling the default size limit for all users. It
4573 defaults to 10% of the available physical memory. This is no
4574 replacement for quotas on tmpfs though (which the kernel
4575 still does not support), as /dev/shm and /tmp are still
4576 shared resources used by both the system and unprivileged
4579 * logind will now automatically turn off automatic suspending
4580 on laptop lid close when more than one display is
4581 connected. This was previously expected to be implemented
4582 individually in desktop environments (such as GNOME),
4583 however has been added to logind now, in order to fix a
4584 boot-time race where a desktop environment might not have
4585 been started yet and thus not been able to take an inhibitor
4586 lock at the time where logind already suspends the system
4587 due to a closed lid.
4589 * logind will now wait at least 30s after each system
4590 suspend/resume cycle, and 3min after system boot before
4591 suspending the system due to a closed laptop lid. This
4592 should give USB docking stations and similar enough time to
4593 be probed and configured after system resume and boot in
4594 order to then act as suspend blocker.
4596 * systemd-run gained a new --property= setting which allows
4597 initialization of resource control properties (and others)
4598 for the created scope or service unit. Example: "systemd-run
4599 --property=BlockIOWeight=10 updatedb" may be used to run
4600 updatedb at a low block IO scheduling weight.
4602 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
4603 now also work in --scope mode.
4605 * When systemd is compiled with kdbus support, basic support
4606 for enforced policies is now in place. (Note that enabling
4607 kdbus still voids your warranty and no API compatibility
4610 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
4611 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
4612 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
4613 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
4614 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
4615 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
4616 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
4617 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
4618 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
4619 Zbigniew Jędrzejewski-Szmek
4621 — Berlin, 2014-03-12
4625 * systemd will now relabel /dev after loading the SMACK policy
4626 according to SMACK rules.
4628 * A new unit file option AppArmorProfile= has been added to
4629 set the AppArmor profile for the processes of a unit.
4631 * A new condition check ConditionArchitecture= has been added
4632 to conditionalize units based on the system architecture, as
4633 reported by uname()'s "machine" field.
4635 * systemd-networkd now supports matching on the system
4636 virtualization, architecture, kernel command line, host name
4639 * logind is now a lot more aggressive when suspending the
4640 machine due to a closed laptop lid. Instead of acting only
4641 on the lid close action, it will continuously watch the lid
4642 status and act on it. This is useful for laptops where the
4643 power button is on the outside of the chassis so that it can
4644 be reached without opening the lid (such as the Lenovo
4645 Yoga). On those machines, logind will now immediately
4646 re-suspend the machine if the power button has been
4647 accidentally pressed while the laptop was suspended and in a
4648 backpack or similar.
4650 * logind will now watch SW_DOCK switches and inhibit reaction
4651 to the lid switch if it is pressed. This means that logind
4652 will not suspend the machine anymore if the lid is closed
4653 and the system is docked, if the laptop supports SW_DOCK
4654 notifications via the input layer. Note that ACPI docking
4655 stations do not generate this currently. Also note that this
4656 logic is usually not fully sufficient and Desktop
4657 Environments should take a lid switch inhibitor lock when an
4658 external display is connected, as systemd will not watch
4661 * nspawn will now make use of the devices cgroup controller by
4662 default, and only permit creation of and access to the usual
4663 API device nodes like /dev/null or /dev/random, as well as
4664 access to (but not creation of) the pty devices.
4666 * We will now ship a default .network file for
4667 systemd-networkd that automatically configures DHCP for
4668 network interfaces created by nspawn's --network-veth or
4669 --network-bridge= switches.
4671 * systemd will now understand the usual M, K, G, T suffixes
4672 according to SI conventions (i.e. to the base 1000) when
4673 referring to throughput and hardware metrics. It will stay
4674 with IEC conventions (i.e. to the base 1024) for software
4675 metrics, according to what is customary according to
4676 Wikipedia. We explicitly document which base applies for
4677 each configuration option.
4679 * The DeviceAllow= setting in unit files now supports a syntax
4680 to whitelist an entire group of devices node majors at once,
4681 based on the /proc/devices listing. For example, with the
4682 string "char-pts", it is now possible to whitelist all
4683 current and future pseudo-TTYs at once.
4685 * sd-event learned a new "post" event source. Event sources of
4686 this type are triggered by the dispatching of any event
4687 source of a type that is not "post". This is useful for
4688 implementing clean-up and check event sources that are
4689 triggered by other work being done in the program.
4691 * systemd-networkd is no longer statically enabled, but uses
4692 the usual [Install] sections so that it can be
4693 enabled/disabled using systemctl. It still is enabled by
4696 * When creating a veth interface pair with systemd-nspawn, the
4697 host side will now be prefixed with "vb-" if
4698 --network-bridge= is used, and with "ve-" if --network-veth
4699 is used. This way, it is easy to distinguish these cases on
4700 the host, for example to apply different configuration to
4701 them with systemd-networkd.
4703 * The compatibility libraries for libsystemd-journal.so,
4704 libsystem-id128.so, libsystemd-login.so and
4705 libsystemd-daemon.so do not make use of IFUNC
4706 anymore. Instead, we now build libsystemd.so multiple times
4707 under these alternative names. This means that the footprint
4708 is drastically increased, but given that these are
4709 transitional compatibility libraries, this should not matter
4710 much. This change has been made necessary to support the ARM
4711 platform for these compatibility libraries, as the ARM
4712 toolchain is not really at the same level as the toolchain
4713 for other architectures like x86 and does not support
4714 IFUNC. Please make sure to use --enable-compat-libs only
4715 during a transitional period!
4717 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
4718 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
4719 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
4720 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
4721 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
4722 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
4723 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
4724 Zbigniew Jędrzejewski-Szmek
4726 — Berlin, 2014-02-24
4730 * A new component "systemd-networkd" has been added that can
4731 be used to configure local network interfaces statically or
4732 via DHCP. It is capable of bringing up bridges, VLANs, and
4733 bonding. Currently, no hook-ups for interactive network
4734 configuration are provided. Use this for your initrd,
4735 container, embedded, or server setup if you need a simple,
4736 yet powerful, network configuration solution. This
4737 configuration subsystem is quite nifty, as it allows wildcard
4738 hotplug matching in interfaces. For example, with a single
4739 configuration snippet, you can configure that all Ethernet
4740 interfaces showing up are automatically added to a bridge,
4741 or similar. It supports link-sensing and more.
4743 * A new tool "systemd-socket-proxyd" has been added which can
4744 act as a bidirectional proxy for TCP sockets. This is
4745 useful for adding socket activation support to services that
4746 do not actually support socket activation, including virtual
4747 machines and the like.
4749 * Add a new tool to save/restore rfkill state on
4752 * Save/restore state of keyboard backlights in addition to
4753 display backlights on shutdown/boot.
4755 * udev learned a new SECLABEL{} construct to label device
4756 nodes with a specific security label when they appear. For
4757 now, only SECLABEL{selinux} is supported, but the syntax is
4758 prepared for additional security frameworks.
4760 * udev gained a new scheme to configure link-level attributes
4761 from files in /etc/systemd/network/*.link. These files can
4762 match against MAC address, device path, driver name and type,
4763 and will apply attributes like the naming policy, link speed,
4764 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
4765 address assignment policy (randomized, ...).
4767 * The configuration of network interface naming rules for
4768 "permanent interface names" has changed: a new NamePolicy=
4769 setting in the [Link] section of .link files determines the
4770 priority of possible naming schemes (onboard, slot, MAC,
4771 path). The default value of this setting is determined by
4772 /usr/lib/net/links/99-default.link. Old
4773 80-net-name-slot.rules udev configuration file has been
4774 removed, so local configuration overriding this file should
4775 be adapted to override 99-default.link instead.
4777 * When the User= switch is used in a unit file, also
4778 initialize $SHELL= based on the user database entry.
4780 * systemd no longer depends on libdbus. All communication is
4781 now done with sd-bus, systemd's low-level bus library
4784 * kdbus support has been added to PID 1 itself. When kdbus is
4785 enabled, this causes PID 1 to set up the system bus and
4786 enable support for a new ".busname" unit type that
4787 encapsulates bus name activation on kdbus. It works a little
4788 bit like ".socket" units, except for bus names. A new
4789 generator has been added that converts classic dbus1 service
4790 activation files automatically into native systemd .busname
4793 * sd-bus: add a light-weight vtable implementation that allows
4794 defining objects on the bus with a simple static const
4795 vtable array of its methods, signals and properties.
4797 * systemd will not generate or install static dbus
4798 introspection data anymore to /usr/share/dbus-1/interfaces,
4799 as the precise format of these files is unclear, and
4800 nothing makes use of it.
4802 * A proxy daemon is now provided to proxy clients connecting
4803 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
4804 compatibility with classic D-Bus.
4806 * A bus driver implementation has been added that supports the
4807 classic D-Bus bus driver calls on kdbus, also for
4808 compatibility purposes.
4810 * A new API "sd-event.h" has been added that implements a
4811 minimal event loop API built around epoll. It provides a
4812 couple of features that direct epoll usage is lacking:
4813 prioritization of events, scales to large numbers of timer
4814 events, per-event timer slack (accuracy), system-wide
4815 coalescing of timer events, exit handlers, watchdog
4816 supervision support using systemd's sd_notify() API, child
4819 * A new API "sd-rntl.h" has been added that provides an API
4820 around the route netlink interface of the kernel, similar in
4821 style to "sd-bus.h".
4823 * A new API "sd-dhcp-client.h" has been added that provides a
4824 small DHCPv4 client-side implementation. This is used by
4827 * There is a new kernel command line option
4828 "systemd.restore_state=0|1". When set to "0", none of the
4829 systemd tools will restore saved runtime state to hardware
4830 devices. More specifically, the rfkill and backlight states
4833 * The FsckPassNo= compatibility option in mount/service units
4834 has been removed. The fstab generator will now add the
4835 necessary dependencies automatically, and does not require
4836 PID1's support for that anymore.
4838 * journalctl gained a new switch, --list-boots, that lists
4839 recent boots with their times and boot IDs.
4841 * The various tools like systemctl, loginctl, timedatectl,
4842 busctl, systemd-run, ... have gained a new switch "-M" to
4843 connect to a specific, local OS container (as direct
4844 connection, without requiring SSH). This works on any
4845 container that is registered with machined, such as those
4846 created by libvirt-lxc or nspawn.
4848 * systemd-run and systemd-analyze also gained support for "-H"
4849 to connect to remote hosts via SSH. This is particularly
4850 useful for systemd-run because it enables queuing of jobs
4851 onto remote systems.
4853 * machinectl gained a new command "login" to open a getty
4854 login in any local container. This works with any container
4855 that is registered with machined (such as those created by
4856 libvirt-lxc or nspawn), and which runs systemd inside.
4858 * machinectl gained a new "reboot" command that may be used to
4859 trigger a reboot on a specific container that is registered
4860 with machined. This works on any container that runs an init
4861 system of some kind.
4863 * systemctl gained a new "list-timers" command to print a nice
4864 listing of installed timer units with the times they elapse
4867 * Alternative reboot() parameters may now be specified on the
4868 "systemctl reboot" command line and are passed to the
4869 reboot() system call.
4871 * systemctl gained a new --job-mode= switch to configure the
4872 mode to queue a job with. This is a more generic version of
4873 --fail, --irreversible, and --ignore-dependencies, which are
4874 still available but not advertised anymore.
4876 * /etc/systemd/system.conf gained new settings to configure
4877 various default timeouts of units, as well as the default
4878 start limit interval and burst. These may still be overridden
4881 * PID1 will now export on the bus profile data of the security
4882 policy upload process (such as the SELinux policy upload to
4885 * journald: when forwarding logs to the console, include
4886 timestamps (following the setting in
4887 /sys/module/printk/parameters/time).
4889 * OnCalendar= in timer units now understands the special
4890 strings "yearly" and "annually". (Both are equivalent)
4892 * The accuracy of timer units is now configurable with the new
4893 AccuracySec= setting. It defaults to 1min.
4895 * A new dependency type JoinsNamespaceOf= has been added that
4896 allows running two services within the same /tmp and network
4897 namespace, if PrivateNetwork= or PrivateTmp= are used.
4899 * A new command "cat" has been added to systemctl. It outputs
4900 the original unit file of a unit, and concatenates the
4901 contents of additional "drop-in" unit file snippets, so that
4902 the full configuration is shown.
4904 * systemctl now supports globbing on the various "list-xyz"
4905 commands, like "list-units" or "list-sockets", as well as on
4906 those commands which take multiple unit names.
4908 * journalctl's --unit= switch gained support for globbing.
4910 * All systemd daemons now make use of the watchdog logic so
4911 that systemd automatically notices when they hang.
4913 * If the $container_ttys environment variable is set,
4914 getty-generator will automatically spawn a getty for each
4915 listed tty. This is useful for container managers to request
4916 login gettys to be spawned on as many ttys as needed.
4918 * %h, %s, %U specifier support is not available anymore when
4919 used in unit files for PID 1. This is because NSS calls are
4920 not safe from PID 1. They stay available for --user
4921 instances of systemd, and as special case for the root user.
4923 * loginctl gained a new "--no-legend" switch to turn off output
4926 * The "sd-login.h" API gained three new calls:
4927 sd_session_is_remote(), sd_session_get_remote_user(),
4928 sd_session_get_remote_host() to query information about
4931 * The udev hardware database now also carries vendor/product
4932 information of SDIO devices.
4934 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
4935 determine whether watchdog notifications are requested by
4938 * Socket-activated per-connection services now include a
4939 short description of the connection parameters in the
4942 * tmpfiles gained a new "--boot" option. When this is not used,
4943 only lines where the command character is not suffixed with
4944 "!" are executed. When this option is specified, those
4945 options are executed too. This partitions tmpfiles
4946 directives into those that can be safely executed at any
4947 time, and those which should be run only at boot (for
4948 example, a line that creates /run/nologin).
4950 * A new API "sd-resolve.h" has been added which provides a simple
4951 asynchronous wrapper around glibc NSS host name resolution
4952 calls, such as getaddrinfo(). In contrast to glibc's
4953 getaddrinfo_a(), it does not use signals. In contrast to most
4954 other asynchronous name resolution libraries, this one does
4955 not reimplement DNS, but reuses NSS, so that alternate
4956 host name resolution systems continue to work, such as mDNS,
4957 LDAP, etc. This API is based on libasyncns, but it has been
4958 cleaned up for inclusion in systemd.
4960 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
4961 "sd-daemon.h" are no longer found in individual libraries
4962 libsystemd-journal.so, libsystemd-login.so,
4963 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
4964 merged them into a single library, libsystemd.so, which
4965 provides all symbols. The reason for this is cyclic
4966 dependencies, as these libraries tend to use each other's
4967 symbols. So far, we have managed to workaround that by linking
4968 a copy of a good part of our code into each of these
4969 libraries again and again, which, however, makes certain
4970 things hard to do, like sharing static variables. Also, it
4971 substantially increases footprint. With this change, there
4972 is only one library for the basic APIs systemd
4973 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
4974 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
4975 library as well, however are subject to the --enable-kdbus
4976 switch (see below). Note that "sd-dhcp-client.h" is not part
4977 of this library (this is because it only consumes, never
4978 provides, services of/to other APIs). To make the transition
4979 easy from the separate libraries to the unified one, we
4980 provide the --enable-compat-libs compile-time switch which
4981 will generate stub libraries that are compatible with the
4982 old ones but redirect all calls to the new one.
4984 * All of the kdbus logic and the new APIs "sd-bus.h",
4985 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
4986 and "sd-utf8.h" are compile-time optional via the
4987 "--enable-kdbus" switch, and they are not compiled in by
4988 default. To make use of kdbus, you have to explicitly enable
4989 the switch. Note however, that neither the kernel nor the
4990 userspace API for all of this is considered stable yet. We
4991 want to maintain the freedom to still change the APIs for
4992 now. By specifying this build-time switch, you acknowledge
4993 that you are aware of the instability of the current
4996 * Also, note that while kdbus is pretty much complete,
4997 it lacks one thing: proper policy support. This means you
4998 can build a fully working system with all features; however,
4999 it will be highly insecure. Policy support will be added in
5000 one of the next releases, at the same time that we will
5001 declare the APIs stable.
5003 * When the kernel command line argument "kdbus" is specified,
5004 systemd will automatically load the kdbus.ko kernel module. At
5005 this stage of development, it is only useful for testing kdbus
5006 and should not be used in production. Note: if "--enable-kdbus"
5007 is specified, and the kdbus.ko kernel module is available, and
5008 "kdbus" is added to the kernel command line, the entire system
5009 runs with kdbus instead of dbus-daemon, with the above mentioned
5010 problem of missing the system policy enforcement. Also a future
5011 version of kdbus.ko or a newer systemd will not be compatible with
5012 each other, and will unlikely be able to boot the machine if only
5013 one of them is updated.
5015 * systemctl gained a new "import-environment" command which
5016 uploads the caller's environment (or parts thereof) into the
5017 service manager so that it is inherited by services started
5018 by the manager. This is useful to upload variables like
5019 $DISPLAY into the user service manager.
5021 * A new PrivateDevices= switch has been added to service units
5022 which allows running a service with a namespaced /dev
5023 directory that does not contain any device nodes for
5024 physical devices. More specifically, it only includes devices
5025 such as /dev/null, /dev/urandom, and /dev/zero which are API
5028 * logind has been extended to support behaviour like VT
5029 switching on seats that do not support a VT. This makes
5030 multi-session available on seats that are not the first seat
5031 (seat0), and on systems where kernel support for VTs has
5032 been disabled at compile-time.
5034 * If a process holds a delay lock for system sleep or shutdown
5035 and fails to release it in time, we will now log its
5036 identity. This makes it easier to identify processes that
5037 cause slow suspends or power-offs.
5039 * When parsing /etc/crypttab, support for a new key-slot=
5040 option as supported by Debian is added. It allows indicating
5041 which LUKS slot to use on disk, speeding up key loading.
5043 * The sd_journald_sendv() API call has been checked and
5044 officially declared to be async-signal-safe so that it may
5045 be invoked from signal handlers for logging purposes.
5047 * Boot-time status output is now enabled automatically after a
5048 short timeout if boot does not progress, in order to give
5049 the user an indication what she or he is waiting for.
5051 * The boot-time output has been improved to show how much time
5052 remains until jobs expire.
5054 * The KillMode= switch in service units gained a new possible
5055 value "mixed". If set, and the unit is shut down, then the
5056 initial SIGTERM signal is sent only to the main daemon
5057 process, while the following SIGKILL signal is sent to
5058 all remaining processes of the service.
5060 * When a scope unit is registered, a new property "Controller"
5061 may be set. If set to a valid bus name, systemd will send a
5062 RequestStop() signal to this name when it would like to shut
5063 down the scope. This may be used to hook manager logic into
5064 the shutdown logic of scope units. Also, scope units may now
5065 be put in a special "abandoned" state, in which case the
5066 manager process which created them takes no further
5067 responsibilities for it.
5069 * When reading unit files, systemd will now verify
5070 the access mode of these files, and warn about certain
5071 suspicious combinations. This has been added to make it
5072 easier to track down packaging bugs where unit files are
5073 marked executable or world-writable.
5075 * systemd-nspawn gained a new "--setenv=" switch to set
5076 container-wide environment variables. The similar option in
5077 systemd-activate was renamed from "--environment=" to
5078 "--setenv=" for consistency.
5080 * systemd-nspawn has been updated to create a new kdbus domain
5081 for each container that is invoked, thus allowing each
5082 container to have its own set of system and user buses,
5083 independent of the host.
5085 * systemd-nspawn gained a new --drop-capability= switch to run
5086 the container with less capabilities than the default. Both
5087 --drop-capability= and --capability= now take the special
5088 string "all" for dropping or keeping all capabilities.
5090 * systemd-nspawn gained new switches for executing containers
5091 with specific SELinux labels set.
5093 * systemd-nspawn gained a new --quiet switch to not generate
5094 any additional output but the container's own console
5097 * systemd-nspawn gained a new --share-system switch to run a
5098 container without PID namespacing enabled.
5100 * systemd-nspawn gained a new --register= switch to control
5101 whether the container is registered with systemd-machined or
5102 not. This is useful for containers that do not run full
5103 OS images, but only specific apps.
5105 * systemd-nspawn gained a new --keep-unit which may be used
5106 when invoked as the only program from a service unit, and
5107 results in registration of the unit service itself in
5108 systemd-machined, instead of a newly opened scope unit.
5110 * systemd-nspawn gained a new --network-interface= switch for
5111 moving arbitrary interfaces to the container. The new
5112 --network-veth switch creates a virtual Ethernet connection
5113 between host and container. The new --network-bridge=
5114 switch then allows assigning the host side of this virtual
5115 Ethernet connection to a bridge device.
5117 * systemd-nspawn gained a new --personality= switch for
5118 setting the kernel personality for the container. This is
5119 useful when running a 32-bit container on a 64-bit host. A
5120 similar option Personality= is now also available for service
5123 * logind will now also track a "Desktop" identifier for each
5124 session which encodes the desktop environment of it. This is
5125 useful for desktop environments that want to identify
5126 multiple running sessions of itself easily.
5128 * A new SELinuxContext= setting for service units has been
5129 added that allows setting a specific SELinux execution
5130 context for a service.
5132 * Most systemd client tools will now honour $SYSTEMD_LESS for
5133 settings of the "less" pager. By default, these tools will
5134 override $LESS to allow certain operations to work, such as
5135 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
5136 influence this logic.
5138 * systemd's "seccomp" hook-up has been changed to make use of
5139 the libseccomp library instead of using its own
5140 implementation. This has benefits for portability among
5143 * For usage together with SystemCallFilter=, a new
5144 SystemCallErrorNumber= setting has been introduced that
5145 allows configuration of a system error number to be returned
5146 on filtered system calls, instead of immediately killing the
5147 process. Also, SystemCallArchitectures= has been added to
5148 limit access to system calls of a particular architecture
5149 (in order to turn off support for unused secondary
5150 architectures). There is also a global
5151 SystemCallArchitectures= setting in system.conf now to turn
5152 off support for non-native system calls system-wide.
5154 * systemd requires a kernel with a working name_to_handle_at(),
5155 please see the kernel config requirements in the README file.
5157 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
5158 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
5159 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
5160 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
5161 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
5162 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
5163 Elia Pinto, Florian Weimer, George McCollister, Goffredo
5164 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
5165 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
5166 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
5167 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
5168 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
5169 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
5170 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
5171 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
5172 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
5173 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
5174 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
5175 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
5176 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
5177 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
5178 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
5179 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
5180 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
5182 — Berlin, 2014-02-20
5186 * logind has gained support for facilitating privileged input
5187 and drm device access for unprivileged clients. This work is
5188 useful to allow Wayland display servers (and similar
5189 programs, such as kmscon) to run under the user's ID and
5190 access input and drm devices which are normally
5191 protected. When this is used (and the kernel is new enough)
5192 logind will "mute" IO on the file descriptors passed to
5193 Wayland as long as it is in the background and "unmute" it
5194 if it returns into the foreground. This allows secure
5195 session switching without allowing background sessions to
5196 eavesdrop on input and display data. This also introduces
5197 session switching support if VT support is turned off in the
5198 kernel, and on seats that are not seat0.
5200 * A new kernel command line option luks.options= is understood
5201 now which allows specifying LUKS options for usage for LUKS
5202 encrypted partitions specified with luks.uuid=.
5204 * tmpfiles.d(5) snippets may now use specifier expansion in
5205 path names. More specifically %m, %b, %H, %v, are now
5206 replaced by the local machine id, boot id, hostname, and
5207 kernel version number.
5209 * A new tmpfiles.d(5) command "m" has been introduced which
5210 may be used to change the owner/group/access mode of a file
5211 or directory if it exists, but do nothing if it does not.
5213 * This release removes high-level support for the
5214 MemorySoftLimit= cgroup setting. The underlying kernel
5215 cgroup attribute memory.soft_limit= is currently badly
5216 designed and likely to be removed from the kernel API in its
5217 current form, hence we should not expose it for now.
5219 * The memory.use_hierarchy cgroup attribute is now enabled for
5220 all cgroups systemd creates in the memory cgroup
5221 hierarchy. This option is likely to be come the built-in
5222 default in the kernel anyway, and the non-hierarchical mode
5223 never made much sense in the intrinsically hierarchical
5226 * A new field _SYSTEMD_SLICE= is logged along with all journal
5227 messages containing the slice a message was generated
5228 from. This is useful to allow easy per-customer filtering of
5229 logs among other things.
5231 * systemd-journald will no longer adjust the group of journal
5232 files it creates to the "systemd-journal" group. Instead we
5233 rely on the journal directory to be owned by the
5234 "systemd-journal" group, and its setgid bit set, so that the
5235 kernel file system layer will automatically enforce that
5236 journal files inherit this group assignment. The reason for
5237 this change is that we cannot allow NSS look-ups from
5238 journald which would be necessary to resolve
5239 "systemd-journal" to a numeric GID, because this might
5240 create deadlocks if NSS involves synchronous queries to
5241 other daemons (such as nscd, or sssd) which in turn are
5242 logging clients of journald and might block on it, which
5243 would then dead lock. A tmpfiles.d(5) snippet included in
5244 systemd will make sure the setgid bit and group are
5245 properly set on the journal directory if it exists on every
5246 boot. However, we recommend adjusting it manually after
5247 upgrades too (or from RPM scriptlets), so that the change is
5248 not delayed until next reboot.
5250 * Backlight and random seed files in /var/lib/ have moved into
5251 the /var/lib/systemd/ directory, in order to centralize all
5252 systemd generated files in one directory.
5254 * Boot time performance measurements (as displayed by
5255 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
5256 performance information if that's available to determine how
5257 much time BIOS and boot loader initialization required. With
5258 a sufficiently new BIOS you hence no longer need to boot
5259 with Gummiboot to get access to such information.
5261 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
5262 Cristian Rodríguez, Dave Reisner, David Herrmann, David
5263 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
5264 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
5265 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
5266 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
5267 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
5269 — Berlin, 2013-10-02
5273 * The Restart= option for services now understands a new
5274 on-watchdog setting, which will restart the service
5275 automatically if the service stops sending out watchdog keep
5276 alive messages (as configured with WatchdogSec=).
5278 * The getty generator (which is responsible for bringing up a
5279 getty on configured serial consoles) will no longer only
5280 start a getty on the primary kernel console but on all
5281 others, too. This makes the order in which console= is
5282 specified on the kernel command line less important.
5284 * libsystemd-logind gained a new sd_session_get_vt() call to
5285 retrieve the VT number of a session.
5287 * If the option "tries=0" is set for an entry of /etc/crypttab
5288 its passphrase is queried indefinitely instead of any
5289 maximum number of tries.
5291 * If a service with a configure PID file terminates its PID
5292 file will now be removed automatically if it still exists
5293 afterwards. This should put an end to stale PID files.
5295 * systemd-run will now also take relative binary path names
5296 for execution and no longer insists on absolute paths.
5298 * InaccessibleDirectories= and ReadOnlyDirectories= now take
5299 paths that are optionally prefixed with "-" to indicate that
5300 it should not be considered a failure if they do not exist.
5302 * journalctl -o (and similar commands) now understands a new
5303 output mode "short-precise", it is similar to "short" but
5304 shows timestamps with usec accuracy.
5306 * The option "discard" (as known from Debian) is now
5307 synonymous to "allow-discards" in /etc/crypttab. In fact,
5308 "discard" is preferred now (since it is easier to remember
5311 * Some licensing clean-ups were made, so that more code is now
5312 LGPL-2.1 licensed than before.
5314 * A minimal tool to save/restore the display backlight
5315 brightness across reboots has been added. It will store the
5316 backlight setting as late as possible at shutdown, and
5317 restore it as early as possible during reboot.
5319 * A logic to automatically discover and enable home and swap
5320 partitions on GPT disks has been added. With this in place
5321 /etc/fstab becomes optional for many setups as systemd can
5322 discover certain partitions located on the root disk
5323 automatically. Home partitions are recognized under their
5324 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
5325 partitions are recognized under their GPT type ID
5326 0657fd6da4ab43c484e50933c84b4f4f.
5328 * systemd will no longer pass any environment from the kernel
5329 or initrd to system services. If you want to set an
5330 environment for all services, do so via the kernel command
5331 line systemd.setenv= assignment.
5333 * The systemd-sysctl tool no longer natively reads the file
5334 /etc/sysctl.conf. If desired, the file should be symlinked
5335 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
5336 legacy support by a symlink rather than built-in code, it
5337 also makes the otherwise hidden order of application of the
5338 different files visible. (Note that this partly reverts to a
5339 pre-198 application order of sysctl knobs!)
5341 * The "systemctl set-log-level" and "systemctl dump" commands
5342 have been moved to systemd-analyze.
5344 * systemd-run learned the new --remain-after-exit switch,
5345 which causes the scope unit not to be cleaned up
5346 automatically after the process terminated.
5348 * tmpfiles learned a new --exclude-prefix= switch to exclude
5349 certain paths from operation.
5351 * journald will now automatically flush all messages to disk
5352 as soon as a message at the log level CRIT, ALERT or EMERG
5355 Contributions from: Andrew Cook, Brandon Philips, Christian
5356 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
5357 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
5358 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
5359 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
5360 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
5361 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
5362 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
5363 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
5364 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
5365 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
5366 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
5367 William Giokas, Zbigniew Jędrzejewski-Szmek
5369 — Berlin, 2013-09-13
5373 * The documentation has been updated to cover the various new
5374 concepts introduced with 205.
5376 * Unit files now understand the new %v specifier which
5377 resolves to the kernel version string as returned by "uname
5380 * systemctl now supports filtering the unit list output by
5381 load state, active state and sub state, using the new
5384 * "systemctl status" will now show the results of the
5385 condition checks (like ConditionPathExists= and similar) of
5386 the last start attempts of the unit. They are also logged to
5389 * "journalctl -b" may now be used to look for boot output of a
5390 specific boot. Try "journalctl -b -1" for the previous boot,
5391 but the syntax is substantially more powerful.
5393 * "journalctl --show-cursor" has been added which prints the
5394 cursor string the last shown log line. This may then be used
5395 with the new "journalctl --after-cursor=" switch to continue
5396 browsing logs from that point on.
5398 * "journalctl --force" may now be used to force regeneration
5401 * Creation of "dead" device nodes has been moved from udev
5402 into kmod and tmpfiles. Previously, udev would read the kmod
5403 databases to pre-generate dead device nodes based on meta
5404 information contained in kernel modules, so that these would
5405 be auto-loaded on access rather then at boot. As this
5406 does not really have much to do with the exposing actual
5407 kernel devices to userspace this has always been slightly
5408 alien in the udev codebase. Following the new scheme kmod
5409 will now generate a runtime snippet for tmpfiles from the
5410 module meta information and it now is tmpfiles' job to the
5411 create the nodes. This also allows overriding access and
5412 other parameters for the nodes using the usual tmpfiles
5413 facilities. As side effect this allows us to remove the
5414 CAP_SYS_MKNOD capability bit from udevd entirely.
5416 * logind's device ACLs may now be applied to these "dead"
5417 devices nodes too, thus finally allowing managed access to
5418 devices such as /dev/snd/sequencer without loading the
5419 backing module right-away.
5421 * A new RPM macro has been added that may be used to apply
5422 tmpfiles configuration during package installation.
5424 * systemd-detect-virt and ConditionVirtualization= now can
5425 detect User-Mode-Linux machines (UML).
5427 * journald will now implicitly log the effective capabilities
5428 set of processes in the message metadata.
5430 * systemd-cryptsetup has gained support for TrueCrypt volumes.
5432 * The initrd interface has been simplified (more specifically,
5433 support for passing performance data via environment
5434 variables and fsck results via files in /run has been
5435 removed). These features were non-essential, and are
5436 nowadays available in a much nicer way by having systemd in
5437 the initrd serialize its state and have the hosts systemd
5438 deserialize it again.
5440 * The udev "keymap" data files and tools to apply keyboard
5441 specific mappings of scan to key codes, and force-release
5442 scan code lists have been entirely replaced by a udev
5443 "keyboard" builtin and a hwdb data file.
5445 * systemd will now honour the kernel's "quiet" command line
5446 argument also during late shutdown, resulting in a
5447 completely silent shutdown when used.
5449 * There's now an option to control the SO_REUSEPORT socket
5450 option in .socket units.
5452 * Instance units will now automatically get a per-template
5453 subslice of system.slice unless something else is explicitly
5454 configured. For example, instances of sshd@.service will now
5455 implicitly be placed in system-sshd.slice rather than
5456 system.slice as before.
5458 * Test coverage support may now be enabled at build time.
5460 Contributions from: Dave Reisner, Frederic Crozat, Harald
5461 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
5462 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
5463 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
5464 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
5465 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
5466 Giokas, Zbigniew Jędrzejewski-Szmek
5468 — Berlin, 2013-07-23
5472 * Two new unit types have been introduced:
5474 Scope units are very similar to service units, however, are
5475 created out of pre-existing processes — instead of PID 1
5476 forking off the processes. By using scope units it is
5477 possible for system services and applications to group their
5478 own child processes (worker processes) in a powerful way
5479 which then maybe used to organize them, or kill them
5480 together, or apply resource limits on them.
5482 Slice units may be used to partition system resources in an
5483 hierarchical fashion and then assign other units to them. By
5484 default there are now three slices: system.slice (for all
5485 system services), user.slice (for all user sessions),
5486 machine.slice (for VMs and containers).
5488 Slices and scopes have been introduced primarily in
5489 context of the work to move cgroup handling to a
5490 single-writer scheme, where only PID 1
5491 creates/removes/manages cgroups.
5493 * There's a new concept of "transient" units. In contrast to
5494 normal units these units are created via an API at runtime,
5495 not from configuration from disk. More specifically this
5496 means it is now possible to run arbitrary programs as
5497 independent services, with all execution parameters passed
5498 in via bus APIs rather than read from disk. Transient units
5499 make systemd substantially more dynamic then it ever was,
5500 and useful as a general batch manager.
5502 * logind has been updated to make use of scope and slice units
5503 for managing user sessions. As a user logs in he will get
5504 his own private slice unit, to which all sessions are added
5505 as scope units. We also added support for automatically
5506 adding an instance of user@.service for the user into the
5507 slice. Effectively logind will no longer create cgroup
5508 hierarchies on its own now, it will defer entirely to PID 1
5509 for this by means of scope, service and slice units. Since
5510 user sessions this way become entities managed by PID 1
5511 the output of "systemctl" is now a lot more comprehensive.
5513 * A new mini-daemon "systemd-machined" has been added which
5514 may be used by virtualization managers to register local
5515 VMs/containers. nspawn has been updated accordingly, and
5516 libvirt will be updated shortly. machined will collect a bit
5517 of meta information about the VMs/containers, and assign
5518 them their own scope unit (see above). The collected
5519 meta-data is then made available via the "machinectl" tool,
5520 and exposed in "ps" and similar tools. machined/machinectl
5521 is compile-time optional.
5523 * As discussed earlier, the low-level cgroup configuration
5524 options ControlGroup=, ControlGroupModify=,
5525 ControlGroupPersistent=, ControlGroupAttribute= have been
5526 removed. Please use high-level attribute settings instead as
5527 well as slice units.
5529 * A new bus call SetUnitProperties() has been added to alter
5530 various runtime parameters of a unit. This is primarily
5531 useful to alter cgroup parameters dynamically in a nice way,
5532 but will be extended later on to make more properties
5533 modifiable at runtime. systemctl gained a new set-properties
5534 command that wraps this call.
5536 * A new tool "systemd-run" has been added which can be used to
5537 run arbitrary command lines as transient services or scopes,
5538 while configuring a number of settings via the command
5539 line. This tool is currently very basic, however already
5540 very useful. We plan to extend this tool to even allow
5541 queuing of execution jobs with time triggers from the
5542 command line, similar in fashion to "at".
5544 * nspawn will now inform the user explicitly that kernels with
5545 audit enabled break containers, and suggest the user to turn
5548 * Support for detecting the IMA and AppArmor security
5549 frameworks with ConditionSecurity= has been added.
5551 * journalctl gained a new "-k" switch for showing only kernel
5552 messages, mimicking dmesg output; in addition to "--user"
5553 and "--system" switches for showing only user's own logs
5556 * systemd-delta can now show information about drop-in
5557 snippets extending unit files.
5559 * libsystemd-bus has been substantially updated but is still
5560 not available as public API.
5562 * systemd will now look for the "debug" argument on the kernel
5563 command line and enable debug logging, similar to what
5564 "systemd.log_level=debug" already did before.
5566 * "systemctl set-default", "systemctl get-default" has been
5567 added to configure the default.target symlink, which
5568 controls what to boot into by default.
5570 * "systemctl set-log-level" has been added as a convenient
5571 way to raise and lower systemd logging threshold.
5573 * "systemd-analyze plot" will now show the time the various
5574 generators needed for execution, as well as information
5575 about the unit file loading.
5577 * libsystemd-journal gained a new sd_journal_open_files() call
5578 for opening specific journal files. journactl also gained a
5579 new switch to expose this new functionality. Previously we
5580 only supported opening all files from a directory, or all
5581 files from the system, as opening individual files only is
5582 racy due to journal file rotation.
5584 * systemd gained the new DefaultEnvironment= setting in
5585 /etc/systemd/system.conf to set environment variables for
5588 * If a privileged process logs a journal message with the
5589 OBJECT_PID= field set, then journald will automatically
5590 augment this with additional OBJECT_UID=, OBJECT_GID=,
5591 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
5592 system services want to log events about specific client
5593 processes. journactl/systemctl has been updated to make use
5594 of this information if all log messages regarding a specific
5597 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
5598 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
5599 Reisner, David Coppa, David King, David Strauss, Eelco
5600 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
5601 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
5602 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
5603 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
5604 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
5605 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
5606 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
5607 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
5608 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
5609 Łukasz Stelmach, 장동준
5613 * The Python bindings gained some minimal support for the APIs
5614 exposed by libsystemd-logind.
5616 * ConditionSecurity= gained support for detecting SMACK. Since
5617 this condition already supports SELinux and AppArmor we only
5618 miss IMA for this. Patches welcome!
5620 Contributions from: Karol Lewandowski, Lennart Poettering,
5621 Zbigniew Jędrzejewski-Szmek
5625 * systemd-nspawn will now create /etc/resolv.conf if
5626 necessary, before bind-mounting the host's file onto it.
5628 * systemd-nspawn will now store meta information about a
5629 container on the container's cgroup as extended attribute
5630 fields, including the root directory.
5632 * The cgroup hierarchy has been reworked in many ways. All
5633 objects any of the components systemd creates in the cgroup
5634 tree are now suffixed. More specifically, user sessions are
5635 now placed in cgroups suffixed with ".session", users in
5636 cgroups suffixed with ".user", and nspawn containers in
5637 cgroups suffixed with ".nspawn". Furthermore, all cgroup
5638 names are now escaped in a simple scheme to avoid collision
5639 of userspace object names with kernel filenames. This work
5640 is preparation for making these objects relocatable in the
5641 cgroup tree, in order to allow easy resource partitioning of
5642 these objects without causing naming conflicts.
5644 * systemctl list-dependencies gained the new switches
5645 --plain, --reverse, --after and --before.
5647 * systemd-inhibit now shows the process name of processes that
5648 have taken an inhibitor lock.
5650 * nss-myhostname will now also resolve "localhost"
5651 implicitly. This makes /etc/hosts an optional file and
5652 nicely handles that on IPv6 ::1 maps to both "localhost" and
5655 * libsystemd-logind.so gained a new call
5656 sd_get_machine_names() to enumerate running containers and
5657 VMs (currently only supported by very new libvirt and
5658 nspawn). sd_login_monitor can now be used to watch
5659 VMs/containers coming and going.
5661 * .include is not allowed recursively anymore, and only in
5662 unit files. Usually it is better to use drop-in snippets in
5663 .d/*.conf anyway, as introduced with systemd 198.
5665 * systemd-analyze gained a new "critical-chain" command that
5666 determines the slowest chain of units run during system
5667 boot-up. It is very useful for tracking down where
5668 optimizing boot time is the most beneficial.
5670 * systemd will no longer allow manipulating service paths in
5671 the name=systemd:/system cgroup tree using ControlGroup= in
5672 units. (But is still fine with it in all other dirs.)
5674 * There's a new systemd-nspawn@.service service file that may
5675 be used to easily run nspawn containers as system
5676 services. With the container's root directory in
5677 /var/lib/container/foobar it is now sufficient to run
5678 "systemctl start systemd-nspawn@foobar.service" to boot it.
5680 * systemd-cgls gained a new parameter "--machine" to list only
5681 the processes within a certain container.
5683 * ConditionSecurity= now can check for "apparmor". We still
5684 are lacking checks for SMACK and IMA for this condition
5685 check though. Patches welcome!
5687 * A new configuration file /etc/systemd/sleep.conf has been
5688 added that may be used to configure which kernel operation
5689 systemd is supposed to execute when "suspend", "hibernate"
5690 or "hybrid-sleep" is requested. This makes the new kernel
5691 "freeze" state accessible to the user.
5693 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
5694 the passed argument if applicable.
5696 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
5697 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
5698 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
5699 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
5700 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
5701 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
5702 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
5707 * The output of 'systemctl list-jobs' got some polishing. The
5708 '--type=' argument may now be passed more than once. A new
5709 command 'systemctl list-sockets' has been added which shows
5710 a list of kernel sockets systemd is listening on with the
5711 socket units they belong to, plus the units these socket
5714 * The experimental libsystemd-bus library got substantial
5715 updates to work in conjunction with the (also experimental)
5716 kdbus kernel project. It works well enough to exchange
5717 messages with some sophistication. Note that kdbus is not
5718 ready yet, and the library is mostly an elaborate test case
5719 for now, and not installable.
5721 * systemd gained a new unit 'systemd-static-nodes.service'
5722 that generates static device nodes earlier during boot, and
5723 can run in conjunction with udev.
5725 * libsystemd-login gained a new call sd_pid_get_user_unit()
5726 to retrieve the user systemd unit a process is running
5727 in. This is useful for systems where systemd is used as
5730 * systemd-nspawn now places all containers in the new /machine
5731 top-level cgroup directory in the name=systemd
5732 hierarchy. libvirt will soon do the same, so that we get a
5733 uniform separation of /system, /user and /machine for system
5734 services, user processes and containers/virtual
5735 machines. This new cgroup hierarchy is also useful to stick
5736 stable names to specific container instances, which can be
5737 recognized later this way (this name may be controlled
5738 via systemd-nspawn's new -M switch). libsystemd-login also
5739 gained a new call sd_pid_get_machine_name() to retrieve the
5740 name of the container/VM a specific process belongs to.
5742 * bootchart can now store its data in the journal.
5744 * libsystemd-journal gained a new call
5745 sd_journal_add_conjunction() for AND expressions to the
5746 matching logic. This can be used to express more complex
5747 logical expressions.
5749 * journactl can now take multiple --unit= and --user-unit=
5752 * The cryptsetup logic now understands the "luks.key=" kernel
5753 command line switch for specifying a file to read the
5754 decryption key from. Also, if a configured key file is not
5755 found the tool will now automatically fall back to prompting
5758 * Python systemd.journal module was updated to wrap recently
5759 added functions from libsystemd-journal. The interface was
5760 changed to bring the low level interface in s.j._Reader
5761 closer to the C API, and the high level interface in
5762 s.j.Reader was updated to wrap and convert all data about
5765 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
5766 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
5767 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
5768 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
5769 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
5770 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
5774 * journalctl --update-catalog now understands a new --root=
5775 option to operate on catalogs found in a different root
5778 * During shutdown after systemd has terminated all running
5779 services a final killing loop kills all remaining left-over
5780 processes. We will now print the name of these processes
5781 when we send SIGKILL to them, since this usually indicates a
5784 * If /etc/crypttab refers to password files stored on
5785 configured mount points automatic dependencies will now be
5786 generated to ensure the specific mount is established first
5787 before the key file is attempted to be read.
5789 * 'systemctl status' will now show information about the
5790 network sockets a socket unit is listening on.
5792 * 'systemctl status' will also shown information about any
5793 drop-in configuration file for units. (Drop-In configuration
5794 files in this context are files such as
5795 /etc/systemd/systemd/foobar.service.d/*.conf)
5797 * systemd-cgtop now optionally shows summed up CPU times of
5798 cgroups. Press '%' while running cgtop to switch between
5799 percentage and absolute mode. This is useful to determine
5800 which cgroups use up the most CPU time over the entire
5801 runtime of the system. systemd-cgtop has also been updated
5802 to be 'pipeable' for processing with further shell tools.
5804 * 'hostnamectl set-hostname' will now allow setting of FQDN
5807 * The formatting and parsing of time span values has been
5808 changed. The parser now understands fractional expressions
5809 such as "5.5h". The formatter will now output fractional
5810 expressions for all time spans under 1min, i.e. "5.123456s"
5811 rather than "5s 123ms 456us". For time spans under 1s
5812 millisecond values are shown, for those under 1ms
5813 microsecond values are shown. This should greatly improve
5814 all time-related output of systemd.
5816 * libsystemd-login and libsystemd-journal gained new
5817 functions for querying the poll() events mask and poll()
5818 timeout value for integration into arbitrary event
5821 * localectl gained the ability to list available X11 keymaps
5822 (models, layouts, variants, options).
5824 * 'systemd-analyze dot' gained the ability to filter for
5825 specific units via shell-style globs, to create smaller,
5826 more useful graphs. I.e. it is now possible to create simple
5827 graphs of all the dependencies between only target units, or
5828 of all units that Avahi has dependencies with.
5830 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
5831 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
5832 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
5833 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
5834 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
5835 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
5836 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
5840 * The boot-time readahead implementation for rotating media
5841 will now read the read-ahead data in multiple passes which
5842 consist of all read requests made in equidistant time
5843 intervals. This means instead of strictly reading read-ahead
5844 data in its physical order on disk we now try to find a
5845 middle ground between physical and access time order.
5847 * /etc/os-release files gained a new BUILD_ID= field for usage
5848 on operating systems that provide continuous builds of OS
5851 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
5852 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
5853 William Douglas, Zbigniew Jędrzejewski-Szmek
5857 * systemd-python gained an API exposing libsystemd-daemon.
5859 * The SMACK setup logic gained support for uploading CIPSO
5862 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
5863 ReadOnlyDirectories= and InaccessibleDirectories= has
5864 changed. The private /tmp and /var/tmp directories are now
5865 shared by all processes of a service (which means
5866 ExecStartPre= may now leave data in /tmp that ExecStart= of
5867 the same service can still access). When a service is
5868 stopped its temporary directories are immediately deleted
5869 (normal clean-up with tmpfiles is still done in addition to
5872 * By default, systemd will now set a couple of sysctl
5873 variables in the kernel: the safe sysrq options are turned
5874 on, IP route verification is turned on, and source routing
5875 disabled. The recently added hardlink and softlink
5876 protection of the kernel is turned on. These settings should
5877 be reasonably safe, and good defaults for all new systems.
5879 * The predictable network naming logic may now be turned off
5880 with a new kernel command line switch: net.ifnames=0.
5882 * A new libsystemd-bus module has been added that implements a
5883 pretty complete D-Bus client library. For details see:
5885 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
5887 * journald will now explicitly flush the journal files to disk
5888 at the latest 5min after each write. The file will then also
5889 be marked offline until the next write. This should increase
5890 reliability in case of a crash. The synchronization delay
5891 can be configured via SyncIntervalSec= in journald.conf.
5893 * There's a new remote-fs-setup.target unit that can be used
5894 to pull in specific services when at least one remote file
5895 system is to be mounted.
5897 * There are new targets timers.target and paths.target as
5898 canonical targets to pull user timer and path units in
5899 from. This complements sockets.target with a similar
5900 purpose for socket units.
5902 * libudev gained a new call udev_device_set_attribute_value()
5903 to set sysfs attributes of a device.
5905 * The udev daemon now sets the default number of worker
5906 processes executed in parallel based on the number of available
5907 CPUs instead of the amount of available RAM. This is supposed
5908 to provide a more reliable default and limit a too aggressive
5909 parallelism for setups with 1000s of devices connected.
5911 Contributions from: Auke Kok, Colin Walters, Cristian
5912 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
5913 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
5914 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
5915 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
5916 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
5917 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
5918 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
5919 Zbigniew Jędrzejewski-Szmek
5923 * Configuration of unit files may now be extended via drop-in
5924 files without having to edit/override the unit files
5925 themselves. More specifically, if the administrator wants to
5926 change one value for a service file foobar.service he can
5927 now do so by dropping in a configuration snippet into
5928 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
5929 will load all these snippets and apply them on top of the
5930 main unit configuration file, possibly extending or
5931 overriding its settings. Using these drop-in snippets is
5932 generally nicer than the two earlier options for changing
5933 unit files locally: copying the files from
5934 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
5935 them there; or creating a new file in /etc/systemd/system/
5936 that incorporates the original one via ".include". Drop-in
5937 snippets into these .d/ directories can be placed in any
5938 directory systemd looks for units in, and the usual
5939 overriding semantics between /usr/lib, /etc and /run apply
5942 * Most unit file settings which take lists of items can now be
5943 reset by assigning the empty string to them. For example,
5944 normally, settings such as Environment=FOO=BAR append a new
5945 environment variable assignment to the environment block,
5946 each time they are used. By assigning Environment= the empty
5947 string the environment block can be reset to empty. This is
5948 particularly useful with the .d/*.conf drop-in snippets
5949 mentioned above, since this adds the ability to reset list
5950 settings from vendor unit files via these drop-ins.
5952 * systemctl gained a new "list-dependencies" command for
5953 listing the dependencies of a unit recursively.
5955 * Inhibitors are now honored and listed by "systemctl
5956 suspend", "systemctl poweroff" (and similar) too, not only
5957 GNOME. These commands will also list active sessions by
5960 * Resource limits (as exposed by the various control group
5961 controllers) can now be controlled dynamically at runtime
5962 for all units. More specifically, you can now use a command
5963 like "systemctl set-cgroup-attr foobar.service cpu.shares
5964 2000" to alter the CPU shares a specific service gets. These
5965 settings are stored persistently on disk, and thus allow the
5966 administrator to easily adjust the resource usage of
5967 services with a few simple commands. This dynamic resource
5968 management logic is also available to other programs via the
5969 bus. Almost any kernel cgroup attribute and controller is
5972 * systemd-vconsole-setup will now copy all font settings to
5973 all allocated VTs, where it previously applied them only to
5976 * libsystemd-login gained the new sd_session_get_tty() API
5979 * This release drops support for a few legacy or
5980 distribution-specific LSB facility names when parsing init
5981 scripts: $x-display-manager, $mail-transfer-agent,
5982 $mail-transport-agent, $mail-transfer-agent, $smtp,
5983 $null. Also, the mail-transfer-agent.target unit backing
5984 this has been removed. Distributions which want to retain
5985 compatibility with this should carry the burden for
5986 supporting this themselves and patch support for these back
5987 in, if they really need to. Also, the facilities $syslog and
5988 $local_fs are now ignored, since systemd does not support
5989 early-boot LSB init scripts anymore, and these facilities
5990 are implied anyway for normal services. syslog.target has
5993 * There are new bus calls on PID1's Manager object for
5994 cancelling jobs, and removing snapshot units. Previously,
5995 both calls were only available on the Job and Snapshot
5998 * systemd-journal-gatewayd gained SSL support.
6000 * The various "environment" files, such as /etc/locale.conf
6001 now support continuation lines with a backslash ("\") as
6002 last character in the line, similarly in style (but different)
6003 to how this is supported in shells.
6005 * For normal user processes the _SYSTEMD_USER_UNIT= field is
6006 now implicitly appended to every log entry logged. systemctl
6007 has been updated to filter by this field when operating on a
6008 user systemd instance.
6010 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
6011 CAP_AUDIT_CONTROL capabilities to the capabilities set for
6012 the container. This makes it easier to boot unmodified
6013 Fedora systems in a container, which however still requires
6014 audit=0 to be passed on the kernel command line. Auditing in
6015 kernel and userspace is unfortunately still too broken in
6016 context of containers, hence we recommend compiling it out
6017 of the kernel or using audit=0. Hopefully this will be fixed
6018 one day for good in the kernel.
6020 * nspawn gained the new --bind= and --bind-ro= parameters to
6021 bind mount specific directories from the host into the
6024 * nspawn will now mount its own devpts file system instance
6025 into the container, in order not to leak pty devices from
6026 the host into the container.
6028 * systemd will now read the firmware boot time performance
6029 information from the EFI variables, if the used boot loader
6030 supports this, and takes it into account for boot performance
6031 analysis via "systemd-analyze". This is currently supported
6032 only in conjunction with Gummiboot, but could be supported
6033 by other boot loaders too. For details see:
6035 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
6037 * A new generator has been added that automatically mounts the
6038 EFI System Partition (ESP) to /boot, if that directory
6039 exists, is empty, and no other file system has been
6040 configured to be mounted there.
6042 * logind will now send out PrepareForSleep(false) out
6043 unconditionally, after coming back from suspend. This may be
6044 used by applications as asynchronous notification for
6045 system resume events.
6047 * "systemctl unlock-sessions" has been added, that allows
6048 unlocking the screens of all user sessions at once, similar
6049 to how "systemctl lock-sessions" already locked all users
6050 sessions. This is backed by a new D-Bus call UnlockSessions().
6052 * "loginctl seat-status" will now show the master device of a
6053 seat. (i.e. the device of a seat that needs to be around for
6054 the seat to be considered available, usually the graphics
6057 * tmpfiles gained a new "X" line type, that allows
6058 configuration of files and directories (with wildcards) that
6059 shall be excluded from automatic cleanup ("aging").
6061 * udev default rules set the device node permissions now only
6062 at "add" events, and do not change them any longer with a
6063 later "change" event.
6065 * The log messages for lid events and power/sleep keypresses
6066 now carry a message ID.
6068 * We now have a substantially larger unit test suite, but this
6069 continues to be work in progress.
6071 * udevadm hwdb gained a new --root= parameter to change the
6072 root directory to operate relative to.
6074 * logind will now issue a background sync() request to the kernel
6075 early at shutdown, so that dirty buffers are flushed to disk early
6076 instead of at the last moment, in order to optimize shutdown
6079 * A new bootctl tool has been added that is an interface for
6080 certain boot loader operations. This is currently a preview
6081 and is likely to be extended into a small mechanism daemon
6082 like timedated, localed, hostnamed, and can be used by
6083 graphical UIs to enumerate available boot options, and
6084 request boot into firmware operations.
6086 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
6087 the rest of the package. It also has been updated to work
6088 correctly in initrds.
6090 * Policykit previously has been runtime optional, and is now
6091 also compile time optional via a configure switch.
6093 * systemd-analyze has been reimplemented in C. Also "systemctl
6094 dot" has moved into systemd-analyze.
6096 * "systemctl status" with no further parameters will now print
6097 the status of all active or failed units.
6099 * Operations such as "systemctl start" can now be executed
6100 with a new mode "--irreversible" which may be used to queue
6101 operations that cannot accidentally be reversed by a later
6102 job queuing. This is by default used to make shutdown
6103 requests more robust.
6105 * The Python API of systemd now gained a new module for
6106 reading journal files.
6108 * A new tool kernel-install has been added that can install
6109 kernel images according to the Boot Loader Specification:
6111 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
6113 * Boot time console output has been improved to provide
6114 animated boot time output for hanging jobs.
6116 * A new tool systemd-activate has been added which can be used
6117 to test socket activation with, directly from the command
6118 line. This should make it much easier to test and debug
6119 socket activation in daemons.
6121 * journalctl gained a new "--reverse" (or -r) option to show
6122 journal output in reverse order (i.e. newest line first).
6124 * journalctl gained a new "--pager-end" (or -e) option to jump
6125 to immediately jump to the end of the journal in the
6126 pager. This is only supported in conjunction with "less".
6128 * journalctl gained a new "--user-unit=" option, that works
6129 similarly to "--unit=" but filters for user units rather than
6132 * A number of unit files to ease adoption of systemd in
6133 initrds has been added. This moves some minimal logic from
6134 the various initrd implementations into systemd proper.
6136 * The journal files are now owned by a new group
6137 "systemd-journal", which exists specifically to allow access
6138 to the journal, and nothing else. Previously, we used the
6139 "adm" group for that, which however possibly covers more
6140 than just journal/log file access. This new group is now
6141 already used by systemd-journal-gatewayd to ensure this
6142 daemon gets access to the journal files and as little else
6143 as possible. Note that "make install" will also set FS ACLs
6144 up for /var/log/journal to give "adm" and "wheel" read
6145 access to it, in addition to "systemd-journal" which owns
6146 the journal files. We recommend that packaging scripts also
6147 add read access to "adm" + "wheel" to /var/log/journal, and
6148 all existing/future journal files. To normal users and
6149 administrators little changes, however packagers need to
6150 ensure to create the "systemd-journal" system group at
6151 package installation time.
6153 * The systemd-journal-gatewayd now runs as unprivileged user
6154 systemd-journal-gateway:systemd-journal-gateway. Packaging
6155 scripts need to create these system user/group at
6158 * timedated now exposes a new boolean property CanNTP that
6159 indicates whether a local NTP service is available or not.
6161 * systemd-detect-virt will now also detect xen PVs
6163 * The pstore file system is now mounted by default, if it is
6166 * In addition to the SELinux and IMA policies we will now also
6167 load SMACK policies at early boot.
6169 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
6170 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
6171 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
6172 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
6173 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
6174 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
6175 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
6176 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
6177 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
6178 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
6179 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
6180 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
6181 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
6182 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
6186 * Timer units now support calendar time events in addition to
6187 monotonic time events. That means you can now trigger a unit
6188 based on a calendar time specification such as "Thu,Fri
6189 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
6190 or fifth day of any month of the year 2013, given that it is
6191 a thursday or friday. This brings timer event support
6192 considerably closer to cron's capabilities. For details on
6193 the supported calendar time specification language see
6196 * udev now supports a number of different naming policies for
6197 network interfaces for predictable names, and a combination
6198 of these policies is now the default. Please see this wiki
6199 document for details:
6201 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
6203 * Auke Kok's bootchart implementation has been added to the
6204 systemd tree. It is an optional component that can graph the
6205 boot in quite some detail. It is one of the best bootchart
6206 implementations around and minimal in its code and
6209 * nss-myhostname has been integrated into the systemd source
6210 tree. nss-myhostname guarantees that the local hostname
6211 always stays resolvable via NSS. It has been a weak
6212 requirement of systemd-hostnamed since a long time, and
6213 since its code is actually trivial we decided to just
6214 include it in systemd's source tree. It can be turned off
6215 with a configure switch.
6217 * The read-ahead logic is now capable of properly detecting
6218 whether a btrfs file system is on SSD or rotating media, in
6219 order to optimize the read-ahead scheme. Previously, it was
6220 only capable of detecting this on traditional file systems
6223 * In udev, additional device properties are now read from the
6224 IAB in addition to the OUI database. Also, Bluetooth company
6225 identities are attached to the devices as well.
6227 * In service files %U may be used as specifier that is
6228 replaced by the configured user name of the service.
6230 * nspawn may now be invoked without a controlling TTY. This
6231 makes it suitable for invocation as its own service. This
6232 may be used to set up a simple containerized server system
6233 using only core OS tools.
6235 * systemd and nspawn can now accept socket file descriptors
6236 when they are started for socket activation. This enables
6237 implementation of socket activated nspawn
6238 containers. i.e. think about autospawning an entire OS image
6239 when the first SSH or HTTP connection is received. We expect
6240 that similar functionality will also be added to libvirt-lxc
6243 * journalctl will now suppress ANSI color codes when
6244 presenting log data.
6246 * systemctl will no longer show control group information for
6247 a unit if the control group is empty anyway.
6249 * logind can now automatically suspend/hibernate/shutdown the
6252 * /etc/machine-info and hostnamed now also expose the chassis
6253 type of the system. This can be used to determine whether
6254 the local system is a laptop, desktop, handset or
6255 tablet. This information may either be configured by the
6256 user/vendor or is automatically determined from ACPI and DMI
6257 information if possible.
6259 * A number of PolicyKit actions are now bound together with
6260 "imply" rules. This should simplify creating UIs because
6261 many actions will now authenticate similar ones as well.
6263 * Unit files learnt a new condition ConditionACPower= which
6264 may be used to conditionalize a unit depending on whether an
6265 AC power source is connected or not, of whether the system
6266 is running on battery power.
6268 * systemctl gained a new "is-failed" verb that may be used in
6269 shell scripts and suchlike to check whether a specific unit
6270 is in the "failed" state.
6272 * The EnvironmentFile= setting in unit files now supports file
6273 globbing, and can hence be used to easily read a number of
6274 environment files at once.
6276 * systemd will no longer detect and recognize specific
6277 distributions. All distribution-specific #ifdeffery has been
6278 removed, systemd is now fully generic and
6279 distribution-agnostic. Effectively, not too much is lost as
6280 a lot of the code is still accessible via explicit configure
6281 switches. However, support for some distribution specific
6282 legacy configuration file formats has been dropped. We
6283 recommend distributions to simply adopt the configuration
6284 files everybody else uses now and convert the old
6285 configuration from packaging scripts. Most distributions
6286 already did that. If that's not possible or desirable,
6287 distributions are welcome to forward port the specific
6288 pieces of code locally from the git history.
6290 * When logging a message about a unit systemd will now always
6291 log the unit name in the message meta data.
6293 * localectl will now also discover system locale data that is
6294 not stored in locale archives, but directly unpacked.
6296 * logind will no longer unconditionally use framebuffer
6297 devices as seat masters, i.e. as devices that are required
6298 to be existing before a seat is considered preset. Instead,
6299 it will now look for all devices that are tagged as
6300 "seat-master" in udev. By default, framebuffer devices will
6301 be marked as such, but depending on local systems, other
6302 devices might be marked as well. This may be used to
6303 integrate graphics cards using closed source drivers (such
6304 as NVidia ones) more nicely into logind. Note however, that
6305 we recommend using the open source NVidia drivers instead,
6306 and no udev rules for the closed-source drivers will be
6307 shipped from us upstream.
6309 Contributions from: Adam Williamson, Alessandro Crismani, Auke
6310 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
6311 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
6312 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
6313 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
6314 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
6315 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
6316 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
6317 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
6318 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
6319 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
6320 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
6325 * udev gained support for loading additional device properties
6326 from an indexed database that is keyed by vendor/product IDs
6327 and similar device identifiers. For the beginning this
6328 "hwdb" is populated with data from the well-known PCI and
6329 USB database, but also includes PNP, ACPI and OID data. In
6330 the longer run this indexed database shall grow into
6331 becoming the one central database for non-essential
6332 userspace device metadata. Previously, data from the PCI/USB
6333 database was only attached to select devices, since the
6334 lookup was a relatively expensive operation due to O(n) time
6335 complexity (with n being the number of entries in the
6336 database). Since this is now O(1), we decided to add in this
6337 data for all devices where this is available, by
6338 default. Note that the indexed database needs to be rebuilt
6339 when new data files are installed. To achieve this you need
6340 to update your packaging scripts to invoke "udevadm hwdb
6341 --update" after installation of hwdb data files. For
6342 RPM-based distributions we introduced the new
6343 %udev_hwdb_update macro for this purpose.
6345 * The Journal gained support for the "Message Catalog", an
6346 indexed database to link up additional information with
6347 journal entries. For further details please check:
6349 https://www.freedesktop.org/wiki/Software/systemd/catalog
6351 The indexed message catalog database also needs to be
6352 rebuilt after installation of message catalog files. Use
6353 "journalctl --update-catalog" for this. For RPM-based
6354 distributions we introduced the %journal_catalog_update
6355 macro for this purpose.
6357 * The Python Journal bindings gained support for the standard
6358 Python logging framework.
6360 * The Journal API gained new functions for checking whether
6361 the underlying file system of a journal file is capable of
6362 properly reporting file change notifications, or whether
6363 applications that want to reflect journal changes "live"
6364 need to recheck journal files continuously in appropriate
6367 * It is now possible to set the "age" field for tmpfiles
6368 entries to 0, indicating that files matching this entry
6369 shall always be removed when the directories are cleaned up.
6371 * coredumpctl gained a new "gdb" verb which invokes gdb
6372 right-away on the selected coredump.
6374 * There's now support for "hybrid sleep" on kernels that
6375 support this, in addition to "suspend" and "hibernate". Use
6376 "systemctl hybrid-sleep" to make use of this.
6378 * logind's HandleSuspendKey= setting (and related settings)
6379 now gained support for a new "lock" setting to simply
6380 request the screen lock on all local sessions, instead of
6381 actually executing a suspend or hibernation.
6383 * systemd will now mount the EFI variables file system by
6386 * Socket units now gained support for configuration of the
6387 SMACK security label.
6389 * timedatectl will now output the time of the last and next
6390 daylight saving change.
6392 * We dropped support for various legacy and distro-specific
6393 concepts, such as insserv, early-boot SysV services
6394 (i.e. those for non-standard runlevels such as 'b' or 'S')
6395 or ArchLinux /etc/rc.conf support. We recommend the
6396 distributions who still need support this to either continue
6397 to maintain the necessary patches downstream, or find a
6398 different solution. (Talk to us if you have questions!)
6400 * Various systemd components will now bypass PolicyKit checks
6401 for root and otherwise handle properly if PolicyKit is not
6402 found to be around. This should fix most issues for
6403 PolicyKit-less systems. Quite frankly this should have been
6404 this way since day one. It is absolutely our intention to
6405 make systemd work fine on PolicyKit-less systems, and we
6406 consider it a bug if something does not work as it should if
6407 PolicyKit is not around.
6409 * For embedded systems it is now possible to build udev and
6410 systemd without blkid and/or kmod support.
6412 * "systemctl switch-root" is now capable of switching root
6413 more than once. I.e. in addition to transitions from the
6414 initrd to the host OS it is now possible to transition to
6415 further OS images from the host. This is useful to implement
6416 offline updating tools.
6418 * Various other additions have been made to the RPM macros
6419 shipped with systemd. Use %udev_rules_update() after
6420 installing new udev rules files. %_udevhwdbdir,
6421 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
6422 %_sysctldir are now available which resolve to the right
6423 directories for packages to place various data files in.
6425 * journalctl gained the new --full switch (in addition to
6426 --all, to disable ellipsation for long messages.
6428 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
6429 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
6430 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
6431 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
6432 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
6433 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
6434 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
6435 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
6436 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
6440 * journalctl gained new --since= and --until= switches to
6441 filter by time. It also now supports nice filtering for
6442 units via --unit=/-u.
6444 * Type=oneshot services may use ExecReload= and do the
6447 * The journal daemon now supports time-based rotation and
6448 vacuuming, in addition to the usual disk-space based
6451 * The journal will now index the available field values for
6452 each field name. This enables clients to show pretty drop
6453 downs of available match values when filtering. The bash
6454 completion of journalctl has been updated
6455 accordingly. journalctl gained a new switch -F to list all
6456 values a certain field takes in the journal database.
6458 * More service events are now written as structured messages
6459 to the journal, and made recognizable via message IDs.
6461 * The timedated, localed and hostnamed mini-services which
6462 previously only provided support for changing time, locale
6463 and hostname settings from graphical DEs such as GNOME now
6464 also have a minimal (but very useful) text-based client
6465 utility each. This is probably the nicest way to changing
6466 these settings from the command line now, especially since
6467 it lists available options and is fully integrated with bash
6470 * There's now a new tool "systemd-coredumpctl" to list and
6471 extract coredumps from the journal.
6473 * We now install a README each in /var/log/ and
6474 /etc/rc.d/init.d explaining where the system logs and init
6475 scripts went. This hopefully should help folks who go to
6476 that dirs and look into the otherwise now empty void and
6477 scratch their heads.
6479 * When user-services are invoked (by systemd --user) the
6480 $MANAGERPID env var is set to the PID of systemd.
6482 * SIGRTMIN+24 when sent to a --user instance will now result
6483 in immediate termination of systemd.
6485 * gatewayd received numerous feature additions such as a
6486 "follow" mode, for live syncing and filtering.
6488 * browse.html now allows filtering and showing detailed
6489 information on specific entries. Keyboard navigation and
6490 mouse screen support has been added.
6492 * gatewayd/journalctl now supports HTML5/JSON
6493 Server-Sent-Events as output.
6495 * The SysV init script compatibility logic will now
6496 heuristically determine whether a script supports the
6497 "reload" verb, and only then make this available as
6500 * "systemctl status --follow" has been removed, use "journalctl
6503 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
6504 have been removed since they are hardly useful to be
6507 * And I'd like to take the opportunity to specifically mention
6508 Zbigniew for his great contributions. Zbigniew, you rock!
6510 Contributions from: Andrew Eikum, Christian Hesse, Colin
6511 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
6512 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
6513 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
6514 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
6515 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
6516 Jędrzejewski-Szmek, Сковорода Никита Андреевич
6520 * If /etc/vconsole.conf is non-existent or empty we will no
6521 longer load any console font or key map at boot by
6522 default. Instead the kernel defaults will be left
6523 intact. This is definitely the right thing to do, as no
6524 configuration should mean no configuration, and hard-coding
6525 font names that are different on all archs is probably a bad
6526 idea. Also, the kernel default key map and font should be
6527 good enough for most cases anyway, and mostly identical to
6528 the userspace fonts/key maps we previously overloaded them
6529 with. If distributions want to continue to default to a
6530 non-kernel font or key map they should ship a default
6531 /etc/vconsole.conf with the appropriate contents.
6533 Contributions from: Colin Walters, Daniel J Walsh, Dave
6534 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
6535 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6539 * journalctl gained a new --cursor= switch to show entries
6540 starting from the specified location in the journal.
6542 * We now enforce a size limit on journal entry fields exported
6543 with "-o json" in journalctl. Fields larger than 4K will be
6544 assigned null. This can be turned off with --all.
6546 * An (optional) journal gateway daemon is now available as
6547 "systemd-journal-gatewayd.service". This service provides
6548 access to the journal via HTTP and JSON. This functionality
6549 will be used to implement live log synchronization in both
6550 pull and push modes, but has various other users too, such
6551 as easy log access for debugging of embedded devices. Right
6552 now it is already useful to retrieve the journal via HTTP:
6554 # systemctl start systemd-journal-gatewayd.service
6555 # wget http://localhost:19531/entries
6557 This will download the journal contents in a
6558 /var/log/messages compatible format. The same as JSON:
6560 # curl -H"Accept: application/json" http://localhost:19531/entries
6562 This service is also accessible via a web browser where a
6563 single static HTML5 app is served that uses the JSON logic
6564 to enable the user to do some basic browsing of the
6565 journal. This will be extended later on. Here's an example
6566 screenshot of this app in its current state:
6568 http://0pointer.de/public/journal-gatewayd
6570 Contributions from: Kay Sievers, Lennart Poettering, Robert
6571 Milasan, Tom Gundersen
6575 * The bash completion logic is now available for journalctl
6578 * We do not mount the "cpuset" controller anymore together with
6579 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
6580 started if no parameters are assigned to it. "cpuset" hence
6581 broke code that assumed it could create "cpu" groups and
6584 * journalctl -f will now subscribe to terminal size changes,
6585 and line break accordingly.
6587 Contributions from: Dave Reisner, Kay Sievers, Lennart
6588 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
6592 * nspawn will now create a symlink /etc/localtime in the
6593 container environment, copying the host's timezone
6594 setting. Previously this has been done via a bind mount, but
6595 since symlinks cannot be bind mounted this has now been
6596 changed to create/update the appropriate symlink.
6598 * journalctl -n's line number argument is now optional, and
6599 will default to 10 if omitted.
6601 * journald will now log the maximum size the journal files may
6602 take up on disk. This is particularly useful if the default
6603 built-in logic of determining this parameter from the file
6604 system size is used. Use "systemctl status
6605 systemd-journald.service" to see this information.
6607 * The multi-seat X wrapper tool has been stripped down. As X
6608 is now capable of enumerating graphics devices via udev in a
6609 seat-aware way the wrapper is not strictly necessary
6610 anymore. A stripped down temporary stop-gap is still shipped
6611 until the upstream display managers have been updated to
6612 fully support the new X logic. Expect this wrapper to be
6613 removed entirely in one of the next releases.
6615 * HandleSleepKey= in logind.conf has been split up into
6616 HandleSuspendKey= and HandleHibernateKey=. The old setting
6617 is not available anymore. X11 and the kernel are
6618 distinguishing between these keys and we should too. This
6619 also means the inhibition lock for these keys has been split
6622 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
6623 Poettering, Lukas Nykryn, Václav Pavlín
6627 * Whenever a unit changes state we will now log this to the
6628 journal and show along the unit's own log output in
6631 * ConditionPathIsMountPoint= can now properly detect bind
6632 mount points too. (Previously, a bind mount of one file
6633 system to another place in the same file system could not be
6634 detected as mount, since they shared struct stat's st_dev
6637 * We will now mount the cgroup controllers cpu, cpuacct,
6638 cpuset and the controllers net_cls, net_prio together by
6641 * nspawn containers will now have a virtualized boot
6642 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
6643 over with a randomized ID at container initialization). This
6644 has the effect of making "journalctl -b" do the right thing
6647 * The JSON output journal serialization has been updated not
6648 to generate "endless" list objects anymore, but rather one
6649 JSON object per line. This is more in line how most JSON
6650 parsers expect JSON objects. The new output mode
6651 "json-pretty" has been added to provide similar output, but
6652 neatly aligned for readability by humans.
6654 * We dropped all explicit sync() invocations in the shutdown
6655 code. The kernel does this implicitly anyway in the kernel
6656 reboot() syscall. halt(8)'s -n option is now a compatibility
6659 * We now support virtualized reboot() in containers, as
6660 supported by newer kernels. We will fall back to exit() if
6661 CAP_SYS_REBOOT is not available to the container. Also,
6662 nspawn makes use of this now and will actually reboot the
6663 container if the containerized OS asks for that.
6665 * journalctl will only show local log output by default
6666 now. Use --merge (-m) to show remote log output, too.
6668 * libsystemd-journal gained the new sd_journal_get_usage()
6669 call to determine the current disk usage of all journal
6670 files. This is exposed in the new "journalctl --disk-usage"
6673 * journald gained a new configuration setting SplitMode= in
6674 journald.conf which may be used to control how user journals
6675 are split off. See journald.conf(5) for details.
6677 * A new condition type ConditionFileNotEmpty= has been added.
6679 * tmpfiles' "w" lines now support file globbing, to write
6680 multiple files at once.
6682 * We added Python bindings for the journal submission
6683 APIs. More Python APIs for a number of selected APIs will
6684 likely follow. Note that we intend to add native bindings
6685 only for the Python language, as we consider it common
6686 enough to deserve bindings shipped within systemd. There are
6687 various projects outside of systemd that provide bindings
6688 for languages such as PHP or Lua.
6690 * Many conditions will now resolve specifiers such as %i. In
6691 addition, PathChanged= and related directives of .path units
6692 now support specifiers as well.
6694 * There's now a new RPM macro definition for the system preset
6697 * journald will now warn if it ca not forward a message to the
6698 syslog daemon because its socket is full.
6700 * timedated will no longer write or process /etc/timezone,
6701 except on Debian. As we do not support late mounted /usr
6702 anymore /etc/localtime always being a symlink is now safe,
6703 and hence the information in /etc/timezone is not necessary
6706 * logind will now always reserve one VT for a text getty (VT6
6707 by default). Previously if more than 6 X sessions where
6708 started they took up all the VTs with auto-spawned gettys,
6709 so that no text gettys were available anymore.
6711 * udev will now automatically inform the btrfs kernel logic
6712 about btrfs RAID components showing up. This should make
6713 simple hotplug based btrfs RAID assembly work.
6715 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
6716 (but not for its children which will stay at the kernel
6717 default). This should allow setups with a lot more listening
6720 * systemd will now always pass the configured timezone to the
6721 kernel at boot. timedated will do the same when the timezone
6724 * logind's inhibition logic has been updated. By default,
6725 logind will now handle the lid switch, the power and sleep
6726 keys all the time, even in graphical sessions. If DEs want
6727 to handle these events on their own they should take the new
6728 handle-power-key, handle-sleep-key and handle-lid-switch
6729 inhibitors during their runtime. A simple way to achieve
6730 that is to invoke the DE wrapped in an invocation of:
6732 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
6734 * Access to unit operations is now checked via SELinux taking
6735 the unit file label and client process label into account.
6737 * systemd will now notify the administrator in the journal
6738 when he over-mounts a non-empty directory.
6740 * There are new specifiers that are resolved in unit files,
6741 for the host name (%H), the machine ID (%m) and the boot ID
6744 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
6745 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
6746 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
6747 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
6748 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
6749 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
6750 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
6754 * Support for reading structured kernel messages from
6755 /dev/kmsg has now been added and is enabled by default.
6757 * Support for reading kernel messages from /proc/kmsg has now
6758 been removed. If you want kernel messages in the journal
6759 make sure to run a recent kernel (>= 3.5) that supports
6760 reading structured messages from /dev/kmsg (see
6761 above). /proc/kmsg is now exclusive property of classic
6762 syslog daemons again.
6764 * The libudev API gained the new
6765 udev_device_new_from_device_id() call.
6767 * The logic for file system namespace (ReadOnlyDirectory=,
6768 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
6769 require pivot_root() anymore. This means fewer temporary
6770 directories are created below /tmp for this feature.
6772 * nspawn containers will now see and receive all submounts
6773 made on the host OS below the root file system of the
6776 * Forward Secure Sealing is now supported for Journal files,
6777 which provide cryptographical sealing of journal files so
6778 that attackers cannot alter log history anymore without this
6779 being detectable. Lennart will soon post a blog story about
6780 this explaining it in more detail.
6782 * There are two new service settings RestartPreventExitStatus=
6783 and SuccessExitStatus= which allow configuration of exit
6784 status (exit code or signal) which will be excepted from the
6785 restart logic, resp. consider successful.
6787 * journalctl gained the new --verify switch that can be used
6788 to check the integrity of the structure of journal files and
6789 (if Forward Secure Sealing is enabled) the contents of
6792 * nspawn containers will now be run with /dev/stdin, /dev/fd/
6793 and similar symlinks pre-created. This makes running shells
6794 as container init process a lot more fun.
6796 * The fstab support can now handle PARTUUID= and PARTLABEL=
6799 * A new ConditionHost= condition has been added to match
6800 against the hostname (with globs) and machine ID. This is
6801 useful for clusters where a single OS image is used to
6802 provision a large number of hosts which shall run slightly
6803 different sets of services.
6805 * Services which hit the restart limit will now be placed in a
6808 Contributions from: Bertram Poettering, Dave Reisner, Huang
6809 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
6810 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
6814 * When running in --user mode systemd will now become a
6815 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
6816 tree a lot more organized.
6818 * A new PartOf= unit dependency type has been introduced that
6819 may be used to group services in a natural way.
6821 * "systemctl enable" may now be used to enable instances of
6824 * journalctl now prints error log levels in red, and
6825 warning/notice log levels in bright white. It also supports
6826 filtering by log level now.
6828 * cgtop gained a new -n switch (similar to top), to configure
6829 the maximum number of iterations to run for. It also gained
6830 -b, to run in batch mode (accepting no input).
6832 * The suffix ".service" may now be omitted on most systemctl
6833 command lines involving service unit names.
6835 * There's a new bus call in logind to lock all sessions, as
6836 well as a loginctl verb for it "lock-sessions".
6838 * libsystemd-logind.so gained a new call sd_journal_perror()
6839 that works similar to libc perror() but logs to the journal
6840 and encodes structured information about the error number.
6842 * /etc/crypttab entries now understand the new keyfile-size=
6845 * shutdown(8) now can send a (configurable) wall message when
6846 a shutdown is cancelled.
6848 * The mount propagation mode for the root file system will now
6849 default to "shared", which is useful to make containers work
6850 nicely out-of-the-box so that they receive new mounts from
6851 the host. This can be undone locally by running "mount
6852 --make-rprivate /" if needed.
6854 * The prefdm.service file has been removed. Distributions
6855 should maintain this unit downstream if they intend to keep
6856 it around. However, we recommend writing normal unit files
6857 for display managers instead.
6859 * Since systemd is a crucial part of the OS we will now
6860 default to a number of compiler switches that improve
6861 security (hardening) such as read-only relocations, stack
6862 protection, and suchlike.
6864 * The TimeoutSec= setting for services is now split into
6865 TimeoutStartSec= and TimeoutStopSec= to allow configuration
6866 of individual time outs for the start and the stop phase of
6869 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
6870 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
6871 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
6872 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
6873 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
6874 Gundersen, Zbigniew Jędrzejewski-Szmek
6878 * The journal and id128 C APIs are now fully documented as man
6881 * Extra safety checks have been added when transitioning from
6882 the initial RAM disk to the main system to avoid accidental
6885 * /etc/crypttab entries now understand the new keyfile-offset=
6888 * systemctl -t can now be used to filter by unit load state.
6890 * The journal C API gained the new sd_journal_wait() call to
6891 make writing synchronous journal clients easier.
6893 * journalctl gained the new -D switch to show journals from a
6896 * journalctl now displays a special marker between log
6897 messages of two different boots.
6899 * The journal is now explicitly flushed to /var via a service
6900 systemd-journal-flush.service, rather than implicitly simply
6901 by seeing /var/log/journal to be writable.
6903 * journalctl (and the journal C APIs) can now match for much
6904 more complex expressions, with alternatives and
6907 * When transitioning from the initial RAM disk to the main
6908 system we will now kill all processes in a killing spree to
6909 ensure no processes stay around by accident.
6911 * Three new specifiers may be used in unit files: %u, %h, %s
6912 resolve to the user name, user home directory resp. user
6913 shell. This is useful for running systemd user instances.
6915 * We now automatically rotate journal files if their data
6916 object hash table gets a fill level > 75%. We also size the
6917 hash table based on the configured maximum file size. This
6918 together should lower hash collisions drastically and thus
6919 speed things up a bit.
6921 * journalctl gained the new "--header" switch to introspect
6922 header data of journal files.
6924 * A new setting SystemCallFilters= has been added to services
6925 which may be used to apply blacklists or whitelists to
6926 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
6928 * nspawn gained a new --link-journal= switch (and quicker: -j)
6929 to link the container journal with the host. This makes it
6930 very easy to centralize log viewing on the host for all
6931 guests while still keeping the journal files separated.
6933 * Many bugfixes and optimizations
6935 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
6936 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
6937 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
6942 * Several tools now understand kernel command line arguments,
6943 which are only read when run in an initial RAM disk. They
6944 usually follow closely their normal counterparts, but are
6947 * There's a new tool to analyze the readahead files that are
6948 automatically generated at boot. Use:
6950 /usr/lib/systemd/systemd-readahead analyze /.readahead
6952 * We now provide an early debug shell on tty9 if this enabled. Use:
6954 systemctl enable debug-shell.service
6956 * All plymouth related units have been moved into the Plymouth
6957 package. Please make sure to upgrade your Plymouth version
6960 * systemd-tmpfiles now supports getting passed the basename of
6961 a configuration file only, in which case it will look for it
6962 in all appropriate directories automatically.
6964 * udevadm info now takes a /dev or /sys path as argument, and
6965 does the right thing. Example:
6967 udevadm info /dev/sda
6968 udevadm info /sys/class/block/sda
6970 * systemctl now prints a warning if a unit is stopped but a
6971 unit that might trigger it continues to run. Example: a
6972 service is stopped but the socket that activates it is left
6975 * "systemctl status" will now mention if the log output was
6976 shortened due to rotation since a service has been started.
6978 * The journal API now exposes functions to determine the
6979 "cutoff" times due to rotation.
6981 * journald now understands SIGUSR1 and SIGUSR2 for triggering
6982 immediately flushing of runtime logs to /var if possible,
6983 resp. for triggering immediate rotation of the journal
6986 * It is now considered an error if a service is attempted to
6987 be stopped that is not loaded.
6989 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
6991 * systemd-analyze now supports Python 3
6993 * tmpfiles now supports cleaning up directories via aging
6994 where the first level dirs are always kept around but
6995 directories beneath it automatically aged. This is enabled
6996 by prefixing the age field with '~'.
6998 * Seat objects now expose CanGraphical, CanTTY properties
6999 which is required to deal with very fast bootups where the
7000 display manager might be running before the graphics drivers
7001 completed initialization.
7003 * Seat objects now expose a State property.
7005 * We now include RPM macros for service enabling/disabling
7006 based on the preset logic. We recommend RPM based
7007 distributions to make use of these macros if possible. This
7008 makes it simpler to reuse RPM spec files across
7011 * We now make sure that the collected systemd unit name is
7012 always valid when services log to the journal via
7015 * There's a new man page kernel-command-line(7) detailing all
7016 command line options we understand.
7018 * The fstab generator may now be disabled at boot by passing
7019 fstab=0 on the kernel command line.
7021 * A new kernel command line option modules-load= is now understood
7022 to load a specific kernel module statically, early at boot.
7024 * Unit names specified on the systemctl command line are now
7025 automatically escaped as needed. Also, if file system or
7026 device paths are specified they are automatically turned
7027 into the appropriate mount or device unit names. Example:
7029 systemctl status /home
7030 systemctl status /dev/sda
7032 * The SysVConsole= configuration option has been removed from
7033 system.conf parsing.
7035 * The SysV search path is no longer exported on the D-Bus
7038 * The Names= option has been removed from unit file parsing.
7040 * There's a new man page bootup(7) detailing the boot process.
7042 * Every unit and every generator we ship with systemd now
7043 comes with full documentation. The self-explanatory boot is
7046 * A couple of services gained "systemd-" prefixes in their
7047 name if they wrap systemd code, rather than only external
7048 code. Among them fsck@.service which is now
7049 systemd-fsck@.service.
7051 * The HaveWatchdog property has been removed from the D-Bus
7054 * systemd.confirm_spawn= on the kernel command line should now
7057 * There's a new man page crypttab(5) which details all options
7058 we actually understand.
7060 * systemd-nspawn gained a new --capability= switch to pass
7061 additional capabilities to the container.
7063 * timedated will now read known NTP implementation unit names
7064 from /usr/lib/systemd/ntp-units.d/*.list,
7065 systemd-timedated-ntp.target has been removed.
7067 * journalctl gained a new switch "-b" that lists log data of
7068 the current boot only.
7070 * The notify socket is in the abstract namespace again, in
7071 order to support daemons which chroot() at start-up.
7073 * There is a new Storage= configuration option for journald
7074 which allows configuration of where log data should go. This
7075 also provides a way to disable journal logging entirely, so
7076 that data collected is only forwarded to the console, the
7077 kernel log buffer or another syslog implementation.
7079 * Many bugfixes and optimizations
7081 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
7082 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
7083 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
7084 Shawn Landden, Tom Gundersen
7088 * "systemctl help <unit>" now shows the man page if one is
7091 * Several new man pages have been added.
7093 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
7094 MaxLevelConsole= can now be specified in
7095 journald.conf. These options allow reducing the amount of
7096 data stored on disk or forwarded by the log level.
7098 * TimerSlackNSec= can now be specified in system.conf for
7099 PID1. This allows system-wide power savings.
7101 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
7102 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
7107 * logind is now capable of (optionally) handling power and
7108 sleep keys as well as the lid switch.
7110 * journalctl now understands the syntax "journalctl
7111 /usr/bin/avahi-daemon" to get all log output of a specific
7114 * CapabilityBoundingSet= in system.conf now also influences
7115 the capability bound set of usermode helpers of the kernel.
7117 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
7118 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
7119 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
7120 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
7124 * Note that we skipped 139 releases here in order to set the
7125 new version to something that is greater than both udev's
7126 and systemd's most recent version number.
7128 * udev: all udev sources are merged into the systemd source tree now.
7129 All future udev development will happen in the systemd tree. It
7130 is still fully supported to use the udev daemon and tools without
7131 systemd running, like in initramfs or other init systems. Building
7132 udev though, will require the *build* of the systemd tree, but
7133 udev can be properly *run* without systemd.
7135 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
7136 should be used to create dead device nodes as workarounds for broken
7139 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
7140 no longer supported. udev_monitor_new_from_netlink() needs to be
7141 used to subscribe to events.
7143 * udev: when udevd is started by systemd, processes which are left
7144 behind by forking them off of udev rules, are unconditionally cleaned
7145 up and killed now after the event handling has finished. Services or
7146 daemons must be started as systemd services. Services can be
7147 pulled-in by udev to get started, but they can no longer be directly
7148 forked by udev rules.
7150 * udev: the daemon binary is called systemd-udevd now and installed
7151 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
7152 to adapt to that, create symlink, or rename the binary after building
7155 * libudev no longer provides these symbols:
7156 udev_monitor_from_socket()
7157 udev_queue_get_failed_list_entry()
7158 udev_get_{dev,sys,run}_path()
7159 The versions number was bumped and symbol versioning introduced.
7161 * systemd-loginctl and systemd-journalctl have been renamed
7162 to loginctl and journalctl to match systemctl.
7164 * The config files: /etc/systemd/systemd-logind.conf and
7165 /etc/systemd/systemd-journald.conf have been renamed to
7166 logind.conf and journald.conf. Package updates should rename
7167 the files to the new names on upgrade.
7169 * For almost all files the license is now LGPL2.1+, changed
7170 from the previous GPL2.0+. Exceptions are some minor stuff
7171 of udev (which will be changed to LGPL2.1 eventually, too),
7172 and the MIT licensed sd-daemon.[ch] library that is suitable
7173 to be used as drop-in files.
7175 * systemd and logind now handle system sleep states, in
7176 particular suspending and hibernating.
7178 * logind now implements a sleep/shutdown/idle inhibiting logic
7179 suitable for a variety of uses. Soonishly Lennart will blog
7180 about this in more detail.
7182 * var-run.mount and var-lock.mount are no longer provided
7183 (which previously bind mounted these directories to their new
7184 places). Distributions which have not converted these
7185 directories to symlinks should consider stealing these files
7186 from git history and add them downstream.
7188 * We introduced the Documentation= field for units and added
7189 this to all our shipped units. This is useful to make it
7190 easier to explore the boot and the purpose of the various
7193 * All smaller setup units (such as
7194 systemd-vconsole-setup.service) now detect properly if they
7195 are run in a container and are skipped when
7196 appropriate. This guarantees an entirely noise-free boot in
7197 Linux container environments such as systemd-nspawn.
7199 * A framework for implementing offline system updates is now
7200 integrated, for details see:
7201 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
7203 * A new service type Type=idle is available now which helps us
7204 avoiding ugly interleaving of getty output and boot status
7207 * There's now a system-wide CapabilityBoundingSet= option to
7208 globally reduce the set of capabilities for the
7209 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
7210 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
7211 even CAP_NET_ADMIN system-wide for secure systems.
7213 * There are now system-wide DefaultLimitXXX= options to
7214 globally change the defaults of the various resource limits
7215 for all units started by PID 1.
7217 * Harald Hoyer's systemd test suite has been integrated into
7218 systemd which allows easy testing of systemd builds in qemu
7219 and nspawn. (This is really awesome! Ask us for details!)
7221 * The fstab parser is now implemented as generator, not inside
7224 * systemctl will now warn you if .mount units generated from
7225 /etc/fstab are out of date due to changes in fstab that
7226 have not been read by systemd yet.
7228 * systemd is now suitable for usage in initrds. Dracut has
7229 already been updated to make use of this. With this in place
7230 initrds get a slight bit faster but primarily are much
7231 easier to introspect and debug since "systemctl status" in
7232 the host system can be used to introspect initrd services,
7233 and the journal from the initrd is kept around too.
7235 * systemd-delta has been added, a tool to explore differences
7236 between user/admin configuration and vendor defaults.
7238 * PrivateTmp= now affects both /tmp and /var/tmp.
7240 * Boot time status messages are now much prettier and feature
7241 proper english language. Booting up systemd has never been
7244 * Read-ahead pack files now include the inode number of all
7245 files to pre-cache. When the inode changes the pre-caching
7246 is not attempted. This should be nicer to deal with updated
7247 packages which might result in changes of read-ahead
7250 * We now temporaritly lower the kernel's read_ahead_kb variable
7251 when collecting read-ahead data to ensure the kernel's
7252 built-in read-ahead does not add noise to our measurements
7253 of necessary blocks to pre-cache.
7255 * There's now RequiresMountsFor= to add automatic dependencies
7256 for all mounts necessary for a specific file system path.
7258 * MountAuto= and SwapAuto= have been removed from
7259 system.conf. Mounting file systems at boot has to take place
7262 * nspawn now learned a new switch --uuid= to set the machine
7263 ID on the command line.
7265 * nspawn now learned the -b switch to automatically search
7268 * vt102 is now the default TERM for serial TTYs, upgraded from
7271 * systemd-logind now works on VT-less systems.
7273 * The build tree has been reorganized. The individual
7274 components now have directories of their own.
7276 * A new condition type ConditionPathIsReadWrite= is now available.
7278 * nspawn learned the new -C switch to create cgroups for the
7279 container in other hierarchies.
7281 * We now have support for hardware watchdogs, configurable in
7284 * The scheduled shutdown logic now has a public API.
7286 * We now mount /tmp as tmpfs by default, but this can be
7287 masked and /etc/fstab can override it.
7289 * Since udisks does not make use of /media anymore we are not
7290 mounting a tmpfs on it anymore.
7292 * journalctl gained a new --local switch to only interleave
7293 locally generated journal files.
7295 * We can now load the IMA policy at boot automatically.
7297 * The GTK tools have been split off into a systemd-ui.
7299 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
7300 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
7301 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
7302 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
7303 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
7304 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
7305 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
7306 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
7307 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
7312 * This is mostly a bugfix release
7314 * Support optional initialization of the machine ID from the
7315 KVM or container configured UUID.
7317 * Support immediate reboots with "systemctl reboot -ff"
7319 * Show /etc/os-release data in systemd-analyze output
7321 * Many bugfixes for the journal, including endianness fixes and
7322 ensuring that disk space enforcement works
7324 * sd-login.h is C++ compatible again
7326 * Extend the /etc/os-release format on request of the Debian
7329 * We now refuse non-UTF8 strings used in various configuration
7330 and unit files. This is done to ensure we do not pass invalid
7331 data over D-Bus or expose it elsewhere.
7333 * Register Mimo USB Screens as suitable for automatic seat
7336 * Read SELinux client context from journal clients in a race
7339 * Reorder configuration file lookup order. /etc now always
7340 overrides /run in order to allow the administrator to always
7341 and unconditionally override vendor-supplied or
7342 automatically generated data.
7344 * The various user visible bits of the journal now have man
7345 pages. We still lack man pages for the journal API calls
7348 * We now ship all man pages in HTML format again in the
7351 Contributions from: Dave Reisner, Dirk Eibach, Frederic
7352 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
7353 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
7358 * This is mostly a bugfix release
7360 * systems lacking /etc/os-release are no longer supported.
7362 * Various functionality updates to libsystemd-login.so
7364 * Track class of PAM logins to distinguish greeters from
7367 Contributions from: Kay Sievers, Lennart Poettering, Michael
7372 * This is an important bugfix release for v41.
7374 * Building man pages is now optional which should be useful
7375 for those building systemd from git but unwilling to install
7378 * Watchdog support for supervising services is now usable. In
7379 a future release support for hardware watchdogs
7380 (i.e. /dev/watchdog) will be added building on this.
7382 * Service start rate limiting is now configurable and can be
7383 turned off per service. When a start rate limit is hit a
7384 reboot can automatically be triggered.
7386 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
7388 Contributions from: Benjamin Franzke, Bill Nottingham,
7389 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
7390 Schmidt, Michał Górny, Piotr Drąg
7394 * The systemd binary is installed /usr/lib/systemd/systemd now;
7395 An existing /sbin/init symlink needs to be adapted with the
7398 * The code that loads kernel modules has been ported to invoke
7399 libkmod directly, instead of modprobe. This means we do not
7400 support systems with module-init-tools anymore.
7402 * Watchdog support is now already useful, but still not
7405 * A new kernel command line option systemd.setenv= is
7406 understood to set system wide environment variables
7407 dynamically at boot.
7409 * We now limit the set of capabilities of systemd-journald.
7411 * We now set SIGPIPE to ignore by default, since it only is
7412 useful in shell pipelines, and has little use in general
7413 code. This can be disabled with IgnoreSIPIPE=no in unit
7416 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
7417 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
7422 * This is mostly a bugfix release
7424 * We now expose the reason why a service failed in the
7425 "Result" D-Bus property.
7427 * Rudimentary service watchdog support (will be completed over
7428 the next few releases.)
7430 * When systemd forks off in order execute some service we will
7431 now immediately changes its argv[0] to reflect which process
7432 it will execute. This is useful to minimize the time window
7433 with a generic argv[0], which makes bootcharts more useful
7435 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
7436 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
7437 Mike Kazantsev, Ray Strode
7441 * This is mostly a test release, but incorporates many
7444 * New systemd-cgtop tool to show control groups by their
7447 * Linking against libacl for ACLs is optional again. If
7448 disabled, support tracking device access for active logins
7449 goes becomes unavailable, and so does access to the user
7450 journals by the respective users.
7452 * If a group "adm" exists, journal files are automatically
7453 owned by them, thus allow members of this group full access
7454 to the system journal as well as all user journals.
7456 * The journal now stores the SELinux context of the logging
7457 client for all entries.
7459 * Add C++ inclusion guards to all public headers
7461 * New output mode "cat" in the journal to print only text
7462 messages, without any meta data like date or time.
7464 * Include tiny X server wrapper as a temporary stop-gap to
7465 teach XOrg udev display enumeration. This is used by display
7466 managers such as gdm, and will go away as soon as XOrg
7467 learned native udev hotplugging for display devices.
7469 * Add new systemd-cat tool for executing arbitrary programs
7470 with STDERR/STDOUT connected to the journal. Can also act as
7471 BSD logger replacement, and does so by default.
7473 * Optionally store all locally generated coredumps in the
7474 journal along with meta data.
7476 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
7477 writing short strings to files (for usage for /sys), and for
7478 creating symlinks, character and block device nodes.
7480 * New unit file option ControlGroupPersistent= to make cgroups
7481 persistent, following the mechanisms outlined in
7482 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
7484 * Support multiple local RTCs in a sane way
7486 * No longer monopolize IO when replaying readahead data on
7487 rotating disks, since we might starve non-file-system IO to
7488 death, since fanotify() will not see accesses done by blkid,
7491 * Do not show kernel threads in systemd-cgls anymore, unless
7492 requested with new -k switch.
7494 Contributions from: Dan Horák, Kay Sievers, Lennart
7495 Poettering, Michal Schmidt
7499 * This is mostly a test release, but incorporates many
7502 * The git repository moved to:
7503 git://anongit.freedesktop.org/systemd/systemd
7504 ssh://git.freedesktop.org/git/systemd/systemd
7506 * First release with the journal
7507 http://0pointer.de/blog/projects/the-journal.html
7509 * The journal replaces both systemd-kmsg-syslogd and
7510 systemd-stdout-bridge.
7512 * New sd_pid_get_unit() API call in libsystemd-logind
7514 * Many systemadm clean-ups
7516 * Introduce remote-fs-pre.target which is ordered before all
7517 remote mounts and may be used to start services before all
7520 * Added Mageia support
7522 * Add bash completion for systemd-loginctl
7524 * Actively monitor PID file creation for daemons which exit in
7525 the parent process before having finished writing the PID
7526 file in the daemon process. Daemons which do this need to be
7527 fixed (i.e. PID file creation must have finished before the
7528 parent exits), but we now react a bit more gracefully to them.
7530 * Add colourful boot output, mimicking the well-known output
7531 of existing distributions.
7533 * New option PassCredentials= for socket units, for
7534 compatibility with a recent kernel ABI breakage.
7536 * /etc/rc.local is now hooked in via a generator binary, and
7537 thus will no longer act as synchronization point during
7540 * systemctl list-unit-files now supports --root=.
7542 * systemd-tmpfiles now understands two new commands: z, Z for
7543 relabelling files according to the SELinux database. This is
7544 useful to apply SELinux labels to specific files in /sys,
7547 * Output of SysV services is now forwarded to both the console
7548 and the journal by default, not only just the console.
7550 * New man pages for all APIs from libsystemd-login.
7552 * The build tree got reorganized and the build system is a
7553 lot more modular allowing embedded setups to specifically
7554 select the components of systemd they are interested in.
7556 * Support for Linux systems lacking the kernel VT subsystem is
7559 * configure's --with-rootdir= got renamed to
7560 --with-rootprefix= to follow the naming used by udev and
7563 * Unless specified otherwise we will now install to /usr instead
7564 of /usr/local by default.
7566 * Processes with '@' in argv[0][0] are now excluded from the
7567 final shut-down killing spree, following the logic explained
7569 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
7571 * All processes remaining in a service cgroup when we enter
7572 the START or START_PRE states are now killed with
7573 SIGKILL. That means it is no longer possible to spawn
7574 background processes from ExecStart= lines (which was never
7575 supported anyway, and bad style).
7577 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
7578 reloading of units together.
7580 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
7581 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
7582 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
7583 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
7584 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek